JUMP RACING

FOR PROFIT

by Peter May

ISBN: 0-900611-995

Raceform Ltd.,
Compton, Newbury, Berkshire, RG20 6NL.

© Peter May 1996

Edited: Graham Wheldon

1
2
ACKNOWLEDGEMENTS

Many thanks to Sara Howes, Mike Fincham, David Rossiter and

Richard Lowther for their helpful comments and ideas.

Published: Raceform Ltd., Compton, Newbury,

Berkshire, RG20 6NL.
Printed by: Greenshires Print Ltd., Kettering, Northants.

3
 CONTENTS

INTRODUCTION 5

PART I - BACKGROUND DATA 9

Profitability .......................................................................... 15
Hurdle Races ...................................................................... 17
Chases ............................................................................... 20
National Hunt Flat Races ................................................... 23
Summary ............................................................................ 24

PART II - SELECTION METHODS FOR

JUMP RACING 25

Recent Form and Beaten Favourites.................................. 27

Weight Analysis ................................................................. 31
Front Runners .................................................................... 39
Sires ................................................................................... 47
The Time Factor ................................................................. 51
Developing Profitable Selection Methods ........................... 57

Concluding Remarks ..................................................... 62

4
 LIST OF TABLES

Table 1: Average Starting Price of Winners by Number

of Runners: All Grades of National Hunt
and Flat Races ....................................................... 13
Table 2: Percentage of Winning Favourites by Number of
Runners: All Grades of National Hunt and
Flat Races .............................................................. 15
Table 3: Average Return by Starting Price ............................. 16
Table 4: Winners to Runners Strike Rate and Average
Return for Hurdle Races: Age Range by
Race Category ........................................................ 19
Table 5: Analysis of Chases by Number of Runners
Within Race Category ............................................. 21
Table 6: Average Return and Strike Rate by Distance Beaten
Last Time: All Grades of National Hunt Races .......... 28
Table 7: Average Return by Distance Beaten and Starting
Price Last Time: Novices’ Hurdle Races ................... 29
Table 8: Winners to Runners Strike Rate by Weight
Carried: All National Hunt Handicap Races .............. 33
Table 9: Winners to Runners Strike Rate by Going:
All National Hunt Handicap Races ........................... 34
Table 10: Effect of Weight Carried by Course .......................... 36
Table 11: Analysis of Front Runners by Going ......................... 42
Table 12: Analysis of Front Runners by Race Distance ............ 43
Table 13: Average Return and Strike Rate for
Probable Front Runners by Racecourse .................. 44
Table 14: Average Return and Strike Rate of Progeny’s
Race Performances by Going .................................. 48
Table 15: Average Return and Strike Rate of Progeny’s
Race Performances by Race \Distance ..................... 49
Table 16: Average Return and Strike Rate of Progeny’s
Race Performances by Course Direction .................. 50
Table 17: Calculation of Going Allowance ............................... 52
Table 18: Strike Rate and Average Return for Good Time
Performers on Their Next Run ................................. 54
Table 19: Strike Rate and Average Return for Good Time
Performers on Their Next Run by Race Grade .......... 56

5
 INTRODUCTION

The cursor flickered on the computer screen, and the machine

posed the now familiar question: ‘Print Results?’. I responded by

pressing the y key and the printer clicked into action. A few

moments later I was looking at my first profitable betting system.

It was June 1986 and I had recently decided to abandon my

attempt to find a winning formula for the casino game Blackjack
which did not involve card-counting, and during the past two
weeks I had entered the results of one season’s novices’ chase
races into the computer. Historically the machine had only ever
analysed agricultural data (apart from many thousands of
Blackjack simulations) but now it was faced with an entirely
different task.

The data which had been extracted from the formbook greatly
simplified the complexity of the race, with only the month, number
of runners in the race, where the favourite and second favourite
finished on their latest starts, the position of the favourite and
second favourite in this race, and the starting prices of the two
horses entered. Three betting variables were added automatically:
return from the favourite backed to win, second favourite to win
and second favourite each way. The aim was to find a
combination of inputs which resulted in a positive return for one,
or all, of the betting variables.

At first glance the system appeared very profitable. It highlighted

well over one hundred bets per year and produced an average
profit of over 30p per £1 staked. So further testing was thought
unnecessary. Mistake number one. Instead, a more important
issue needed to be addressed: how much money could I expect to
win? A computer simulation was devised to answer this question
and it predicted a retirement

6
date for me of 1994. According to the computer, by 1994 I would
have amassed sufficient money to allow me to leave work and live
comfortably from the winnings produced by the system. The
simulation was based on an initial stake of £10 (from a bank of
£200) and would rise to a figure exceeding £1000 by 1994. But
eight years seemed a long time to be tied to the office, I wanted to
leave earlier. A simple solution presented itself - increase the
initial stake. My normal bet size was between £25 and £50, but I
would place £100 on the system bets. Mistake number two.

I had never anticipated the start of a new racing season with such
optimism. After what seemed an interminable wait, my first
betting opportunity finally arose on Saturday, August 2nd at
Newton Abbot. The horse: Karnatak. The price: 13/8. The
result: a comfortable success. The conclusion: easy money.

On the following Thursday I placed my second bet, and collected

from my second winner (price 5/4). A fantastic start, and now I
really believed that I had found a winning formula. The third bet
lost. So did the fourth. And the fifth. The sixth, Lord Laurence,
cruised into the lead at the fourteenth, but fell at the next, the
penultimate fence. Obviously I had expected a few losers, a ratio
of one winner to one loser in fact. But four consecutive losers
came as quite a shock, and put me into the red. So for the next
bet, St Colme at Cartmel, I decided to halve the stake. Mistake
number three. St Colme won easily at evens.

The St Colme race had taught me a valuable lesson; back to £100

per bet. The next bet went down and I had lost over £230 within
three weeks, substantially more than I was used to losing. I
therefore decided to play a little more carefully in the future. After
a great deal of form analysis I decided to ignore the next
recommendation completely. It won at 8/13. Another half stake
on bet 10 which won at 11/4. At this point I

7
was over £100 down (after tax), but had I kept to the system’s
guidelines and (level) staking plan, I would have been in profit by
£124.

This pattern continued throughout the remainder of the season, and

by the following summer I had lost money. By sticking to the
rules of the system and staking plan, however, I would have made
a profit of almost £2000, after tax, a fact which only served to
deepen my disappointment.

I have recounted this sequence of events to illustrate how simple it

is to lose money backing racehorses, even with a winning formula.
The three crucial errors I made are easy to spot in retrospect.
Further testing of the system would have increased my confidence
in its ability to return a profit, and may have highlighted ways to
reduce the likelihood of long losing runs. Increasing the stake way
beyond my normal bet put additional strain on the whole process.
A long losing run at half the usual stake is easier to handle than at
twice the normal stake. The third mistake is unforgivable. Using
my own judgement to determine whether I followed the system’s
advice only serves to introduce an additional bias into the
procedure, effectively making the system redundant.

For the next six years I managed to run the system properly, and it
produced a reasonable profit each season. However, the average
return gradually decreased due to the bookmakers’ awareness of
the weakness the system was exploiting, and by 1995 it was no
longer profitable. However, I have replaced the approach with
many other equally profitable selection methods which I detail in
the remainder of this book.

In Part I some general issues concerning the level of profitability

of National Hunt racing in Great Britain are outlined as
background data. The three main race categories: hurdles races,
chases and National Hunt flat races are examined

8
individually as well as age and race size profiles. Several
profitable selection methods are presented in Part II. These
approaches include methods based on beaten favourites, running
styles, and the use of race times. Other issues covered include the
effect of weight carried in handicap races and information which
can be gleaned from an examination of the sire’s record. The
section concludes with a brief summary of the key points of system
development.

Notation

Throughout this text the following notation will be used to

illustrate returns from several bets: six bets with two winners at
3/1 and 5/1 would be summarised as follows:

Bets Wins (%) Return/£

6 2 (33.3) 0.67

In the above example, the six bets produced a return of £4 to a £1

stake (i.e. +3+5-1-1-1-1 = 4), therefore the average return per bet
is 4 ö 6 (total return divided by the number of bets) = 0.67 (or 67p
per £1 staked). It should be noted that a return of 0.09 would
simply be sufficient to cover the cost of tax (at 9%), anything less
would result in a loss for off-course bettors. Finally, all bets
referred to throughout this book are win singles unless otherwise
stated.

9
 PART I

BACKGROUND DATA

10
BACKGROUND DATA

There are approximately 3,500 races run under National Hunt

rules in Great Britain each season, and this figure excludes the
many hundreds of Point to Point events which take place
throughout the country. Although National Hunt racing is
sometimes referred to as the winter game, the season now runs
throughout the year. The requirement for the racing authorities to
generate income at every possible opportunity, via betting levy and
admission charges, has reduced the mid-summer break from seven
weeks to just four days.

During the 1995/96 season 43 racecourses staged National Hunt

racing. The figure will be reduced by one in 1996/97; Nottingham
racecourse will only be used for Flat racing. Such a large number
of courses produces a wide variation in course configuration. For
example, consider the two south Devon courses: Newton Abbot
and Exeter. The former is a flat, tight, left-handed course suiting
the type of horse which prefers racing up with the pace. In
contrast, Exeter racecourse, located on the side of Haldon Hill
offers an entirely different configuration: right-handed with a stiff
uphill run-in of 300 yards. This variation in courses adds another
dimension

11
to race analysis, requiring the punter to determine whether his/her
selection will be suited by the track in addition to the normal
considerations of form and fitness etc. This is especially
important for front runners, a feature which is discussed in detail
in Part II.

Horses need to be at least three years old before they can race
under National Hunt rules with the career of a Jumps racehorse
tending to last longer than that of its Flat racing counterpart.
Generally starting in novices’ hurdle races, the National Hunt
horse can then progress through handicap hurdle races to novices’
chases and finally handicap chases. Naturally, this career path
takes several seasons to complete, and as such the National Hunt
horse races more often than the average Flat racing animal. This
is the main attraction for many Jumps racing supporters. Unlike
Flat racing, where the very best horses seldom race beyond their
three-year-old season, the National Hunt champions return year
after year to test their ability against the younger generations.
Imagine the 1993 King George VI and Queen Elizabeth Diamond
Stakes at Ascot attracting the 7 year old Nashwan, in addition to
Generous, Salsabil, User Friendly and the 1993 Epsom Derby
winner, Commander In Chief. Quite a race! This reluctance to
race the best Flat horses later in life is understandable since their
stud value would rapidly diminish should the animal fail to run up
to its best. When comparing the potential prize money which
could be won to stud values, racing a champion Flat horse beyond
its three-year-old season is very much a long odds on gamble.
However, this is no compensation for the Flat racing fan.
Fortunately, though, such contests occur quite frequently in
National Hunt racing. Take, for example, the 1995 King George
VI Tripleprint Chase, run at Sandown due to the abandonment of
Kempton. This race featured the 1995 Cheltenham Gold Cup
winner: Master Oats; the top chaser in Ireland: Merry Gale; the
winners of the King George VI race for the previous two seasons:
Barton Bank and Algan;

12
and the favourite for the 1996 Cheltenham Gold Cup: One Man.
The Racing Post described the race as ‘a titanic contest’. It is no
wonder that National Hunt followers become supporters of the
horses as opposed to just observers of races.

This tendency for the National Hunt horse to race over more
seasons and in more races means the race analyst has a greater
volume of form with which to work. More form implies that the
punter should be able to more accurately evaluate the animal’s
likes and dislikes, such as optimum going and race distance.
Consequently, determining the horse most likely to win a race
should be easier. However, this does not mean it is easier to make
money. The bookmakers are also able to assess the chances of
each runner more precisely, and as a result the prices on offer will
reflect these better informed assessments. To illustrate this point,
consider Table 1. This table shows the average price of all
winners in National Hunt and Flat races for several seasons
covering approximately 28,000 races in total:

Table 1: Average Starting Price of Winners by Number of

Runners: All Grades of National Hunt and Flat Races

Average Starting Price of Winners

Runners National Hunt Flat
2-4 2.05 2.01
5-9 4.34 4.57
10-15 6.51 6.95
16+ 8.57 9.83
All 5.13 6.09

Apart from to 2-4 runner fields, for which the difference is

insignificant, the average price of the winner is less for National
Hunt races than for Flat races. This reflects the improvement in
the accuracy of race assessments. In other words there is less
uncertainty in National Hunt races.

13
Fortunately, uncertainty still exists, and it is this uncertainty which
leads to the opportunity to win money. In many games of chance,
such as roulette, dice and the National Lottery, the uncertainty has
been removed. Whilst the outcome of any one spin of the roulette
wheel, roll of the dice, or drop of the lottery ball is unknown
before the event, the probability of each outcome can be computed
exactly. For example, with an unbiased dice, the probability of a
six being thrown on the next roll is 1/6. This probability is
mathematically defined and proven. With horseracing, even the
reasonably well defined National Hunt racing, this probability
remains inexact. It is this very fact which allows us to profit from
the sport, providing we bet sensibly.

Some people find the above result surprising. How can the
outcome of National Hunt races be easier to predict than the
results of Flat races, after all the horses have the added problem of
jumping fences? One answer to this question concerns the relative
levels of difficulty associated with the races. In theory, the more
difficult the sport or game of skill, the more likely it is to be
dominated by the best players. Taken to an extreme, an event that
requires no skill whatsoever and hence depends purely on luck,
could be won by any of the competitors. However, an event of
immense difficulty will only be won by those with the highest
level of skill. Chess, for instance, is dominated by a few top
players. Clearly, jumping 18 fences in a three mile steeplechase is
more difficult than running over a straight five furlongs, requiring
the runners to possess an acceptable level of speed as well as
jumping fluency. Consequently, the more talented participants
should dominate. To illustrate this point consider Table 2. This
table details the percentage of winning favourites for National
Hunt and Flat racing:

14
 Table 2: Percentage of Winning Favourites by Number of
Runners: All Grades of National Hunt and Flat Races

Percentage of Winning Favourites

Runners National Hunt Flat
2-4 51.5 54.4
5-9 37.2 36.3
10-15 32.2 27.8
16+ 25.7 21.1
All 36.0 31.8

The theory seems to hold up, with the strike rate for Jumps race
favourites, at 36.0%, significantly higher than for Flat race
favourites, at 31.8%. It is also interesting to note how the gap
between the pairs of percentages seems to widen as the number of
runners increases. This reinforces our theory since races with
more runners are, in general, harder to win.

However, it is still difficult not to feel anxious as our selection

approaches each hurdle or fence, with this level of anxiety seeming
to increase in proportion to the size of bet. So, what is the chance
of our selection falling? Based on an analysis of approximately
88,000 race performances, the proportion of fallers in all types of
National Hunt racing is about 4.4%. This is surprisingly low,
implying only 4 fallers every 10 races. The proportion for
unseated rider is even less at 2.4%, and only 0.3% of runners are
brought down. Therefore, in total, approximately 7% of runners
fail to complete due to falling, unseating or being brought down.

Profitability

We have established that National Hunt racing is more reliable

than Flat racing and that we should not worry unduly about our
horse falling, but to what degree are these facts reflected in the
prices offered by the bookmakers? The following table

15
presents the average return per £1 staked for almost 88,000 race
performances over Jumps:

Table 3: Average Return by Starting Price

Starting Average
Price Winners (%) Runners Return/£1

Odds On 1200 (57.2) 2098 -0.07

Evens-2/1 2070 (35.7) 5791 -0.09
85/40-5/1 3319 (20.1) 16539 -0.10
11/2-10/1 1868 (10.0) 18646 -0.15
11/1-20/1 765 (3.9) 19512 -0.39
22/1-40/1 193 (1.5) 13044 -0.56
50/1+ 55 (0.5) 12037 -0.74

Total 9470 (10.8) 87667 -0.33

From Table 3 we can see that by selecting horses at random in

National Hunt races we could expect to lose approximately 33p
per £1 staked. This is the result of the bookmakers’ pricing
policy. The average loss for Flat racing is approximately 30p per
£1 staked (see Flat Racing For Profit, published by Raceform).

Although this loss of 33p per £1 staked is extremely high in

relation to many casino games, it compares favourably with the
National Lottery (50p per £1) and football pools betting at 70p
per £1. Premium Bonds, on the other hand, offer a much safer bet
with an average gain of 5p per £1 staked and no risk attached.

On the positive side, Table 3 reinforces the view that National

Hunt racing is predictable. This is illustrated by the high
correlation between the odds on offer and the strike rate. As the
winners to runners strike rate increases, the odds decrease. If the
outcome of the races were not predictable, by

16
bookmakers and punters alike, such a strong relationship would
not be apparent.

It is interesting to note from Table 3 that the average return for the
higher priced horses is poorer than for the shorter priced animals.
This simply reflects the greater degree of under pricing associated
with these runners. In other words, in order to accurately mirror
the true probability of success, most 33/1 shots should really be
priced at 50/1 or greater. As the price decreases the discrepancy
between the fair price and starting price diminishes. Therefore,
restricting our random selection process to horses starting at 10/1
or less reduces the loss to just 10½p per £1 staked. Whilst a 25/1
shot may look an attractive betting option, these longer priced
horses tend, in general, to offer poor value compared to the shorter
priced animals.

Hurdle Races

Hurdle races constitute the majority of all National Hunt races,

54% in fact. Broadly speaking, these races are contested by ex-
Flat race horses, and National Hunt bred animals for whom
hurdling is merely a stepping stone to steeplechasing. Hurdle
races can be conveniently categorised into four groups: conditions
races, novices’ races, other non-handicap races and handicaps.
Conditions races are non-handicap events which include the best
hurdle races staged each year, such as the Champion Hurdle and
Stayers’ Hurdle. In addition to the graded races, conditions events
include contests for four year old second season hurdlers as well as
trials for the championship races. Novices’ hurdle races were,
historically, for horses which had not won a hurdle race prior to
the current season. However, due to the introduction of summer
racing this condition was revised to not having won a hurdle race
prior to the 1st May. Consequently, under the new conditions,

17
horses could have won a hurdle race in the previous season and
still qualify for novices’ events in the current season. The
category other non-handicap races refers to the remaining races for
which horses are not allotted weights by the British Horseracing
Board’s (BHB) handicappers. Therefore, this category includes,
amongst others, selling, maiden and claiming races. Handicap
races are weight adjusted events. Once qualified for a handicap
the BHB‘s team of handicappers rate the performances of the
animal and evaluate its handicap mark. This mark will change as
the season progresses and the horse runs in more races. The
handicap mark determines the weight the horse is set to carry
relative to the other runners in the race. The aim of the weight
adjustment is to produce more competitive races by penalising the
better animals. Handicap hurdle races (which include novices’
handicaps, selling handicaps etc.), and the use of weight to slow
horses down, is discussed in more detail in Part II.

The minimum distance of a hurdle race is two miles, with the

longest race at a little over 3 miles 3 furlongs. Almost 60% of all
hurdle races are over a distance of 16-18 furlongs with the runners
expected to jump between 8 and 10 flights of hurdles. Less than
6½% of races are over a distance exceeding 3 miles, requiring the
horses to jump at least 12 flights.

Horses aged three years and upwards are qualified for hurdle
races. However, 44% of all runners are aged between five and
seven years. This seemingly low average age range is simply due
to a large proportion of hurdlers progressing to steeplechasing
after one or two seasons over the smaller obstacles. The older
hurdlers tend to be animals which are not suited, either physically
or mentally, to fences. Table 4 shows the distribution of runners
by age range for almost 7,000 races:

18
Table 4: Winners to Runners Strike Rate and Average Return for
Hurdle Races: Age Range by Race Category

Age Cond. Nov. Other Hcap. All

(yrs.) Hdl. Hdl. Hdl. Hdl. Races

3 W 0 304 96 19 419
R 0 3261 1012 90 4363
Ret. - -0.41 -0.24 +0.39 -0.36

4 W 17 715 207 420 1359

R 145 7305 2319 3447 13216
Ret. +0.18 -0.42 -0.48 -0.26 -0.38

5..7 W 64 1272 324 1584 3244

R 344 13898 3656 12999 30897
Ret. -0.28 -0.44 -0.41 -0.16 -0.32

8..10 W 54 302 169 1138 1663

R 398 4493 1981 13110 19982
Ret. -0.32 -0.60 -0.46 -0.32 -0.40

11..12 W 1 2 13 37 53
R 20 45 97 880 1042
Ret. +0.70 -0.81 +0.16 -0.63 -0.54

13+ W 1 2 3 7 13
R 7 11 22 176 216
Ret. -0.36 +0.82 -0.62 -0.72 -0.62

Total W 137 2597 812 3205 6751

R 914 29013 9087 30702 69716
Ret. -0.20 -0.46 -0.42 -0.26 -0.36

W indicates the number of winners, R the number of runners, and Ret

the average return per £1 staked.

One of the most significant features revealed by Table 4 is the

decline in average return for the older horses. Consider, for
instance, those hurdlers aged 11 years and older: 1258 race
performances resulted in only 66 wins, a strike rate of 5.2%.

19
This compares to a strike rate of 9.8% for their younger
opponents. However, this poorer strike rate is not reflected in
higher prices offered by the bookmakers, hence the abysmal return
per bet. Specifically, a loss of 55p per £1 staked for the older
animals compared to a loss of 36p per £1 staked for the younger
runners. This illustrates the degree of poor value generally offered
about older horses.

Another interesting feature is the profitable return from backing all

runners aged three years taking part in handicap hurdle races. The
average return of 39p per £1 staked reflects how much this type of
runner is under rated, by the punters and bookmakers, resulting in
odds which are, in fact, generous. Three-year-old hurdlers often
will be considered too weak and inexperienced compared with
their older rivals. Clearly, this judgement is misplaced, which is
an error we can exploit.

With an average loss of 46p per £1 staked, novices’ hurdle races

would seem to offer the poorest betting medium. However, this
loss can be explained, in part, by a higher than average number of
runners per race, and more high priced runners than other types of
hurdle race. In fact, novices’ hurdle races are, in general, easier to
assess with a winners to runners strike rate for favourites of 44%
over the past four seasons, compared to an average of 37% for all
hurdle races for the same period. In terms of profit, backing all
novices’ hurdle race favourites since the 1992/93 season would
have resulted in a loss of only 4p per £1 staked.

Chases

Although hurdle racing can be exciting at times, it is

steeplechasing which usually provides the spectacle for National
Hunt racing. If you are unsure about this statement,

20
cast your thoughts back to the closing stages of the 1995 Mumm
Melling Chase at Aintree. The three top 2 mile chasers, Viking
Flagship, Deep Sensation and Martha’s Son, approached the final
fence at full racing speed and jumped it in unison before battling
out the finish with just a length separating the three at the line.
That’s real horseracing.

Chases make up approximately 42% of all National Hunt races

and range in distance from 2 miles, requiring the horses to possess
speed and jumping fluency, to the 4½ miles of the Grand National
where the emphasis is on stamina.

To facilitate meaningful analyses, chases have been broken down

into five categories. The first four: conditions races, novices’
races, other non-handicap races and handicaps, are the same as
for hurdle races with the same definitions. The fifth category is
hunter chases. These events are for horses which have been
classified as hunters during the hunting season and are restricted to
amateur riders.

An initial analysis of chases by race category and number of

runners reveals several interesting trends. Consider Table 5:

Table 5: Analysis of Chases by Number of Runners

Within Race Category

Number of Cond. Nov. Other Hcap. Hunter All

Runners Chase Chase Chase Chase Chase Races

2..4 W 51 269 13 526 36 895

R 183 994 48 1893 129 3247
Ret. -0.07 -0.25 -0.15 -0.09 -0.30 -0.15

5..7 W 54 661 45 1243 133 2136

R 312 4000 276 7375 811 12774
Ret. -0.23 -0.27 -0.29 -0.16 -0.32 -0.21

8..12 W 16 628 58 882 226 1810

R 156 6008 541 8304 2158 17167
Ret. -0.36 -0.38 -0.25 -0.22 -0.42 -0.30

21
 13..16 W 4 122 19 144 96 385
R 60 1737 278 2016 1347 5438
Ret. -0.60 -0.46 -0.50 -0.31 -0.47 -0.41

17+ W 0 12 9 41 20 82
R 0 215 170 813 402 1600
Ret. - -0.38 -0.54 -0.32 -0.64 -0.43

Total W 125 1692 144 2836 511 5308

R 711 12954 1313 20401 4847 40226
Ret. -0.25 -0.35 -0.34 -0.20 -0.43 -0.28

W indicates the number of winners, R the number of runners and Ret the average
return per £1 staked.

Clearly, as the number of runners per race increases so does the

bookmakers’ over-round. This is a trend which is continued
through both codes of racing, Flat and Jumps, and as mentioned in
the previous section, it accounts for the poor return for novices’
hurdle races. Hunter chases seem to offer a very poor average rate
of return, with a loss of 43p per £1 staked, which is, in fact, worse
than the National Lottery when the effect of tax is added.
Unusually, this poor return is maintained in the very small fields.
From Table 5 it can be seen that in fields of 2-4 runners the
average loss for handicap chases is 9p per £1 staked, whereas for
hunter chases the figure is 30p per £1, a huge difference.

A reason for this poor rate of return can be found by examining the
price profile of the hunter chase runners. These races usually
feature a large proportion of horses with very little chance of
winning, probabilities which are not accurately reflected by the
odds on offer (i.e. long shots still under priced, illustrated by
Table 3). Consequently, the bookmakers’ over-round is increased.
However, the hunter chase category has the highest proportion of
winning favourites, at a little over 48%, compared to 46% for
novices’ chases and figures in the mid 30’s for the other
categories. In fact, simply backing all hunter chase favourites over
the past

22
few seasons would have returned a profit of 7p per £1 staked,
before tax. All other race types produced a loss of between 3p and
23p per £1 staked.

The average age range of a chaser is 8 to 10 years old. As with

the hurdlers, the older chasers win less frequently than the younger
horses and this poor strike rate is not reflected in higher prices
offered by the bookmakers. Chasers aged up to 10 years old have
a winners to runners strike rate of 14% and an average loss of 25p
per £1 staked. In comparison, the strike rate for older horses, 11
years and upwards, is only 8% and the average return is a loss of
44p per £1 staked. Clearly, the age of the animal is a factor which
should be carefully considered before placing a bet.

National Hunt Flat Races

National Hunt Flat races are designed to give National Hunt bred
horses experience of racing before they attempt hurdles or fences.
These races attract a high bookmakers’ over-round with a random
selection producing an average loss of 47p per £1 staked over the
past few seasons. This level of under pricing is due in part to the
fact that a large proportion of the runners are unknown quantities,
and the bookmakers are not prepared to take any chances about
this type of horse. To illustrate this point, by selecting horses at
random in the 11/2-10/1 price band for all type of National Hunt
race would return a loss of 15p per £1 staked. For National Hunt
Flat races this loss is an incredible 30p per £1 staked.

As a result of such prohibitive pricing, identifying profitable

betting strategies for National Hunt Flat races is an extremely
difficult task.

23
Summary

In this section I have tried to show the degree to which the odds
are stacked in favour of the bookmakers, and how the average rate
of return varies depending on the number of runners in the race,
the starting price, and age of the horse. Whilst the starting price
provides a good guide to the chance of success there are instances
where it severely under estimates the true odds of runners, most
notably the older horses. We have also seen one case, three year
old horses running in handicap hurdle races, where the average
starting price exceeds a fair price providing a profitable betting
opportunity.

In the following section, we concentrate on developing selection

methods which should result in a positive return. We have already
seen that on average the bookmaker bets to an over-round of 33%
per race, and has at his/her disposal all the information available
to the general public. Therefore, using conventional methods for
selecting horses (i.e. checking the suitability of the going, race
distance etc.) is unlikely to return a profit. For the off-course
punter to make a long term profit by evaluating races in the
traditional way, either the bookmaker needs to make an appalling
mistake or the punter needs to be exceptionally talented at
analysing races. To put the level of skill required into perspective,
the top professional gamblers only make 10p to 15p per £1 staked.
Consequently, the best route to a profitable return is to adopt an
unconventional approach to racehorse selection and consider
information which is often overlooked by the bookmaker, and
other punters. This type of approach to betting is examined in
Part II.

24
 PART II

SELECTION METHODS FOR

JUMP RACING

25
SELECTION METHODS FOR
JUMP RACING

In this section we examine various facets of National Hunt racing

and develop several profitable selection methods. The section
starts with an analysis of recent form and the importance of the
animal’s starting price on its previous race.

Recent Form and Beaten Favourites

Recent form is treated as one of the most important aspects of race

analysis. In almost every book aimed at providing the reader with
methods for picking winners there will be a section relating to
recent form. Usually the author will suggest that the punter
ensures that his/her selection has ‘good recent form’ before
placing a bet. However, defining good or poor recent form is not
an easy task. A guide, though, can be gleaned from considering
the distance the horse was beaten on its latest run. Table 6
illustrates the relationship between the chance of success and the
distance the horse was beaten on its latest run:

26
 Table 6: Average Return and Strike Rate by Distance Beaten
Last Time: All Grades of National Hunt Races

†
Distance
Beaten Average
Last Time Winners (%) Runners Return/£1

0 2070 (25.5) 8121 -0.12

0.01-2.0 515 (20.6) 2495 -0.16
2.01-5.0 687 (18.7) 3677 -0.15
5.01-10.0 856 (16.2) 5286 -0.16
10.01-20.0 1140 (12.2) 9320 -0.25
20.01+ 1626 (6.6) 24599 -0.41
Not Complete 721 (6.0) 11971 -0.46
‡
Unraced 1855 (8.4) 22208 -0.37

Total 9470 (10.8) 87667 -0.33

† ‡
distance beaten is given in horse lengths. unraced this season

As you may expect, the above table does not reveal a profitable
betting strategy, however it does provide some valuable
information. The most interesting feature of Table 6 is the decline
in strike rate and average return as the distance beaten increases.
In other words, the starting price does not reflect the lower
probability of success of those horses well beaten last time.

A second component of recent form which should be considered is

the price the horse started for its latest race. We have already seen
that the starting price of a horse provides a good guide to its
chance of success (see Table 3). This trend is maintained for the
price the animal started on its latest run and is particularly
significant for novices’ hurdle races. Table 7 combines starting
price and distance beaten last time for almost 2,000 novices’
hurdle races:

27
 Table 7: Average Return by Distance Beaten and Starting Price
Last Time: Novices’ Hurdle Races

†
Distance Starting Price Last Time
Beaten Odds Evens- 85/40- 11/2- All
Last Time On 2/1 5/1 10/1 11/1+ Prices

0 W 122 125 172 87 39 545

R 367 422 625 357 254 2025
Ret. -0.18 -0.01 -0.09 -0.20 -0.28 -0.13

0.01-2.0 W 16 44 53 27 21 161
R 56 121 225 196 175 773
Ret. -0.19 +0.33 -0.13 -0.53 -0.49 -0.25

2.01-5.0 W 10 36 57 53 47 203
R 35 120 283 255 310 1003
Ret. +0.03 -0.09 -0.25 -0.18 -0.29 -0.22

5.01-10.0 W 21 33 86 74 56 270
R 55 113 390 394 534 1486
Ret. +0.13 +0.35 -0.04 -0.20 -0.47 -0.20

10.01+ W 14 56 145 171 291 677

R 71 275 1128 1819 7925 11218
Ret. -0.34 -0.03 -0.30 -0.35 -0.64 -0.54

Not W 8 12 32 36 39 127
Complete R 28 74 267 368 2581 3318
Ret -0.28 -0.36 -0.49 -0.04 -0.70 -0.60

Total W 191 306 545 448 493 1983

R 612 1125 2918 3389 11779 19823
Ret. -0.17 +0.03 -0.22 -0.28 -0.63 -0.46

W indicates the number of winners, R the number of runners and Ret the average
† distance beaten is given in horse lengths.
return per £1 staked.

Several interesting features are exhibited in Table 7. Firstly, the

average return remains reasonably constant for horses beaten up to
10 lengths but drops markedly to a loss of 54p per £1 staked for
horses beaten over 10 lengths on their previous run, and to a loss
of 60p for those which failed to complete last time. Secondly, for
runners which started at odds against on their previous start there
is a steady decline in the average return as this price increases.
And for animals which started at over 10/1 most recently the
average return

28
drops to a very poor loss of 63p per £1 staked. These are clearly
horses to avoid.

However, profitable selection methods are also appearing, notably

horses beaten on their last start by less than 10 lengths whilst
starting at 2/1 or less. From Table 7, we can see that this
approach would have resulted in 160 winners from 500 races, a
strike rate of almost 32% with a return of 13p per £1 staked.

In Part I we found that the higher priced horses generally offer

poorer value. Removing the horses which started at 10/1 or higher
and applying the restrictions outlined above decreases the number
of bets by 70 to 430 at the expense of only one winner. The strike
rate improves to 37% with a return of 28p per £1 staked.
Therefore,

consider backing any horse starting at less than

10/1 in a novices’ hurdle race which, on its latest
start, was beaten no more than 10 lengths at a
starting price of 2/1 or less: expected profit 28p per
£1 staked.

This selection method involves a fair amount of searching through

past histories, but fortunately it can be simplified by restricting the
horses eligible for selection to beaten favourites. Both of the trade
papers indicate beaten favourites in the race card reducing the
amount of searching required. Therefore,

consider backing any horse starting at less than

10/1 in a novices’ hurdle race which, on its latest
start, was a beaten favourite finishing within 10
lengths of the winner: expected profit 25p per £1
staked.

29
There is only a slight difference in the expected return and number
of bets for these two selection methods, and as mentioned earlier
the second approach should be easier to implement.

Weight Analysis

In horseracing, Jumps or Flat, the standard measure of ability is

the official rating, also known as the handicap mark. Ultimately
all horses receive a handicap mark assigned by a team of
handicappers employed by the BHB, whether they run in handicap
races or not. The rating is a convenient way of expressing the
superiority of one animal relative to another. For example, a horse
rated 85 would be considered to possess more ability than an
animal rated 70. For handicap races, this rating determines the
amount of weight each horse will carry. In the previous example,
the horse rated 85 would carry 15 lbs. more weight than the horse
rated 70. It should be noted that the handicap marks do not
determine the amount of weight to be carried, just the weight one
horse will carry relative to another. Consequently, it would be
possible for these two horses to be set to carry 11-04 (11 stone 4
lbs.) and 10-03 in one race, and 11-10 and 10-09 in another race
on the same day. The absolute weight carried is determined by the
BHB rating of the highest rated horse in the race. In these two
hypothetical races, the top rated horse in the first race would be
rated higher than the top rated horse in the second race.

This is an important distinction to make, and it is imperative to

remember that a horse carrying 12-0, for instance, is not
necessarily badly handicapped. Likewise, an animal set to carry
10-0 is not guaranteed to be well handicapped. The weight carried
by a horse simply reflects the strength of the

30
race. Just because a handicapper is set to carry 12-0 is no reason
to assume it cannot win.

There are occasions when runners in handicaps carry more weight

than their official rating requires. This can happen when the
rating requires the animal to carry a weight which is below the
minimum set for the race. The minimum weight is normally 10-0
although it can be increased to 10-04 or 10-07 for certain races.
In such cases horses which are out of the handicap effectively
carry overweight. For example, if the top weight in a race was set
to carry 11-10 and was rated 120, a horse rated 80 should only
carry 8-08 but would have to carry the minimum weight, 10-0, an
overweight of 20 lbs. These horses are generally considered to be
badly handicapped.

Official ratings have most effect in handicap races. The aim of

handicaps is to give all horses the same chance of winning by
penalising the better animals by giving them more weight to carry.
Consequently, many gamblers do not find handicap races a viable
betting medium. For instance, in the book Braddock’s Complete
Guide to Horse Race Selection and Betting (Longman, 1983),
Peter Braddock recommends that handicaps “must not be
considered for serious purposes of selection”. This is a rather
extreme view, although it did discourage me from betting in
handicaps for many years.

However, in Betting For A Living (Aesculus Press, 1992), Nick

Mordin suggests that weight carried should be ignored when
assessing form and provides evidence from this, and the previous
century, to support the claim. At first I dismissed this idea
without a second thought, however, I am beginning to believe that
it may have some validity. We should be able to test this theory
using independent handicap ratings as provided by organisations
such as Timeform and Raceform. If Mordin’s contention is true,
there should be no difference between the winners to runners strike
rate for the top rated

31
selections after adjusting for weight (the usual approach) and for
the top rated selections without any weight adjustment. Analysing
the ratings from an independent source for almost 800 handicap
races produced the following results:

Strike Rate For Weight Adjusted: 24%

Top Rated Horse Weight Ignored: 23%

The minimal difference between the two percentages supports the

hypothesis that weight carried has no effect. It also questions the
value of high priced ratings services, especially since the top
weight in Jumps handicap races regardless of its position in the
ratings, wins on average 20% of races.

As a further check we can look at the winners to runners strike rate

for all National Hunt handicap runners by weight carried. Again,
if weight has a significant effect, and the handicappers are
reasonably accurate in their assessments, this profile should be
uniform over the different weight categories. On the other hand, if
weight has only a minimal effect, the better horses, carrying higher
weights, should win more often. Consider Table 8 which
illustrates the winners to runners strike rate for almost 6,000
handicap races:

Table 8: Winners to Runners Strike Rate by Weight Carried:

All National Hunt Handicap Races

Weight
†
Carried Winners (%) Runners

12-01+ 27 (38.6) 70
11-08..12.00 1540 (16.7) 9218
11-00..11-07 1246 (13.6) 9181
10-08..10-13 1278 (12.1) 10594
10-00..10-07 1526 (9.1) 16711
..9-13 286 (6.6) 4306

All Weights 5903 (11.8) 50080

† weight carried includes claims by jockeys and overweight.

32
Clearly, the horses carrying most weight win more often which
indicates that the handicapping process is not resulting in total
uniformity. So perhaps Nick Mordin is correct, and for National
Hunt races at least, we should not place too much emphasis on the
effect of the amount of weight carried when assessing the chances
of horses. And to base the assessment of the runners’ abilities on
the official ratings before adjustment for weight.

Another long held belief is that horses find it harder to carry

weight on soft ground. Braddock comments: ‘it is more difficult
for horses to carry big weights on soft or heavy ground.
Conversely, when the ground is firm ... top weights are more likely
to prevail.’ Mordin agrees with this statement, and it seems to
possess a degree of logical merit. Table 9 shows the number of
winners for almost 6,000 runners that carried 11-10 or more in a
National Hunt handicap races, together with the difference from
the expected number of winners:

Table 9: Winners to Runners Strike Rate by Going:

All National Hunt Handicap Races

Diff. From Adjusted

Going Winners (%) Runners Expected Difference

Hard 5 (38.5) 13 2.0 68.9

Firm 110 (23.6) 466 23.3 26.9
Good to firm 276 (21.5) 1284 82.2 42.4
Good 298 (16.4) 1812 61.0 25.8
Good to soft 168 (16.8) 1000 38.1 29.3
Soft 151 (17.9) 844 32.5 27.4
Heavy 86 (21.0) 409 22.4 35.3

All 1094 (18.8) 5828 261.6 31.4

For horses which carried 11-10 or more in handicaps. The expected

number of winners is calculated directly from the number of runners
in each race analysed.

33
The first feature to note regarding Table 9 is that all elements in
the first Difference column are positive. This illustrates the higher
than average strike rate for top weights compared to other runners.
The total figure of 261 implies that there were 261 more winners
than would have been expected if the horses had been selected at
random from all weight bands. Although the figures in the
difference column tend to decrease towards the softer going
categories, this is due to the decline in the number of soft/heavy
ground races analysed. In fact, when this effect is removed the
pattern becomes quite uniform.

The Adjusted Difference column expresses the difference as a

percentage of the expected number of winners. In other words, the
figures indicate the percentage by which the actual number of
winners exceed the expected number. For instance, for the heavy
going category there were 29.3% more winners than expected.
The high figure for hard going is not particularly reliable due to
the small sample size, however, the remaining figures are
reasonably uniform. It could be argued that the firm ground figure
is high, supporting the case that horses carry weight better on
firmer ground, though the good to soft going percentage, at
35.3%, may also be considered high. Consequently, there is no
evidence from the above data to conclude that the top weights
enjoy a greater advantage over the other runners when the going is
firm, and neither are they greatly inconvenienced when the going
turns soft.

The final factor to consider under the heading of Weight Analysis

concerns the effect of course configurations on the ability of the
animal to carry weight. It has been suggested that tight turning
courses help an animal to carry greater weights since the speed of
the runners on these types of track will be lower, on average, when
compared to the, so called, galloping tracks. However,
justification of this assumption

34
requires accurate definitions of the terms tight track and
galloping track. Ideally, courses would be measured to a
reasonable degree of precision, and data relating to the sharpness
of the bends recorded. In the absence of these figures, we could
use the definitions presented in the official formbook, or those
provided by Timeform to categorise the courses. However, I am
not totally happy with these descriptions, which are made, I
believe, simply by observation. An alternative approach is to
simply analyse all courses individually and base any deductions on
these data. The following table lists all National Hunt courses
together with the winners to runners strike rate, the average return
per £1 staked together with the percentage difference between the
number of winners recorded and the number of winners expected
for all handicap runners which carried 11-10 or more:

Table 10: Effect of Weight Carried by Course

Average
Course Winners (%) Runners Return/£1 % Diff.

Worcester 57 (23.8) 239 +0.10 79.1

Hereford 34 (23.1) 147 +0.01 75.9
Windsor 19 (20.0) 95 +0.09 74.6
Newcastle 27 (27.3) 99 -0.06 74.0
Perth 30 (28.0) 107 +0.17 65.1
Exeter 37 (21.5) 172 -0.12 56.6
Kelso 25 (23.6) 106 +0.12 54.9
Doncaster 14 (18.9) 74 -0.01 51.5
Fontwell 41 (21.7) 189 -0.14 48.1
Towcester 34 (20.9) 163 -0.03 45.9
Chepstow 29 (22.3) 130 -0.10 43.8
Sedgefield 43 (19.9) 216 -0.17 43.3
Ludlow 23 (16.8) 137 -0.36 42.9
Lingfield 19 (22.6) 84 -0.25 40.0
Bangor 28 (20.9) 134 -0.22 39.9
Nottingham 18 (16.5) 109 -0.03 36.9
Carlisle 21 (20.2) 104 -0.25 35.0
Ayr 32 (21.3) 150 -0.10 34.7
Wincanton 26 (19.8) 131 -0.16 34.5
Plumpton 34 (22.7) 150 -0.26 33.5
Market Rasen 31 (17.5) 177 -0.10 32.3

35
 Ascot 23 (18.1) 127 -0.14 31.3
Southwell 26 (20.8) 125 -0.29 30.9
Wetherby 36 (20.2) 178 -0.17 30.3
Uttoxeter 44 (17.7) 249 -0.20 29.6
Cartmel 9 (20.5) 44 -0.02 23.3
Catterick 14 (14.6) 96 +0.01 21.4
Stratford 30 (16.9) 177 -0.16 21.3
Musselburgh 14 (15.4) 91 -0.26 20.8
Huntingdon 32 (17.7) 181 -0.29 20.1
Haydock 23 (21.3) 108 -0.30 15.9
Newton Abbot 49 (17.9) 274 -0.38 14.5
Newbury 22 (16.9) 130 -0.23 6.7
Kempton 13 (14.9) 87 -0.28 5.7
Warwick 18 (14.4) 125 -0.12 5.0
Hexham 17 (13.5) 126 -0.52 3.9
Aintree 8 (12.1) 66 -0.51 1.5
Cheltenham 24 (13.0) 185 -0.40 0.3
Fakenham 11 (12.2) 90 -0.35 0.0
Folkestone 13 (14.4) 90 -0.47 -4.8
Taunton 14 (9.8) 143 -0.29 -11.1
Sandown 18 (14.2) 127 -0.46 -12.2
Leicester 8 (10.5) 76 -0.48 -22.9

For horses which carried 11-10 or more in handicaps.

Table 10 is sorted by the percentage difference column, therefore,

the first named courses are best for top weights. Unlike Table 9,
there are several negative figures in this column. These indicate
the courses where carrying a large weight is actually detrimental to
the chances of the horse winning (i.e. fewer horses won than
expected). To emphasise this point further, the average return
column shows a huge loss of 48p per £1 staked for all bets on
horses carrying 11-10 or more in handicaps at Leicester. This
compares to a profit of 17p per £1 staked at Perth.

Table 10 is only a guide to the likely effect of racecourse

configuration on the chances of the top weights. There are several
other factors which have not been considered and which could
contribute to the order of the courses given above. For instance,
there may be a difference in the average number of runners per
race between the tracks which would

36
introduce a slight bias. But more importantly, racing may, in
general, be less competitive at some courses with fewer horses
carrying large weights per race (i.e. bigger gaps, in pounds,
between the top weight and next highest weighted horse).
However, if we are prepared to allow for a small degree of error,
in what is a very uncertain domain, the results presented in Table
10 should be useful when evaluating the chances of handicappers.

So far we have identified a theory regarding weight carrying which

runs counter to conventional logic, and have illustrated that it
possesses a degree of merit. Consequently, we should be able to
turn this theory into a profitable betting system. The key factors
we need to consider are weight carried and racecourse
configuration, the going does not appear to have a significant
effect.

Using Table 10 it is possible to identify the most successful tracks

for top weights. Setting a percentage difference threshold at 70%
isolates four courses which are considered significantly better for
top weights. These courses are: Worcester, Hereford, Windsor
and Newcastle. Considering all handicap chase runners which
carried 12-0 or more at these courses, starting at less than 10/1,
(the reasons for imposing the price restriction have been discussed
in previous sections) over the past few seasons would have
resulted in 249 bets with 79 winners, a strike rate of 32%, and an
average return of 26p per £1 staked. Therefore,

consider backing any horse starting at less than

10/1 in a handicap chase at Worcester, Hereford,
Windsor or Newcastle which is carrying 12 stone or
more: expected profit 26p per £1 staked.

37
The evidence published in Betting For A Living, based on tests
undertaken in the 19th century, indicated that weight only slowed
horses to an appreciable degree when the amount reached 14
stone. From the above tables it appears that this argument could
still apply today, over one hundred years later.

In summary, the horses which carry more weight win more often.
Furthermore, we found that using the weight allotted as an ability
rating is as accurate as the tested independent ratings service. The
advantage afforded to the top weights does not appear to be
reduced when the going turns soft, but can be further enhanced by
the configuration of the racecourse.

Front Runners

In the Beaten Favourites section we found a profitable betting

strategy which ignored current form, and the competition from the
other runners in the race. It simply focused on two items of data
from the animal’s performance history. This does not sit happily
with conventional racehorse selection which requires the bettor to
evaluate many attributes associated with each runner, such as
suitability of going and race distance, in addition to overall level
of ability, and then to make a comparison based on these data to
determine the most likely winner. However, we found that the
Beaten Favourites method produces a reasonable profit which is
almost entirely due to its unconventional form and the fact that the
price at which the horse started in its previous race is not
considered to be a significant discriminating factor by most
punters and bookmakers. An obvious approach to develop
alternative profitable selection methods is to consider other data
items which are overlooked by the betting public, and determine
their effectiveness at returning a profit. In this section we look at

38
another commonly ignored variable, the style of running and
specifically front runners.

In human athletics, other than sprints, front running is not

particularly popular. Human middle distance athletes seem to
perform better if they are able to run just behind another
competitor for the majority of the race. For instance, pace makers
are always used for record breaking attempts to ensure that the
better athletes do not have to run from the front. On the other
hand, in horseracing front running is quite popular. This
difference could be due to contrasting training regimes or the
psychological differences between humans and horses. A human
athlete knows the exact distance of the race, the distance he/she
has still to run and the fact that the other runners will try to pass
him/her. Horses are aware of none of these facts, and are simply
running. There are, of course, exceptional cases, where particular
horses will not run so well when out in front, and if they are to win
need to be held up until the final stages of the race. But in
general, horses, without the acute mental awareness of the human
athlete, should be able to run just as well, or better, from the front
as from any other position in the field.

One popular misconception concerning front runners is the speed

they are running relative to the other runners. It is a long held
belief by some race observers that front runners are actually
running faster than the following pack, and hence the horses which
are held up are conserving their energy. If this were true, the gap
between the front runner and the pack would be gradually
increasing with the front runner drawing further clear. If the gap
remains constant, it means that all the horses are running at, or
about, the same speed.

There are many advantages to front running which were clearly

illustrated by the Martin Pipe - Peter Scudamore partnership
during the late 1980’s. For example, front runners

39
cannot be brought down. Admittedly, less than 1% of runners are
brought down (see Part I) so the advantage is only slight, but
nonetheless it is an advantage. Front runners do not suffer from
being unsighted when approaching a hurdle or fence making the
jump much harder to execute and increasing the chances of falling.
Master Oats exemplified this point when winning the 1995 Gold
Cup. In the initial stages of the race Master Oats was held up
behind several horses and his jumping was far from perfect. But
as soon as Norman Williamson moved him to the outside his
jumping improved markedly and success was virtually assured.
Other advantages which can be exploited by the jockey include:
taking the shortest route, selecting the best ground on which to
race, dictating the pace of the race and, most importantly, starting
the race with several lengths lead over the other runners.

So, if we believe that front runners hold a distinct advantage over

their competitors, and that this advantage is not universally
accepted, we should be able to develop a profitable betting
strategy based on this assumption.

Before analysing the data it is important to precisely define the

term front runner (this aspect of the analysis is covered in detail
in the Developing Profitable Selection Methods section). From
the information in the formbook, how can we tell if the race was
won by a front runner? Clearly, we need to look at the comments-
in-running. These comments are designed to provide a brief
summary of how the race was run and should contain the
information we require. However, analysing these comments is
not easy, and in terms of race comments, the definition of a front
runner is, to a certain extent, ambiguous. For instance, would a
chaser, which took up the lead at the second fence and maintained
the lead to the line, be considered a front runner? What about an
animal which took up the lead at the third obstacle, was briefly

40
headed two fences out, but regained the lead to win? Obviously,
we need to establish a set of rules which satisfy our own definition
of a front runner and apply them strictly.

For the purposes of this analysis, I have defined a front runner to

be an animal for which the comment-in-running contained either of
the following:

made all
or made virtually all

This is very simplistic, and will miss many horses which could be
considered to be front runners. However, the definition is
unambiguous and can be applied without too much difficulty.

Using this definition about 12% of National Hunt races, in recent

seasons, have been won by front runners. This compares to 10%
in 1983/84, an increase of about 60 races. Analysing front
runners by going should indicate if it is easier to win from the
front on certain types of ground. Table 11 presents the number of
races won by front runners by going:

Table 11: Analysis of Front Runners by Going

Number of
Front Running Number
Going Winners of Races (%)

Hard 8 30 26.7
Firm 161 1035 15.6
Good to firm 350 2747 12.7
Good 377 3698 10.2
Good to soft 201 1961 10.2
Soft 209 1914 10.9
Heavy 134 851 15.7

All 1440 12236 11.8

41
From Table 11 we can see that the proportion of races won by
front runners does not vary appreciably (the figure for hard going
is based on a small sample). And after adjusting for the average
field size for each going category, the percentages become even
more uniform, leading us to conclude that the type of surface does
not effect the chances of a front runner. The conclusion is
different, however, when the race distance is considered. Table 12
shows the number of races won by front runners within race
distance:

Table 12: Analysis of Front Runners by Race Distance

Number of
Race Front Running Number
Distance Winners of Races (%)

16f..18f 676 5524 12.2

19f..21f 344 2821 12.2
22f..24f 267 2544 10.5
25f+ 150 1313 11.4

All 1437 12202 11.8

Although the percentages appear quite uniform, when they are

adjusted for the average number of runners within each race
distance category a different picture emerges:

Race Distance 16..18f 19..21f 22..24f 25f+

Adjusted % 17.3 15.6 14.5 13.8

It is not surprising to see that front runners win more often in the
shorter races. The problem remaining is to convert this
information into a profitable selection method.

At this point we need to determine how we will identify probable

future front runners. Our current definition only relates to the race
in which the horse is running, which is of no use for betting
purposes. From a betting perspective, we

42
need to identify before the race which horse(s) are likely to front
run. Naturally, an animal’s past race performances is the best
guide. Using the comments-in-running for previous race
performances the definition of a likely front runner could have
been stated as follows:

a horse is identified as a probable front runner if it won

its most recent outing and the terms made all or made
virtually all appeared in its comment-in-running.

Again, this is one of many definitions which could be used and has
been selected due to its simple, unambiguous form. Analysing all
horses which qualify as probable front runners for four seasons
produced the following results: 343 winners from 1225 bets, a
strike rate of 28%, with a loss of 17p per £1 staked. This is a very
poor result, however with a little refining we should be able to
turn this into a profit.

In the previous section we found that horses are able to carry big
weights more easily at particular courses. It is also true that front
runners are suited by particular types of racecourse, Table 13
illustrates this fact:

Table 13: Average Return and Strike Rate for

Probable Front Runners by Racecourse

Average
Course Winners (%) Runners Return/£1

Aintree 4 (13.3) 30 -0.52

Ascot 6 (14.3) 42 -0.49
Ayr 3 (15.0) 20 -0.72
Bangor 9 (32.1) 28 -0.17
Carlisle 9 (42.9) 21 +0.12
Cartmel 3 (16.7) 18 -0.72
Catterick 7 (46.7) 15 +1.13
Cheltenham 24 (19.8) 121 -0.30
Chepstow 5 (20.0) 25 -0.53
Doncaster 7 (35.0) 20 +0.75

43
 Exeter 11 (34.4) 32 -0.13
Fakenham 1 (9.1) 11 -0.59
Folkestone 1 (25.0) 4 -0.67
Fontwell 14 (45.2) 31 +0.08
Haydock 9 (23.7) 38 -0.49
Hereford 12 (40.0) 30 +0.32
Hexham 2 (40.0) 5 +0.21
Huntingdon 13 (33.3) 39 +0.13
Kelso 9 (42.9) 21 +0.03
Kempton 18 (32.7) 55 -0.11
Leicester 3 (21.4) 14 -0.52
Lingfield 4 (36.4) 11 -0.17
Ludlow 6 (27.3) 22 -0.53
Market Rasen 8 (25.8) 31 -0.26
Musselburgh 5 (71.4) 7 +1.05
Newbury 10 (23.8) 42 -0.10
Newcastle 4 (19.1) 21 -0.31
Newton Abbot 14 (34.2) 41 -0.31
Nottingham 5 (35.7) 14 -0.07
Perth 8 (47.1) 17 +0.38
Plumpton 13 (40.6) 32 +0.02
Sandown 14 (32.6) 43 +0.27
Sedgefield 6 (25.0) 24 -0.21
Southwell 2 (16.7) 12 -0.38
Stratford 8 (17.4) 46 -0.50
Taunton 6 (25.0) 24 -0.21
Towcester 3 (15.8) 19 -0.70
Uttoxeter 19 (33.3) 57 -0.07
Warwick 8 (25.8) 31 -0.39
Wetherby 8 (19.5) 41 -0.38
Wincanton 7 (28.0) 25 +0.06
Windsor 0 (0.0) 5 -1.00
Worcester 10 (27.0) 37 +0.05

All 343 28.0 1225 -0.17

From Table 13 the tracks which suit front runners can easily be
seen. The percentage strike rate ranges from 0% at Windsor to
71% at Musselburgh. Consequently, the front runners’ courses
can be classified by comparing the strike for each track with the
overall average strike rate. Setting a

44
threshold at 35%, all courses with a winners to runners strike rate
exceeding this threshold would be classified as good tracks for
front runners. (This is an arbitrary threshold and can be increased
or lowered to vary the analysis.) The courses which exceed the
35% threshold are listed below (excluding Nottingham which no
longer stages national Hunt racing):

Carlisle, Catterick, Doncaster, Fontwell, Hereford,

Hexham, Kelso, Lingfield, Musselburgh, Perth and
Plumpton.

Backing all qualifying horses at these tracks results in 90 winners

from 210 bets, a strike rate of 43%, with an average return per £1
staked of 29p. Therefore,

consider backing any horse classified as a probable

front runner at Carlisle, Catterick, Doncaster,
Fontwell, Hereford, Hexham, Kelso, Lingfield,
Musselburgh, Perth or Plumpton: expected profit
29p per £1 staked.

The above result is quite pleasing, however, it is possible to

improve both the strike rate and the average return. By restricting
the analysis to handicap races results in 55 winners from 135 bets,
a strike rate of 41%, with a return of 43p per £1 staked. We have
already seen that front runners are better suited by shorter distance
races, and by further restricting our bets to handicap races over 16-
18 furlongs improves the return to: 32 winners from 65 bets, a
strike rate of 49%, with an average return of 72p per £1 staked.

In summary, horses which have won races by front running are

worthy of close consideration. There are many advantages to front
running, not least the lead which can be

45
poached at the start of a race. Front runners are better suited to
shorter races (i.e. 16 to 18 furlongs), and by particular racecourse
configurations.

Sires

In Flat racing the influences of the sire, and to a certain extent the
dam, are used to provide clues to the likes and dislikes of the
progeny. For instance, as the Derby nears there is always
speculation in the trade press regarding the likely levels of stamina
possessed by the leading contenders based on the stamina
exhibited by their sires, dams and grandsires.

In National Hunt racing, the influence of the animal’s parents is

not such an important topic. This is because the horses tend to run
more often and, hence, their preferences can be determined directly
from the form. However, it is a worthwhile exercise to research
the influences of the sires of National Hunt horses, and generate
information which could be incorporated into betting systems. A
full examination of all National Hunt sires is beyond the scope of
this text, however, 60 of the top sires have been examined to
produce the following tables. The sires quoted are those for whom
the influence is significant for each analysis component.

Table 14 illustrates the influence of the sire on its progeny’s

preference for particular states of going.

46
 Table 14: Average Return and Strike Rate of Progeny’s Race
Performances by Going

Fast Ground Soft Ground

Sire (%) Return/£1 (%) Return/£1

Alias Smith (USA) 4.2 -0.72 12.8 +0.57

Black Minstrel 9.3 -0.46 15.4 -0.21
Blakeney 16.8 -0.02 11.6 -0.47
Broadsword (USA) 10.6 +0.11 4.4 -0.82
Buckskin (FR) 14.4 -0.20 20.5 +0.20
Crash Course 7.1 -0.72 12.6 -0.34
Deep Run 17.7 -0.03 11.6 -0.20
Dominion 11.3 -0.34 16.5 -0.28
Electric 15.6 -0.50 8.4 -0.50
Gunner B 3.8 -0.65 13.3 +0.04
Henbit (USA) 20.4 -0.00 10.8 -0.24
Julio Mariner 15.9 +0.02 7.4 -0.79
Kambalda 22.0 -0.02 12.1 -0.27
Kemal (FR) 11.7 +0.01 16.3 +0.01
Kris 17.5 -0.17 11.9 -0.27
Le Bavard (FR) 14.3 +0.18 6.3 -0.69
Nishapour (FR) 13.2 -0.33 6.6 -0.74
Over The River (FR) 15.4 -0.09 8.6 -0.29
Phardante (FR) 19.1 +0.05 10.7 -0.31
Strong Gale 19.8 +0.01 10.9 -0.39
Torus 16.1 +0.92 11.5 -0.43
True Song 13.2 -0.55 4.5 -0.21
Uncle Pokey 4.0 -0.81 15.6 -0.02

where fast ground is defined to be hard, firm or good to firm and soft
ground is defined to be soft or heavy going in the Official formbook.

There are remarkable differences between some of the percentages

presented in the above table. For instance, only 3.8% of the race
performances on fast going by the progeny of Gunner B have
resulted in wins, whereas on soft going this figure increases to
13.3%. It is also interesting to note that it would have been
profitable to back all progeny of several sires on either soft or fast
going. For example, a level £1 stake bet on all the runners sired
by Buckskin (FR) would have resulted in an average return of 20p
per £1. An analysis by race distance is equally as informative.
Table 15 presents this information:

47
 Table 15: Average Return and Strike Rate of Progeny’s Race
Performances by Race Distance

Race Distance in furlongs

Sire 16..20f 21..24f 25f+

Alias Smith (USA) % 3.5 9.6 17.1

Ret -0.49 -0.36 -0.16
Ardross % 18.9 16.8 12.9
Ret +0.19 +0.18 +0.11
Baron Blakeney % 9.8 9.2 1.8
Ret -0.43 -0.44 -0.90
Black Minstrel % 13.4 11.8 4.9
Ret -0.05 -0.27 -0.76
Derring Rose % 4.0 12.0 17.7
Ret -0.84 -0.36 -0.14
Dominion % 14.7 13.0 7.1
Ret -0.29 -0.08 -0.76
Furry Glen % 15.2 13.4 10.1
Ret -0.23 -0.27 -0.38
Green Shoon % 8.1 9.6 15.4
Ret -0.46 -0.43 +0.05
Gunner B % 10.9 2.6 0.0
Ret -0.16 -0.78 -1.00
Le Bavard (FR) % 7.3 9.2 11.7
Ret -0.40 -0.29 -0.43
Oats % 12.5 14.0 21.5
Ret -0.16 -0.16 +0.05
Roselier (FR) % 12.5 14.4 20.7
Ret -0.08 -0.26 +0.50

where Ret denotes the average return per £1 staked.

Based on the above data, it is clear that stamina traits are passed
from sire to offspring. For example, the progeny of Baron
Blakeney do not appear suited by extreme distances in excess of 3
miles. In contrast, the runners sired by Oats seem to excel in long
distance races.

A third and slightly more obscure analysis, concerns course

directions. Some animals prefer to run in a particular direction,
either right- or left-handed. The following table seems to indicate
that this preference can be inherited from the sire.

Table 16: Average Return and Strike Rate of Progeny’s Race

Performances by Course Direction

48
 Course Direction
Left-Handed Right-Handed
Sire (%) Return/£1 (%) Return/£1

Absalom 7.8 -0.69 12.4 -0.44

Ardross 20.3 +0.34 11.4 -0.25
Black Minstrel 9.8 -0.40 15.0 +0.03
Blakeney 9.9 -0.43 15.3 -0.35
Cruise Missile 2.3 -0.57 7.4 -0.69
Furry Glen 15.1 -0.23 11.4 -0.33
Green Shoon 12.4 -0.16 7.8 -0.57
Gunner B 11.0 -0.15 5.3 -0.57
Kemal (FR) 14.4 -0.18 9.8 -0.42
Le Moss 15.4 +0.09 12.6 -0.23
Roselier (FR) 12.7 -0.15 16.5 +0.05
Scallywag 11.0 -0.42 6.8 -0.53
The Parson 15.2 -0.19 11.0 -0.29
True Song 7.4 -0.63 12.2 -0.25
Uncle Pokey 6.8 -0.55 10.9 -0.20

Due to the large number of race performances analysed (over

100,000) all the above differences in strike rate between the
course directions are significant. The most striking difference is
that for Ardross. A winners to runners strike rate of 20.4% on
left-handed courses compared to only 11.4% on right-handed
tracks.

The above tables illustrate the traits which can be inherited by

horses from their sires. Whilst this information alone cannot be
used to formulate a profitable betting strategy, it should be
considered as a component of conventional race analysis,
especially for previously unraced horses, and to supplement other
systematic selection methods.

49
The Time Factor

In National Hunt racing, previous race times are not generally

considered to be of primary importance when analysing races due
to the problematic nature of incorporating the data into a sound
analytical formula. There are many difficulties associated with
using race times to assist in race analysis, specifically the
unreliable nature of the going assessments and the variation in the
racecourse configurations across the country. However, it is still
possible to devise selection methods which rely on previously
achieved race times which are worthy of consideration.

The variability of racecourse configuration can, to some extent, be

nullified by the introduction of a race standard time. Both the
trade papers, as well as Timeform and Raceform, produce standard
times for each race distance on every course in Great Britain.
Such times are supposed to represent the likely time a horse rated
135 by the BHB would take to run the course on good to firm
going. Using these standards facilitates the comparison of race
times produced by different horses on different tracks. However it
does not take account of varying ground conditions.

Naturally, races run on different ground conditions will produce

different times. For instance, races on soft or heavy ground can
result in times over 50 seconds slower than the standard, whereas
those on a fast surface can be run in times significantly faster than
the standard. Clearly, the accurate comparison of such race
performances is not an easy problem to overcome.

The calculation of a going allowance, based on the races staged on

the course on a particular day, is often suggested as a solution to
this problem. This figure can then be used to normalise for the
effect of the going at different courses and

50
facilitate the comparison of the race performances. To determine
the going allowance, the differences between the actual race times
and the standard times are first divided by the race distances and
then averaged after any freak figures have been removed.

Therefore, for a six race card, the times may have been as follows:

Table 17: Calculation of Going Allowance

Race Race Standard Diff. per

Dist. Time (s) Time (s) Diff. (s) Furlong (s)

16.5 259.5 237.0 22.5 1.36

20.5 314.5 284.0 30.5 1.49
19.5 316.5 283.0 33.5 1.72
16.5 265.0 228.0 37.0 2.24
24.0 406.4 337.0 69.4 2.89
16.5 260.6 228.0 32.6 1.98

It could be argued that the time for the fifth race is an outlier and
should be removed from the going allowance calculation due to its
extreme value. Consequently, the going allowance becomes the
average of the other five times, namely 1.76 seconds per furlong.
In fact, Nick Mordin (in Betting For A Living) generally takes the
average of the three fastest times excluding any outliers, although
there are no hard and fast rules governing the number of races you
use.

Once the going allowance has been calculated it is possible to

calculate a speed figure for any runner on the card. Firstly, it is
necessary to adjust the time of the winner by the going allowance,
to produce a normalised time. Therefore, if the going allowance
indicates that the track is riding faster than the norm, the winning
time is increased accordingly. For instance, if, in a 16 furlong
hurdle race, the track is found to be riding 0.1 seconds per furlong
faster than the standard, the

51
time of the winner should be increased by 1.6 seconds (i.e. 16 x
0.1). This new time being an estimate of the time the horse would
have recorded on average going.

To convert the time to a rating it is necessary to consider the

standard race time for the distance. At this point, it is necessary to
set the scale for the ratings. Normally, a horse producing a time
equivalent to the standard would be given a rating of 100.
However, any figure could be used to determine the rating scale.
The next step is to determine the difference between the standard
time and the adjusted time, in seconds. This figure is multiplied
by 5 (lengths run per second). The resultant figure is either added
to, or subtracted from, 100 depending on whether the adjusted
time is faster, or slower than the standard time. Finally, the figure
should be adjusted for distance beaten by the winner. For a more
detailed description of calculating speed figures see Betting For A
Living by Nick Mordin.

Whilst this method for generating speed ratings is well founded in

theory, it relies heavily on two potentially unreliable components,
namely the standard time and going allowance. In fact, the actual
race time itself is far from perfect, depending as it does, in the
majority of cases, on hand-timing.

The following method for selecting horses with winning potential

utilises the race time and standard time but not the going
allowance, which, due to its approximate method of calculation, is
the component subject to the highest degree of error.

The selection method is based on the horses which produce the

best race times of the day at each racecourse. The best time of the
day is simply the fastest time relative to the standards. For
example, given the six races:

52
 Difference from
Race Number Standard Time (s)
1 +11.8
2 +8.0
3 +2.9
4 +6.2
5 +7.9
6 +5.0

the best race time of the day was recorded in race 3 since this time
is closest to the standard and all race times exceeded their
comparable standards. In the next example, the best race time was
recorded in race 4, the race which produced a time 3.9 seconds
faster than the standard (hence the minus sign). The next best time
of the day would be race 1 at 1.1 seconds under the standard.

Difference from
Race Number Standard Time
1 -1.1
2 +5.5
3 +2.6
4 -3.9
5 +9.9
6 +11.0

The following table charts the performance of horses which win

the race producing the best time of the day, on their next outing:

Table 18: Strike Rate and Average Return for Good Time
Performers on Their Next Run

†
Time Average
Difference Winners (%) Runners Return/£1

0.00-4.00 269 (29.6) 910 -0.01

4.01-5.00 39 (29.3) 133 -0.11
5.01-6.00 25 (31.6) 79 +0.14
6.01-7.00 22 (36.7) 60 +0.48

53
 7.01-8.00 18 (42.9) 42 +0.08
8.01+ 24 (31.2) 77 +0.20

Total 397 (30.5) 1301 +0.03

†
the time difference is the number of seconds between the best time
(the qualifying time) and the next best time of the day after adjusting by
the standard times.

Clearly, there is some merit in this approach. The strike rate of

over 30% is more than adequate, however, an average return of
just 3p per £1 staked would translate to a loss for off-course
punters. An obvious modification is to restrict the selections to
those horses which win races with adjusted time differences of
over 5 seconds faster than the next best time of the day. This
would result in 89 winners from 258 bets, a strike rate of 34.5%,
with an average return of 23p per £1 staked. Therefore,

consider backing the winner of the fastest race of

the day, providing the adjusted time difference was
more than 5 seconds faster than the next best time
of the day, on its next run: expected profit 23p per
£1 staked.

It would be logical to conclude that horses in the highest time

categories should be the best animals to follow due to their
obvious superiority over the other runners at the course. However,
the average return for the two highest time categories (i.e. 7.01-8.0
and 8.01+) is, in fact, inferior to the 6.01-7.0 seconds category.
One reason for this anomaly is that horses which produce
outstanding times, record times etc., are well reported in the press
and are brought to the attention of punters and bookmakers. As a
result subsequent starting prices for these horses will be reduced
to reflect this

54
good performance. An alternative explanation is that such race
times indicate errors in the standard time for the course.

An analysis of the 258 selections by race grade should help to

improve the average return. Table 19 presents this analysis:

Table 19: Strike Rate and Average Return for Good Time
Performers on Their Next Run by Race Grade

†
Race Average
Grade Winners (%) Runners Return/£1

Cond. Hdl. 4 (66.7) 6 +2.64

Clm. Hdl. 2 (66.7) 3 +1.67
Nov. Hdl. 14 (38.9) 36 +0.37
Sell. Hdl. 4 (44.4) 9 +2.89
Hcap. Hdl 16 (24.6) 65 -0.01
Hunter Chs. 0 (0.0) 3 -1.00
Cond. Chs. 2 (33.3) 6 -0.48
Clm. Chs. 0 (0.0) 1 -1.00
Nov. Chs. 22 (46.8) 47 +0.23
Sell. Chs. 0 (-) 0 -
Hcap.Chs. 25 (30.5) 82 -0.06

Total 89 (34.5) 258 +0.23

†
the race grade relates to the race type in which the good time was
recorded.

From Table 18 it can be seen that non handicap races seem to

provide a better guide than handicaps. By restricting the analysis
to non handicaps improves the average return to 58p per £1
staked, although the number of bets is reduced.

In this section we have seen how difficult it is to compare times of

different races due to the effect of the going and the wide variety of
racecourse configurations. However, it is still possible to develop
profitable selection methods using race times and race standards.
The method outlined highlights

55
horses which have produced significantly better race times than the
other runners on the same card. Whilst these runners are worth
following, the level of profitability does not increase linearly with
the time superiority. This fact should be remembered when
considering backing horses which have produced record times in
the past.

Developing Profitable Selection Methods

This final section of Part II is devoted to providing the reader with

a set of guidelines for developing, testing and implementing
profitable horseracing selection methods.

Development Stage

Simplicity - When designing a selection method it is important

not to make it too complex. As the number of variables increases
(i.e. distance beaten last time, going last time, going today, race
distance last time, race distance today, days since last run etc.) the
amount of data needed to give meaningful results increases.
Complex systems are also difficult to implement, requiring a great
deal of searching through previous form to determine the selection.

Uniqueness - Try to find new relationships between the data.

Using the normal data items in conventional fashion will, in all
probability, produce losing systems. The odds on offer account
for the established trends, and consequently there is no value for
the punter. For example, Table 6 illustrates the relationship
between average return and the distance the horse was beaten on
its latest run. For each category of distance beaten within the table
the average return is negative. This trend is well known and is
fully accounted for in the bookmakers’ odds.

56
Logic - When searching for unique approaches it is imperative to
ensure that the variables used are sensible. For instance, whilst
backing all horses ridden by bearded jockeys may have returned a
profit in the past, facial hair can hardly be considered a reliable
discriminating factor with regard to the chances of racing success.

Correlation - Be wary of ill-founded relationships and hidden

correlations. As an example, horses running in August tend to
produce faster race times than those running in February, therefore
they must be better animals. Whilst this conclusion seems a
logical deduction from the available evidence, it is an
unconsidered variable which produces this result, namely the
going. Faster going produces faster race times and the ground is
likely to ride firmer in August than February, hence the faster
times.

Trainers and Jockeys - Basing systems on trainer and/or jockeys

is not particularly reliable for the reasons highlighted in
Correlation. There is a great deal of correlation between trainers,
jockeys and owners: the best jockeys ride for the best trainers and
the most dominant owners tend to have their horses ridden by the
best jockeys. Consequently, a very profitable selection method
can suddenly become very costly if a main owner removes his/her
horses from a trainer, or a riding arrangement is terminated.

Measurement - Base the selection methods on quantifiable

variables, for example data items such as historical wins to runs
strike rate of the horse. Try to avoid qualitative items such as
suitability of going, distance etc., unless these terms can be
defined precisely. Opinion is difficult to analyse, and it is far from
constant. If you feel that suitability of the going is important, then
define exactly what constitutes suitable and unsuitable going for a
horse. For instance, the following definition could be used for
going suitability:

57
 the going may be assessed as suitable if and only if the
horse has won on an identical surface in the past ten
runs.

However, this in itself will introduce further problems. Using this

definition discriminates, to a certain extent, between horses which
have won in the past ten runs and those which have not.
Consequently, an underlying relationship has already been
introduced.

Data Collection - There are three issues concerned with data

collection: quality, quantity and spread. Clearly, the data collected
needs to be accurate and unbiased. This seems obvious, but it is
not necessarily a straightforward task to undertake, especially
when extracting data manually from a formbook. It is very easy to
omit particular races during the extraction process reasoning:
‘Well, I wouldn’t have bet in that race anyway’. Naturally,
sufficient data needs to be collected to support the complexity of
the system. However, too many seasons’ data can introduce bias
due to the ongoing changing nature of racing. For instance, in Flat
handicaps, horses carrying large penalties were good betting
opportunities, however the introduction of the five day entry
system in 1988 ruined this method. Paradoxically, it is not
advisable to extract data from just one season. Certain conditions
of the year, weather, quality of horse etc., will bias any analysis.
The data needs to be spread evenly over a few recent seasons.

Long Shots - In general, I do not like systems which select long

priced winners for two reasons. Developing approaches based on
long priced horses requires huge amounts of back data due to the
high variance of the average return statistic. In fact, if a selection
method has selected a long priced winner (i.e. 20/1 or greater) I
remove it from the analysis. It only takes one or two of these
selections to turn a very poor

58
system into a seemingly profitable method. Furthermore, when
running the system these selections are often ignored, especially
during long losing runs.

Strike rate - Aim for an approach with a reasonable winner to

bets strike rate. A strike rate of just 10% is not really high
enough, the losing runs will occur too frequently and could be
extremely long. Aim for a strike rate of at least 20%.

Testing Stage

There are two different ways of testing a new selection method.

The first involves partitioning the data set before the development
process. One part can be used for development and the other for
testing. For instance, if the data have been collected from four
seasons, the first three seasons could be used to develop and refine
the system with the fourth season used for testing. The second
method, and the one I prefer, is to test the system live. In other
words, develop the approach on all available and relevant data,
and conduct the testing phase during the following year by either
keeping a record of the system’s performance or by actually
backing the selections to small stakes. In addition to providing a
real test environment, this approach is also one of education for
the punter. By closely monitoring the system through the
placement of bets, the punter will get a better feel for the method
and perhaps identify other highly correlated variables. For
instance the likelihood of long losing runs, the way he/she reacts
after several losing bets etc. Once satisfied with the validity of the
system the punter can then increase the stakes.

Running The System

It is very easy to break the rules of the system, especially during

long losing runs. This behaviour should be avoided. Providing
the system has been based on well-founded ideas,

59
and has been adequately tested, there is no need to doubt its ability
to return a profit. Consequently, the devised staking plan should
be adhered to. There is nothing worse than finding, at the end of
the year, that you have lost money, and that if you had kept to the
system’s rules you would have made a profit.

60
 CONCLUDING REMARKS

Throughout this book I have attempted to show that whilst it is

possible to make a profit from Jumps racing, it does not
necessarily come from the conventional approach to winner
finding. Simply backing a high proportion of winners does not
ensure a long term profit.

The conventional approach to race analysis requires the punter to

determine, amongst numerous other factors, whether the selection
is suited by the race distance, going, track configuration etc., all of
which have been considered by the bookmakers before pricing up
the race. Consequently, punters and bookmakers are working with
the same data, and if they both adopt similar approaches to
solving the race surely the punter must come off worse in the long
run. A good example concerns the selection method based on
time. Horses which produce outstanding times are not as
profitable to follow as those which merely produce good times
since this fact is widely reported and reflected in subsequent
starting prices. A good, but not record breaking time, is often
overlooked by bookmakers and not incorporated into the price
assessment.

So how can the punter make a profit? Well, there are three
options. The first requires the punter to become so knowledgeable
about horseracing that his/her assessment of the true probabilities
is vastly superior to the bookmakers’, to bet only on-course
(avoiding tax) and always get the best price. This is asking a great
deal, but it is achievable. The second option is to acquire
information that is unknown to the public and bookmakers. For
instance, to know when a fancied horse is not trying. However, I
expect that such information would not come cheaply. The third
option is to adopt a different perspective on race analysis, and use
information which is not normally considered, to form alternative
selection methods which return long term profits.

Over many years I have tried the first and second approaches to
backing horses, although recently I have relied almost entirely on
the systematic approach. Occasionally, tempted by the sound
theoretical basis of conventional race analysis, I attempt to study

61
form. However, I usually regret these lapses.

Whether you choose to adopt a systematic approach to betting or

not I hope that you have found some worthwhile material in this
text to incorporate into your betting strategies.

62