See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/342976689

Context based Adoption of Ranking and Indexing Measures for Cricket Team
Ranks

Article in Computers, Materials & Continua · July 2020

DOI: 10.32604/cmc.2020.010789

CITATIONS READS
2 839

7 authors, including:

Raja Sher Afgun Usmani Muhammad Sher Ramzan

BlueBlaze.Earth King Abdulaziz University
29 PUBLICATIONS 513 CITATIONS 151 PUBLICATIONS 2,998 CITATIONS

SEE PROFILE SEE PROFILE

Anwar Ghani Imran Khan

International Islamic University, Islamabad International Islamic University, Islamabad
85 PUBLICATIONS 1,449 CITATIONS 31 PUBLICATIONS 753 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Anwar Ghani on 23 August 2020.

The user has requested enhancement of the downloaded file.

Computers, Materials & Continua CMC, vol.65, no.2, pp.1113-1136, 2020

Context Based Adoption of Ranking and Indexing Measures for

Cricket Team Ranks

Raja Sher Afgun Usmani1, Syed Muhammad Saqlain Shah 1, *, Muhammad Sher
Ramzan2, Abdullah Saad AL-Malaise AL-Ghamdi2, Anwar Ghani1, Imran Khan1
and Farrukh Saleem2

Abstract: There is an international cricket governing body that ranks the expertise of all
the cricket playing nations, known as the International Cricket Council (ICC). The
ranking system followed by the ICC relies on the winnings and defeats of the teams. The
model used by the ICC to implement rankings is deficient in certain key respects. It
ignores key factors like winning margin and strength of the opposition. Various measures
of the ranking concept are presented in this research. The proposed methods adopt the
concepts of h-Index and PageRank for presenting more comprehensive ranking metrics.
The proposed approaches not only rank the teams on their losing/winning stats but also
take into consideration the margin of winning and the quality of the opposition. Three
cricket team ranking techniques are presented i.e., (1) Cricket Team-Index (ct-index), (2)
Cricket Team Rank (CTR) and (3) Weighted Cricket Team Rank (WCTR). The proposed
metrics are validated through the collection of cricket dataset, extracted from Cricinfo,
having instances for all the three formats of the game i.e., T20 International (T20i), One
Day International (ODI) and Test matches. The comparative analysis between the
proposed and existing techniques, for all the three formats, is presented as well.

Keywords: Indexing, ranking, cricket teams, runs and wickets.

1 Introduction
Since the very first game played in the history of sports, it was important to keep the
records of the competition. In history, various ranking systems have been used to rank
teams and players, but more comprehensive sports ranking systems have been around for
nearly 80 years. During the early years of technical ranking, calculations were rendered on
paper rather than on a machine. Rating systems use a variety of methods to rank teams, and
the most widely used criterion is power ranking. Power rating is the power of the team
relative to the teams competing in the same division or league. In power rating, an analyst

1
Department of CS&SE, International Islamic University, H-10 Sector, Islamabad, 44000, Pakistan.
2
Department of Information System, Faculty of Computing and Information Technology, King Abdulaziz
University, Jeddah, 21589, Saudi Arabia.
*
Corresponding Author: Syed Muhammad Saqlain Shah. Email: [email protected].
Received: 29 March 2020; Accepted: 14 July 2020.

CMC. doi:10.32604/cmc.2020.010789 www.techscience.com/journal/cmc

1114 CMC, vol.65, no.2, pp.1113-1136, 2020

attempts to find a transitive relationship in a given dataset. For instance, if Team One wins
over Team Two and Team Two wins over Team Three, it can be said that: (team one>team
two>team three). However, complications may occur while relying on a system that is
completely dependent upon winning and losing. If Team Three wins a game played against
Team One, the relationship in the data is intransitive, as (team one>team two>team
three>team one) and if it is the only data available, violation in ranking may take place.
Situations like this may repetitively prevail in sports and need to be tackled.
The International Cricket Council (ICC) is the governing body of crickets. In the past, it
managed team ranking for all cricket playing nations, using an impromptu system that is
based primarily on winning and losing. The ICC ranking process was simply a system
that was used to regulate all international cricket matches on a regular schedule.
Afterward, a new concept was introduced and implemented, all the teams are assigned a
certain amount of points based on their opponent's performance as well as the result of
the match. The ICC implemented the idea to ensure that such dead rubbers still have
some significance. In the past, if a team won the first three matches of a five-game series,
they didn’t have much to play for in the final two. The series was decided and there was
no advantage in winning it 5-0 compared to 3-2. Teams could rest key players and give
inexperienced players a chance to get away. But now, the 5-0 series gives them more
points and more opportunities to move up the points table.
This paper proposes the Cricket Team Index for the cricket team ranking, adoption of the
h-index [Hirsch (2005)]. The h-index is a state-of-the-art indexing strategy that is used to
measure the productivity and citation impact of scholars, based on their most cited work
i.e., the research papers and the number of citations received in other publications. This
paper maps the citations, used in the h-index, to the winning margin in terms of the
number of wickets and runs. Higher the average citations, the higher h-index. Therefore,
the higher the winning margin, the higher it should be ranked. This paper argues that a
single run can’t be worth a wicket, so the average wicket worth is computed using batting
statistics from the last two years. The ct-index only considers the statistical figures of the
wickets and runs in terms of winning margin, but the strength of the opponent teams is
neglected. To measure the rank and strength of the team, this paper proposes the cricket
team rank (CTR) which is an adoption of the PageRank [Page, Brin, Motwani et al.
(1999)]. Instinctively, the more matches a team wins in competition to a stronger team,
the higher its rank will be. Team Ranking observes the strength and impotence of teams
while neglecting the numeric figures of runs and wickets by which a match is won. The
third proposed technique is Weighted Cricket Team Rank (WCTR) which is also a
modification of PageRank like CTR but it includes weight considering the figures of runs
and wickets by which a match is won.
The rest of the paper is organized as follows: Section two presents a literature review,
section three discusses the current ranking methods and the proposed methods in more
detail, and section four provides the data which was used to experiment and the results of
the experiments. Section five provides a discussion and brief analysis of the results while
section six concludes the presented research.
Context Based Adoption of Ranking and Indexing Measures 1115

2 Related work
As sport is such a finely tuned competitive endeavor, and because multiple millions of
dollars can be connected to just one match, the task of accurate team ranking is critical.
By relying upon outdated ranking techniques, the rankings are not reliable. The h-index
[Hirsch (2005)] and PageRank [Farooq, Khan, Malik et al. (2016); Page, Brin, Motwani
et al. (1999)] approaches are more modern and deliver more reliable results. The ranking
is a practice used for almost all sports, and different methods for producing ranks are
presented in the past. Looking at the batsman’s performance using a parametric control
chart, Bracewell and Ruggiero documented interesting outcomes [Bracewell and
Ruggiero (2009)]. Qader et al. [Qader, Zaidan, Zaidan et al. (2017)] presented a technique
for ranking football players. They used multiple criteria for decision making, i.e., 12 tests
belonging to the three categories (five fitness, three anthropometrics, and four skill tests).
As test data, twenty-four players from U17 were taken, and the results were similar to the
existing system. Applying a social network analysis, Duch et al. [Duch, Waitzman and
Amaral (2010)] created a method of ranking an individual soccer player. Previous
researchers have attempted to use PageRank for delivering a reliable ranking of various
cricket teams [Mukherjee (2012)] and/or cricket players, but these attempts did not
harness the power the h-index brings to ranking teams, nor did they employ a graphical or
non-graphical evaluation routine to calculate the values of runs and wickets. Mukherjee
[Mukherjee (2012)] concluded that there is no real way to accurately determine rank
based on only the number of wins. Quality of a win is also important in creating a metric
to analyze a team’s strength in play. Using the PageRank algorithm, the author created a
formula better to understand the strength of a team and its captain. Likewise, Borooah et
al. [Borooah and Mangan (2010)] see that the existing traditional ranking system has
several drawbacks. When creating a batsman ranking system that relies on the batting
average alone, the system does not take into account the time factor throughout the
matches. A batsman with a consistent scoring of lower value might fare better, at least
temporarily, than a batsman who has a typically high average but suffers from a rough
patch. In the current system, authors claim that the runs a player scores for his team are
entirely discounted, and should be represented with value. The proposed research is an
attempt to resolve these perceived flaws. Amin et al. [Amin and Sharma (2014)]
presented a cricket batsman ranking mechanism for the Indian Premier League (IPL). The
authors adopted ordered weighted averaging (OWA) parameter by using the highest score,
batting average, strike rate, number of fours, and sixes hit by the batsman. The OWA
score was subject to regression for the final ranking of the player. Pradhan et al. [Pradhan,
Paul, Maheswari et al. (2017)] argued that h-index and popular adaptations were good at
ranking highly-cited authors but not much successful in resolving the ties in between
medium and low cited authors. As the majority of the authors comes under the category
of low-medium cited ones, they proposed methodology, C3-index, to resolve the ties
between low and medium cited categories and predicting the future rankings of the
authors during their earlier career. It was shown that the proposed C3-index remained
more consistent and efficient than h-index and its well-known adoptions. Citation-based
metrics like Relative Citation Ratio (RCR) are used as alternate ranking techniques to
different PageRank adoptions. Dadelo et al. [Dadelo, Turskis, Zavadskas et al. (2014)]
argued that current Basketball player ranking systems lack objectivity as they use
1116 CMC, vol.65, no.2, pp.1113-1136, 2020

situational factors (performance statistics) of the game. They proposed a multi-criteria

systematic solution that uses the Technique for Order of Preference by Similarity to Ideal
Solution (TOPSIS) and design principles for algorithm based on the method. Mukherjee
[Mukherjee (2014)] argued that the rating of bowlers and batsmen in cricket is done by
their wickets and runs averages respectively but it does not take into account the ‘quality’
of those statistics i.e., rank of the batsman dismissed or rank of bowler against which a
batsman scored. He proposed a refined method to quantify the ‘quality’ of the statistics
used for ranking by using the application of Social Network Analysis (SNA) to rate the
players in team performance. Min et al. [Min, Kim, Choe et al. (2008)] presented a
methodology to predict the outcome of a match by combining the Bayesian classifier
along with rule-based reasoning. They have combined them by looking into the fact that
the results are not only stochastic but the team planning can be represented by the rules.
They tested their framework over the football matches and called that system as Football
Result Expert System.

2.1 H-index and it’s significant extensions

When attempting to measure the value and productivity of a researcher or scientist against the
broader scientific community, it was complicated because there were both quantitative
comparisons and qualitative comparisons to be made. The h-index [Hirsch (2005)] was
designed to accommodate that exact situation, and allow for an evaluation of the number of
papers by a specific scientist, as well as their impact on the field. The impact question was
studied by looking at the number of citations of that scientist’s work among others. The h-
index is a useful tool, but when the data sets reach a certain high level of reference, there are
unreliable results. Egghe [Egghe (2006)] created the g-index to compensate for this flaw.
However, both indexing techniques, i.e., g-index and h-index, compromised the calculated
time in which the paper was published and received the citation. As a result, Burrell [Burrell
(2007)] brought forth the m-quotient, which included the career length in the current h-index,
worked through dividing the h-index value by the total time of his research activity. But this
technique has not yet been implemented in other areas. Adding to the field of study, Daud et
al. [Daud, Muhammad, Dawood et al. (2015)] proposed the t-index. By adopting the h-index,
they created a weighted environment in which they could consider the values of the runs and
the wickets by which a match is won. However, the t-index uses the same weight for both
runs and wickets, which is ineffective. Even laypeople are aware that wickets should be given
a higher weight than should runs.
Nykl et al. [Nykl, Campr and Ježek (2015)] proposed a personalized method to rank
authors of scientific papers using the journal values. They used the adoption of the
PageRank algorithm as well as other popular measures like h-index, citation count,
publication count, and publication’s author count to rank authors. Pérez-Rosés et al.
[Pérez-Rosés, Sebé and Ribó (2016)] proposed authority score computation method to
rank profiles using their skills and endorsements. The authors’ calculated authority score
by keeping in view the relations between different skills and ranking is done using a
PageRank algorithm on weighted graphs generated against different skills. Degree and
other centrality-based heuristics are commonly used in literature to estimate the impact of
individuals on social media, Zhang et al. [Zhang, Wang, Jin et al. (2015)] argued that
Context Based Adoption of Ranking and Indexing Measures 1117

these techniques have major design flaws and proposed heuristic scheme based on
PageRank to maximize the impact on social media.

2.2 PageRank and it’s major extensions

PageRank was developed as a means by which web pages could be evaluated
comparatively. PageRank is tasked with calculating the relative strength of a web page,
ignoring the frequency with which the page is requested. Frequency of requests can be
important; however, so Haveliwala [Haveliwala (2002)] created a modified version in
2002. In the modification, the value of the page was read as a relationship between the
page and its linked pages. Manaskasemsak et al. [Manaskasemsak, Rungsawang and
Yamana (2011)] proposed the PageRank Time Weighted approach to calculate the impact
of the page over time. For instance, the factor of how new the page is in relationship to
other pages, events such as special events are respected as drivers and trends allow for the
incorporation of revision counts in the evaluation of a page’s impact.
The PageRank algorithm helps the world monetize and commoditize the web page. PageRank
helps to create a metric by which web site owners or buyers can value a web page. Text and
word ranking are used by many researchers in various fields [Gao, Wang and Chen (2019);
Xiang, Wu, Li et al. (2018)]. Features, attributes and Z-number ranking based techniques are
presented as well [Ezadi and Allahviranloob (2018); Wang, Ren, Davis et al. (2017); Yeh
(2018)]. TextRank, an adoption of PageRank, can even evaluate the data in natural language
and can extract keywords and phrases from the various documents [Mihalcea and Tarau
(2004)]. It is concluded that the results obtained were an accurate reflection of the
proportionate results expected. In addition, Haley [Haley (2016)] explored the negative
impacts of h-index ranking for a scholar keeping in mind the uncertainty, insurance, and
lobbies involved. The author provided an economist’s views on how the scholar h-index
impact faculty promotions, awards, and other incentives. Cerchiello et al. [Cerchiello and
Giudici (2014)] argued that significant research is done using h-index, but few have studied
its statistical properties and implications. In order to focus on said issues, the author proposed
a statistical approach to come up with h-index distribution by focusing on two major h-index
components, i.e., the total papers produced and count vector of their citation, by introducing
convolution models. Springer [Springer (2016)] proposed h-Index adoption to assess the
impact of science, technology, and engineering (STE) on an institution to provide a basis for
funding. Iván et al. [Iván and Grolmusz (2011)] took the framework on a protein interaction
network. Daud et al. [Daud, Muhammad, Dawood et al. (2015)] suggested a Team Rank (TR)
and Weighted Team Rank (WTR) approaches. The TR approach was only resulting in false
points for teams that win more, with teams that winless being “dampened”. Another way to
illustrate the problem: If a team plays two matches and wins one match, do they not deserve a
better result than a ranking of 0.5? If a team plays fifty games and wins twenty-five, they will
be ranked the same way, with a 0.5 ranking. This is inherently unfair, even though it is
mathematically correct under the ranking system.

2.3 ICC cricket teams ranking system

International Cricket Council (ICC) is using the following system of rating formulae for
ODI and Test matches, which determines a champion and provides team rankings.
1118 CMC, vol.65, no.2, pp.1113-1136, 2020

2.3.1 ODI matches

Before the match, the ODI Method looks at the points of the teams in play. If they are
different by less than forty points, the match will award fifty bonus points plus the
amount of the losing team’s ranking, to the winner. In the case of a tie, the teams will
each be awarded the opposing team’s rank score. If the two team’s rank scores are
different by more than forty points, the leading team will be awarded ten bonus points,
plus the losing team’s ranking score, should they win. Should the leading team lose, they
will be penalized ninety points. Likewise, should the lower-ranked team win, they will be
awarded ninety points. Should the trailing team lose, they will be penalized with ninety
score ratings. Should there be a tie, the leading (or stronger) team is penalized forty
points, and, the trailing (or weaker) team is awarded forty points.
The new scores from the match are added to the team’s pre-match ranking. The number of
matches played is updated for each time, dropping off any matches and points that accrue
from a period longer than three years in the past. For each match, the winning team is
awarded one extra point for winning. Should there be a draw, one-half point is awarded to
each team. The updated score is then divided by the updated match score. The rating is thus
updated for each time. The teams are organized by strength using this rating scale.

2.3.2 Test matches

If the pre-match scores of the two teams are different by less than forty points, then the
analyst will multiply the series scores by fifty, adding that number to the winning team’s
series result times fifty points less than the opponent’s score. If the two teams’ scores are
different by more than forty points, the winning team’s series result is multiplied by the
stronger team’s score plus ten points. That result is then added to the opposing team’s
series score, which total is then multiplied by the team’s rating less ninety points. For the
weaker team, a similar formula is used, adding ninety points to the team’s ranking, then
adding that total to the opposing team’s series score and multiplying by the team’s rating
minus ten points. The new scores are updated in the rankings when they are added to the
pre-match scores. Matches and points that are outside of the three-year threshold are
removed. The Match Total is updated by incrementing the number of matches in the
series by one. Then the updated scores are divided by the updated match totals. For each
match, the winning team is awarded one extra point for winning. Should there be a draw,
one-half point is awarded to each team. The team that wins the series gets a bonus point.
If the series ends on a draw, then each team is awarded a half a point.

2.4 Research limitations

From the above presented existing techniques, it may be observed that there is still room for
research in cricket teams ranking. One of the reference research by Daud et al. [Daud,
Muhammad, Dawood et al. (2015)] presented their criteria for cricket team ranking i.e., t-
index, TR and WTR. We have already identified some inconsistencies [Daud, Muhammad,
Dawood et al. (2015)] and presented corrections for the observed inconsistency [Saqlain
and Usmani (2017)]. Even now, there are several limitations to the cricket team ranking
criteria, and the objective of the presented research is to focus on those limitations. A list of
the limitations in the existing team ranking methodologies is as follows:
Context Based Adoption of Ranking and Indexing Measures 1119

• t-index gives the same weight to winning margins in terms of runs and wickets,
which is not correct as winning by seven runs is a close margin while winning by
seven wickets is a considerable margin. This needs to be tackled.
• In case of presenting TeamRank (TR) technique, authors [Daud, Muhammad,
Dawood et al. (2015)] used a constant damping factor for all the teams and only
considered the winning ratio of a team A against B to winning ratios of other teams
against B. TR does involve how many matches B played against A and the other
teams. This gives false results as if B is a newer team which, if has not played the
matches ultimately did not lose many matches from other teams. Winning from such
a team A should not be awarded maximum reward.
• While presenting their techniques of WTR and UWTR [Daud, Muhammad, Dawood
et al. (2015)] did the same as said above that they did not use the strength of the
opponent team along with winning margins in a proper manner.
• ICC ranks the teams based on the strength of the teams, and they do not count the
winning margin while ranking the teams.

3 Proposed measurements for cricket team ranking

In this section three rank measuring techniques for the cricket teams are proposed i.e., (1)
Cricket Team Index (ct-index), (2) Cricket Team Rank (CTR), (3) Weighted Cricket
Team Rank (WCTR). The proposed ranking measurements are adoptions of standard h-
index [Hirsch (2005)] and PageRank algorithm [Page, Brin, Motwani et al. (1999)].

3.1 Measuring team ranks through cricket team-index (ct-index)

Classifying the productivity of a scientist or researcher based on their relative importance
in the field was subjective until Hirsch [Hirsch (2005)] proposed the h-index in 2005. The
number of published papers and citations of their work by other researchers is used to
calculate impact and productivity. In addition, the breadth of publication and citation
reflects the researcher’s reputation and credibility in the field and the scientific
community at large. The h-index is one of the most important ranking systems,
particularly because of its ability to take into account co-authors and co-researchers in a
network. h-index is calculated through:
�N𝑐𝑐 T
ℎ= 𝑎𝑎
(1)
where NcT denotes the scientist’s citation sum for his/her papers and is a proportionality
constant, with “a” values between 3 and 5, but its value is usually set as 4, which gives
non-fraction value in the denominator.
Definition 1. Given the set of teams T = {T1, T2, ......,Tn}, the measure ct-index ranks the
team Tj ∈ T on the basis of the sum of winning margins of runs, Tr, and the sum of
winning margin of wickets, Tw, from other teams i.e., T-{Tj}.
ct-index is an adoption of h-index [Hirsch (2005)]. The ct-index uses the same method of
thinking that the h-index does. Replace papers with the total winning margin in terms of
runs and wickets of the winning team, and replace the “team” with the authorteam with
1120 CMC, vol.65, no.2, pp.1113-1136, 2020

the highest margin of win should score higher, and in fact, that is what happens when
using the h-index.
ct-index is adopted as follows,

𝑐𝑐𝑐𝑐 − 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �𝑇𝑇𝑤𝑤𝑎𝑎+𝑇𝑇𝑟𝑟 (2)

Tw is the sum of winning margin by wickets, and Tr is the sum of winning margin by runs.
The value of “a” can be chosen between 1 and 5. We have used a=4 for experimentation.
This is chosen to avoid the fractional value in the denominator of Eq. (2). The value of a
wicket is assigned by accumulating the batting records of each team for three years (in
this case, 2013-2015). By calculating the average of runs scored against lost wickets, it is
determined that the value of a single wicket is 30.02 runs for the ODI matches, 32.04
runs for the Test matches, and 21.45 runs for the T20i matches. When calculating the ct-
index, it is necessary to substitute a consistent value of wickets.

3.2 Measuring team ranks through Cricket Team Rank (CTR)

Page et al. [Page, Brin, Motwani et al. (1999)] brought forth a ranking algorithm known
as PageRank that is used to rank web pages. PageRank [Page, Brin, Motwani et al. (1999)]
is said to be one of the most significant graph-based page rank algorithms. The idea
behind web page ranking is deemed as simple; it conducts the linkage of a web page with
various web pages that cater to the same subject. Here it must be noted that the in-links
provided in a page are more significant than the web page itself. The rank of any page
(node) can be calculated by using the following formula,
(1−d) PR(Ti ) PR(Tn )
PR(A) = N
+ d� CTi
+ ⋯+ CTn
� (3)
where PR(A) is PageRank of A, PR(Ti) are PageRanks of pages which are providing the
links to page A. CTi is the number of out-links given by a page Ti to other web pages in the
network, N is the total number of pages, and d is the damping factor having value 0.85.
Definition 2. Given the set of teams T = {T1, T2, ......,Tn}, CTR measurement ranks the
team A ∈ T based on statistics of results (win/loss), R(Ui) and R(Oi) ∀i≤1≤n, and
dynamic damping factor di for every opponent team Ti.
CTR is the adoption of the PageRank algorithm. If a team has won matches from stronger
teams (provided those teams were also victorious against stronger teams), the CTR of the
winning team should be high.
CTR of a team “A” is calculated as:
R(Ti ) R(Un )
CTR(A) = d �di + ⋯ + dn � (4)
R(Oi ) R(On)

where,
𝐺𝐺𝐺𝐺(𝑈𝑈𝑖𝑖 ) 𝐺𝐺𝐺𝐺(𝑂𝑂𝑖𝑖 )
𝑅𝑅(𝑈𝑈𝑖𝑖 ) = 𝑅𝑅(𝑂𝑂𝑖𝑖 ) =
𝑇𝑇𝑇𝑇(𝑈𝑈𝑖𝑖 ) 𝑇𝑇𝑇𝑇(𝑂𝑂𝑖𝑖 )
CTR(A) is the Cricket Team Rank of team A, GL(Ui) is the number of games lost against A
by team i and TG(Ui) are the total games played between A and team i, consequently, R(Ui) is
the ratio between GL(Ui) and TG(Ui). GL(Oi) is the games lost against other opponents
Context Based Adoption of Ranking and Indexing Measures 1121

(excluding A) by team i and TG(Oi) is the games played between team i and other opponents
(excluding A), so R(Oi) is the ratio between GL(Oi) and TG(Oi). Where di is the damping
factor, the value of di depends on the number of matches played by the opponent team. If the
number of matches played by the opponent is greater than or equal to mean matches, the
value of di is 1. If the number of matches played by the opponent is less than the mean
matches, the value of di is the ratio of the number of matches played by the opponent and the
number of mean matches. This is to handle the situations in which winning from a new team
who has not played enough matches, resultantly haven’t lost many matches should not be
given a high weightage. The benefit is reduced as di would be in a fraction for new teams.

3.3 Measuring team ranks through Weighted Cricket Team Rank (WCTR)
Weighted Cricket Team Rank (WCTR) uses winning margins of a team in terms of
wickets and runs while calculating its rank. The WCTR is defined as:
Definition 3. Given the set of teams T={T1, T2, ......,Tn}, WCTR measurement ranks the
team A∈T based on statistics of results (win/loss), R(Ui) and R(Oi) and statistics of the
margin of win/loss, M(Ui) and M(Oi) ∀i≤1≤n, and dynamic damping factor di for every
opponent team Ti.
WCTR is an improvised form of CTR that relies upon weightage. The weights are added
by taking into account the margin of matches lost by runs/wickets by the opponent teams.
The proposed WCTR asserts that an opponent team Ti should have a higher impact on the
ranking of the team A if it loses to team A by a big margin of runs/wickets but loses to
other teams at low margin and lower impact if it loses to team A by a small margin of
wickets/runs but loses to other teams by a big margin of wickets/runs. The WCTR score
of a team A is calculated as:
R(Ui ) M(Ui ) R(Un ) M(Un )
WCTR(A) = �di � ∗ �+ ⋯ + dn � ∗ �� (5)
R(Oi ) M(Oi ) R(On ) M(On )

where,
𝐺𝐺𝐺𝐺(𝑈𝑈𝑖𝑖 ) 𝐺𝐺𝐺𝐺(𝑂𝑂𝑖𝑖 )
𝑅𝑅(𝑈𝑈𝑖𝑖 ) = 𝑅𝑅(𝑂𝑂𝑖𝑖 ) =
𝑇𝑇𝑇𝑇(𝑈𝑈𝑖𝑖 ) 𝑇𝑇𝑇𝑇(𝑂𝑂𝑖𝑖 )
𝑀𝑀𝑀𝑀𝑀𝑀(𝑈𝑈𝑖𝑖 ) 𝑀𝑀𝑀𝑀𝑀𝑀(𝑂𝑂𝑖𝑖 )
𝑀𝑀(𝑈𝑈𝑖𝑖 ) = 𝑀𝑀(𝑂𝑂𝑖𝑖 ) =
𝑇𝑇𝑇𝑇𝑇𝑇(𝑈𝑈𝑖𝑖 ) 𝑀𝑀𝑀𝑀𝑀𝑀(𝑂𝑂𝑖𝑖 )
TG(Ui) is the number of total games played between team A and team i while GL(Ui)
represents the number of games lost by team i to team A. Consequently, R(Ui) is the ratio
between GL(Ui) and TG(Ui). GL(Oi) is the number of games lost against other opponents
(excluding A) by team i and TG(Oi) is the games played between team i and other
opponents (excluding A). Therefore, R(Oi) is the ratio between GL(Oi) and TG(Oi).
MGL(Ui) is the losing margin in games lost by team i against A and MTG(Ui)is the sum
of margin in total games played between A and ith team. Consequently, M(Ui) is the ratio
between MGL(Ui) and MTG(Ui). MGL(Oi) is the losing margin in games lost against
other opponents (excluding A) by team i and MTG(Oi) is the sum of margins in games
played between team i and other opponents (excluding A). Therefore, R(Oi) is the ratio
between GL(Oi) and TG(Oi). Where di is the damping factor; the value of di depends on
the number of matches played by the opponent team. If the number of matches played by
1122 CMC, vol.65, no.2, pp.1113-1136, 2020

the opponent is greater than or equal to mean matches, the value of di is 1. If the number
of matches played by the opponent is less than the mean matches, the value of di is the
ratio of the number of matches played by the opponent and the number of mean matches.

4 Experiments
The research in this paper used a dataset that is specified in this section. This section also
illustrates the results of each of the techniques investigated in this research and discusses
the results in the context of international cricket (ODI, Test, and T20i matches). This
section also discusses the use of a damping factor for each of the proposed techniques. A
comparative analysis of the techniques is presented, as well.

4.1 Dataset
The experiments are conducted using the CricInfo website’s dataset. This data
corresponds to the data used in the latest rankings provided by the ICC of ODI, Test and
T20i matches (as of July 20, 2016). The batting statistics are captured from January 2013-
December 2015 from each international match, and these statistics were used to
determine the weighted average of a single wicket.

4.2 Results and discussions

The results achieved through the proposed techniques, i.e., ct-index, CTR, and WCTR,
are presented in the following subsection. The experimental results are shown for all three
formats of international cricket matches.

4.2.1 ODI matches

In this sub-section, the comparative results are presented for all three proposed techniques
over the ODI dataset. The details of the data are explained in Section 4.1.
Table 4: ODI team rankings through proposed ranking measurements
ct-index CTR WCTR
Rank Team Score Team Score Team Score
1 India 46.23 New Zealand 16.91 Australia 16.57
2 South Africa 42.05 Australia 16.03 New Zealand 16.35
3 New Zealand 41.19 South Africa 14.15 India 14.75
4 Australia 41.09 India 13.80 South Africa 14.45
5 Sri Lanka 38.23 Sri Lanka 11.79 Sri Lanka 12.25
6 England 37.25 West Indies 10.30 West Indies 9.91
7 Bangladesh 33.49 Bangladesh 10.00 Bangladesh 9.75
8 Pakistan 31.94 England 9.46 England 9.49
9 West Indies 24.68 Pakistan 9.26 Pakistan 8.41
10 Zimbabwe 20.29 Zimbabwe 5.18 Zimbabwe 5.10

Tab. 4 shows the different rankings using the different rank measuring techniques, for
ODI matches. Conceptually the team rankings may be visualized into two halves. In the
Context Based Adoption of Ranking and Indexing Measures 1123

top five of the rankings, from all the three proposed techniques, are Australia, India, New
Zealand, South Africa, and Sri Lanka. There are, however, differences in the results
achieved from each formula. The ct-index results in India ranking as the number one
team. India’s wins are by high-margin in terms of runs and wickets. CTR ranks New
Zealand higher since New Zealand won the most matches against other high-ranking
teams compared to other member teams. The WCTR ranking ranks Australia as the top-
ranking team. The WCTR took into account Australia’s multiple wins against highly
ranked teams, and it also won many matches with high margins in terms of runs and
wickets. The latter half of Tab. 4 illustrates the lower-achieving teams. No matter which
method is used, the same teams result at the bottom of the list: Bangladesh, England,
Pakistan, West Indies, and Zimbabwe. No matter which method is used, Zimbabwe
maintains its rank as the tenth i.e., bottom team. This is due to the fact that Zimbabwe
won matches against lower-ranked competitors, won those by low margins of runs and
wickets, and did not win many matches against stronger ranked teams.

4.2.2 Test matches

The following are the ranking results of our proposed methods for test cricket matches
based on the data set explained in the previous section.
Table 5: Test team rankings through proposed ranking measurements
ct-index CTR WCTR
Rank Team Score Team Score Team Score
1 England 49.85 South Africa 8.40 South Africa 8.59
2 Australia 49.40 Pakistan 7.79 Pakistan 7.31
3 South Africa 46.29 England 6.84 England 6.62
4 India 38.53 India 6.53 India 6.15
5 Pakistan 36.18 Australia 6.26 Australia 5.70
6 Sri Lanka 36.09 New Zealand 4.64 Sri Lanka 4.26
7 New Zealand 34.80 Sri Lanka 4.49 New Zealand 4.17
8 West Indies 28.27 West Indies 2.27 West Indies 2.36
9 Bangladesh 15.98 Bangladesh 0.10 Bangladesh 0.08
10 Zimbabwe 00.00 Zimbabwe 0.00 Zimbabwe 0.00

The rank of the test cricketing nations based on the “test cricket data” is shown in Tab. 5.
The ranking shows the ranking and scores that are gained through the proposed
measurements. The results show that the ranking is easily read divided into two halves
using the Test cricket data. The top-ranking half and the bottom ranking half of the teams
each have five teams. The highest-ranking are Australia, England, India, Pakistan and
South Africa. When the ct-index, CTR and WCTR methods are examined on the same
data set, the highest-ranking nations are England, South Africa and Pakistan. Using ct-
index, the English team scores highest at 49.85. This high score is a result of England’s
wins with high margins. The CTR and WCTR methods resulted in South Africa being
issued the highest performance ranking. South Africa won a high number of matches,
1124 CMC, vol.65, no.2, pp.1113-1136, 2020

with a very strong ratio against other strong teams. The lower half of the ranking list
resulted in the same team names no matter what evaluation technique was used. The
teams in the second half of the results were: Bangladesh, New Zealand, Sri Lanka, West
Indies and Zimbabwe. For each of the three different methods tested, Zimbabwe was the
lowest ranking team. This is because Zimbabwe did not emerge the victor in any test
cricket match during the test data time frame.

4.2.3 T20 international (T20i) matches

The following are the ranking results of our proposed methods for T20i matches based on
the data set explained in the previous section.
Table 6: T20 team rankings through proposed ranking measurements
ct-index CTR WCTR
Rank Team Score Team Score Team Score
1 India 26.35 England 12.20 England 13.21
2 Afghanistan 23.22 New Zealand 11.01 West Indies 11.72
3 South Africa 23.14 South Africa 10.96 South Africa 11.49
4 West Indies 22.23 West Indies 10.76 New Zealand 10.00
5 New Zealand 21.47 Sri Lanka 9.24 India 7.89
6 Pakistan 19.22 India 7.29 Sri Lanka 6.68
7 Sri Lanka 18.47 Afghanistan 6.00 Bangladesh 6.46
8 Bangladesh 17.03 Bangladesh 5.34 Afghanistan 6.44
9 England 16.37 Pakistan 4.85 Pakistan 5.58
10 Australia 16.19 Australia 3.99 Australia 4.38

Tab. 6 is showing the cricket team ranking results of the proposed measurements when
applied over the T20i matches dataset. Unlike the ODI and test matches, we cannot divide
the achieved ranking into two of the groups. Using the ct-index, the Indian team is ranked
number one. Team India won many matches with extremely high margins. For CTR and
WCTR evaluations, the English team stayed on the top of the rankings. England won many
matches against strong teams, and their WCTR margins were stronger against strong teams.
Australia landed at the bottom of the list for all of the three different proposed methods.
The reason that Australia fell to the bottom of the list was that their number of wins was
lower, and the matches which they did win were of lower value. The value assessment is
based on the low ranking scores of the teams that they opposed and won over, and the
lower margins in the area of runs and wickets which those wins employed.

4.3 Comparative analysis of the proposed and existing rank measuring techniques
The goal of this paper is to determine the differences in the ICC ranking, which currently
dominates the international ranking platforms, and the ranking options outlined by Daud
et al. [Daud, Muhammad, Dawood et al. (2015)] to determine the most effective
technique. The paper evaluated the proposed techniques (ct-index, CTR, WCTR),
Context Based Adoption of Ranking and Indexing Measures 1125

techniques proposed by Daud et al. [Daud, Muhammad, Dawood et al. (2015)] and ICC
Team Rankings. We normalized the scores (0-1), achieved by all the techniques.

4.3.1 Comparison with ICC cricket ranking

The ICC implements an impromptu system that is completely dependent upon winning and
losing. ICC cricket ranking system is used for ranking of cricket teams competing in all three
international cricket platforms i.e., ODI, Test and T20i matches. In this section, this paper
will compare the results of proposed techniques with current ICC cricket team rankings.
ODI Matches. Fig. 1 illustrate that different techniques have a different impact on the
ranking of teams. For example, Australia dropped from 1st in ICC ranking to 4th in ct-index.
This illustrates that the margin of victory for Australia was considerably less than other
teams like India who topped the ranking in the ct-index method. Also, Australia remains
top in CTR but narrowly misses out in WCTR due to the impact of the winning margin.
Test Matches. As illustrated in Fig. 2, techniques have a different impact on the rating of
the teams. For example, South Africa jumped from 6th in ICC ranking to 3rd in ct-index.
This illustrates that the margin of victory for South Africa was considerably higher than
the other teams like India and Pakistan, who dropped the ranking in the ct-index method.
Also, South Africa tops in CTR and WCTR, illustrating the benefit of not losing less
against strong teams.

Figure 1: Comparison of ct-index, CTR, WCTR and ICC ODI cricket rankings
1126 CMC, vol.65, no.2, pp.1113-1136, 2020

Figure 2: Comparison of ct-index, CTR, WCTR and ICC Test rankings

T20 International Matches. From Fig. 3 it may be observed that New Zealand dropped
from 1st in ICC ranking to 5th in ct-index. This illustrates that the margin of victory for
New Zealand was considerably less than other teams like India who topped the ranking in
ct-index method. Also, New Zealand narrowly misses a top spot in CTR and drops to 4th
in WCTR, illustrating the impact of the margin of victory in WCTR.

.
Figure 3: Comparison of ct-index, CTR, WCTR and ICC T20i rankings

4.3.2 Comparative analysis of proposed Ranking Measurements with Daud et al. [Daud,
Muhammad, Dawood et al. (2015)]
Daud et al. [Daud, Muhammad, Dawood et al. (2015)] proposed four different ranking
measurements. Here a comparison of the proposed methods with three of the relevant ones is
presented, and for a fair comparison, experiments are performed on the same dataset as done
by Daud et al. [Daud, Muhammad, Dawood et al. (2015)]. The data was collected for the
duration of (2010-Mid 2012). The results are presented for the ODI matches only.
Context Based Adoption of Ranking and Indexing Measures 1127

Comparing ct-index with t-index

The first method to compare is the Cricket Team-Index (ct-index) with Team-index (t-
index) [Daud, Muhammad, Dawood et al. (2015)]. The t-index is weak in ranking that it
uses the same weighting mechanism to wickets and runs. It is common knowledge that
wickets have more value than runs in the outcome of a game. As outlined in the deeper
discussion of the data, the valuation of each wicket is arrived at by calculating the
average of the batting information from the past three years. Comparative rankings are
presented in Tab. 7:
Table 7: Comparative analysis of proposed(ct-index) with Daud et al. [Daud,
Muhammad, Dawood et al. (2015)] (t-index) for ODI matches
ct-index t-index
Rank Team Score Team Score
1 India 45.7137 Australia 3.81
2 Sri Lanka 42.4863 South Africa 3.54
3 Pakistan 39.2493 Sri Lanka 3.54
4 Australia 38.6496 Pakistan 3.46
5 South Africa 38.0653 India 3.35
6 England 32.7768 New Zealand 3.08
7 New Zealand 31.4107 England 2.96
8 Ireland 30.8754 West Indies 2.78
9 Bangladesh 27.8593 Bangladesh 2.18
10 West Indies 27.8438 Zimbabwe 2.12

The results, Tab. 7, clearly show an increase in the index values due to the impact of
using the weightage of wicket instead of using wicket and run with the same weightage.
For example, India jumped from 5th in t-index to 1st in ct-index due to high margin
victories by runs/wickets.

Comparing Cricket Team Rank (CTR) with TeamRank (TR)

When comparing the Cricket Team Rank (CTR) with the Team Rank (TR), there is a
proposed adoption of the PageRank algorithm in the CTR method. This allows the
additional consideration of the ratios of lost matches between the teams playing, and lost
matches against the other teams. The assumption was that the team which lost less would
be the strongest. Daud et al. [Daud, Muhammad, Dawood et al. (2015)] took advantage of
a static damping factor. This brought about a result that was inaccurate and lower for
“normal” teams but gave teams with less of a playing history too much credit. For
example, if a team plays 2 matches and wins 1, its rank will be 0.5 according to this
calculation, the team that played 50 matches and won 25 has the same rank. To
compensate for this issue, we used a dynamic damping factor (di). Tab. 8 shows the
comparative team rankings achieved by both the techniques:
1128 CMC, vol.65, no.2, pp.1113-1136, 2020

Table 8: Comparative analysis of proposed (CTR) with Daud et al. [Daud, Muhammad,
Dawood et al. (2015)] (TR)
CTR TR
Rank Team Score Team Score
1 Pakistan 10.8223 Australia 0.133
2 South Africa 10.8122 India 0.121
3 India 10.1737 Pakistan 0.119
4 Australia 9.8100 Sri Lanka 0.114
5 Sri Lanka 8.8970 South Africa 0.110
6 England 8.7769 New Zealand 0.098
7 Bangladesh 6.6100 England 0.094
8 Zimbabwe 5.4226 West Indies 0.081
9 West Indies 5.1413 Zimbabwe 0.063
10 Ireland 3.5656 Bangladesh 0.061

Comparing Weighted Cricket Team Rank (WCTR) with Weighted Team Rank (WTR)
The final comparison is between the Weighted Cricket Team Rank (WCTR) and the
Weighted Team Rank (WTR). The WCTR is a later evolution, or extended formula,
based upon CTR. The important characteristic of WCTR is that it allows teams which
win by the biggest margin, and lose by the smallest margin, to receive the highest
rankings. Daud et al. [Daud, Muhammad, Dawood et al. (2015)] supposed that WTR
would have the same impact, but in fact, WTR has the same issue as the t-index method.
Namely, there cannot be the same ranking issued to runs and wickets. The following
comparison shows the differences achieved when employing the two methods:
Table 9: Comparative analysis of proposed (WCTR) with Daud et al. [Daud, Muhammad,
Dawood et al. (2015)] (WTR)
WCTR WTR
Rank Team Score Team Score
1 South Africa 11.9653 South Africa 0.051270
2 Pakistan 10.1000 Sri Lanka 0.051269
3 India 9.5706 Australia 0.051260
4 Australia 8.5990 India 0.051255
5 Sri Lanka 8.2380 New Zealand 0.051247
6 England 8.0268 England 0.051242
7 Bangladesh 6.3928 Pakistan 0.051204
8 Zimbabwe 5.3062 West Indies 0.051192
9 West Indies 5.1967 Bangladesh 0.050954
10 Ireland 4.4877 Zimbabwe 0.050824
Context Based Adoption of Ranking and Indexing Measures 1129

The results presented in Tab. 9 clearly show the increase in ranking due to the impact of
using the weightage of wicket instead of using wicket and run with the same weightage.
For example, Sri Lanka is dropped from 2nd in WTR to 5th in WCTR due to low margin
victories by runs/wickets.

5 Discussion
In this section, a detailed discussion of the proposed techniques is presented. The
proposed techniques are elaborated by choosing and evaluating example data.

5.1 Presenting relation between winning margins

While presenting ct-index, a transformation between runs and wickets in proposed. Daud
et al. [Daud, Muhammad, Dawood et al. (2015)] used the same weightage for both of the
runs and wickets in terms of winning margins. This is totally against the spirit and
statistics of the cricket game. It is clear noting that the winning margin for a team batting
second can be in terms of wickets while winning margin for the team batting first can be
in terms of runs. There are eleven player sides in cricket games, opponents have to get out
ten players to bowl out the whole team. The maximum number of wickets for a team is
ten and this can be a maximum winning margin while team bats second and chases the
runs successfully without losing its single-player out. On the other hand, a team that bats
first and restricts its opponent to lower than its runs scored is considered a winner in
terms of runs. This winning margin can vary from 1 to hundreds of runs. The weight of
the quantities whose ranges are quite different should not be the same. By considering an
example, a discussion is presented.
In Tab. 10, statistics for two teams A and B are presented. Both the teams A and B won
five matches each while batting first and second. The total winning margins for team A in
terms of runs and wickets are 220 and 37, respectively, while the same for team B are 250
and 26, respectively. If the weights of both winning runs and wickets are kept same then
the ct-index score for both teams would be:

𝑇𝑇𝑟𝑟 + 𝑇𝑇𝑤𝑤 220 + 37

𝑐𝑐𝑐𝑐 − 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐴𝐴) = � =� = 8.01
𝑎𝑎 4

𝑇𝑇𝑟𝑟 + 𝑇𝑇𝑤𝑤 250 + 26

𝑐𝑐𝑐𝑐 − 𝑖𝑖𝑖𝑖𝑖𝑖𝑒𝑒𝑥𝑥(𝐵𝐵) = � =� = 8.30
𝑎𝑎 4
 1130 CMC, vol.65, no.2, pp.1113-1136, 2020

Table 10: Example data for presenting the relation between winning margins
Team Team Winning while batting First Team Winning while batting Second
Runs Wickets Runs Wickets Winning Total Runs Wickets Runs Wickets Winning Total
Scored Lost Scored Lost Margins Winning Scored Lost Scored Lost Margins Winning
by by (Runs) Margins by by (Wickets) Margins
Team Opponent (Runs) Opponent Team (Wickets)
Tr Tw
A 345 5 300 7 45 220 200 10 201 0 10 37
330 7 310 10 20 270 7 271 5 5
350 9 320 10 30 230 8 231 4 6
300 7 200 10 100 300 9 301 3 7
270 6 245 9 25 170 10 171 1 9
B 260 3 185 10 75 250 230 8 231 5 5 26
300 8 220 7 80 200 10 201 8 2
310 7 255 6 55 330 5 331 6 4
300 10 280 7 20 250 6 251 1 9
280 5 250 3 30 270 6 271 4 6

Team B would be ranked higher than team B. It is not the correct ranking as team A has a
higher margin of winning in terms of wickets than team B. The difference between
wicket margin is 11, which must be weighted quite higher than the winning runs margin
which is just 25. To find out the relation between runs and wickets, we calculated total
runs scored by both the teams, whether batting first or second. In the same manner, total
wickets lost by both the teams for all the matches are calculated.
TR=Total Runs scored by both the teams=10520
TW=Total Wickets lost by both the teams=288
Runs Per Wicket (RPW)= (TR)/ (TW)=10520/288=36.53
The ranking through the proposed technique, ct-index, incorporating relation between
runs and wickets is calculated as:
𝑇𝑇𝑤𝑤 + 𝑇𝑇𝑟𝑟 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅 220 + 37 ∗ 36.53 220 + 1351.61
𝑐𝑐𝑐𝑐 − 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐴𝐴) = � =� =� = 19.82
𝑎𝑎 4 4

𝑇𝑇𝑤𝑤 + 𝑇𝑇𝑟𝑟 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅 250 + 26 ∗ 36.53 220 + 949.78

𝑐𝑐𝑐𝑐 − 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝐵𝐵) = � =� =� = 17.32
𝑎𝑎 4 4
Team A is ranked higher than B which is more realistic as team A won their matches by
quite higher margins of wickets than team B.

5.2 Presenting and incorporating strength of opposition teams

A team is supposed to be a strong one if it wins more matches than it loses. When a team
wins against strong opposition, it should be awarded more points than one’s winning a
weaker team. Daud et al. [Daud, Muhammad, Dawood et al. (2015)] in their technique,
TR, managed to adopt the concept of opposition strength while awarding the points to
Context Based Adoption of Ranking and Indexing Measures 1131

winners. It may be observed from Eq. (3) the contribution from team ti while calculating
𝑃𝑃𝑃𝑃(𝑇𝑇𝑖𝑖 )
the TR score of team A, that it depends upon the following ratio i.e., . 𝑃𝑃𝑃𝑃(𝑇𝑇𝑖𝑖 )
𝐶𝐶𝑇𝑇𝑖𝑖
represents matches lost by ith team against A while CTi is the total number of matches lost
by Ti. The existing technique does not count for the number of matches team Ti played
rather only the result count in the form of losing is taken. There are many situations
where the results are not the correct ones. To illustrate the inherited problems in ranking
while not incorporating the statistics of total played matches an example data in Tab. 11
is presented below:
Table 11: Example data for determining the limitations of Daud et al. [Daud,
Muhammad, Dawood et al. (2015)]
Team Total games Total games Total games Total games Contribution
lost against played against lost by Team played by made by team
Team A Team A (CTi) Team in Ranking A
(PR(Ti))

B 3 5 6 10 3/6=0.33
C 3 5 6 30 3/6=0.33

As shown in (3), while calculating the rank through TR(A) [Daud, Muhammad, Dawood
et al. (2015)], all the other teams make their contribution to it. The contribution made by
B would be PR(B)/CTB; the values from Tab. 11 would return it as 0.33.
Same as the contribution made by team C towards the ranking of team A would be 0.33.
It doesn’t look great as team C looks far stronger than team B. Although both the teams
have the same statistics for games lost against A and total games lost against all teams.
Team B lost quite frequently, for instance, lost 3 matches from A while only 5 matches
were played in between them. The overall behavior of team B is not different, it lost 6
matches in all while its total matches are 10. On the other hand, team C has quite better
statistics, i.e., it lost 3 matches to team A while won 7 matches. Overall performance of
team C is the same as against A, i.e., in all teams, C played 30 matches and won 24
matches and lost only 6 matches. The contribution made by both the teams is the same as
TR [Daud, Muhammad, Dawood et al. (2015)] does not use the overall statistics of the
teams in terms of the total number of matches played against team A.
Table 12: Example data for illustrating opposition team strength
Team Total games Total games Games lost Games lost R(Ui ) R(Oi ) Contribution made
lost against played against against GL(Ui ) GL(Oi ) by team in Ranking
Team A against Teams other Teams other = TG(Ui ) =
TG(Oi ) A
(GL(Ui)) Team A than A than A R(Ui )
(TG(Ui)) (GL(Oi)) (TG(Oi)) R(Oi )
B 3 5 3 5 0.6 0.6 1
C 3 5 3 25 0.6 0.12 5

The proposed CTR and WCTR solved the above-discussed issues by incorporating the
number of matches played by ith team against team A along with the total number of
1132 CMC, vol.65, no.2, pp.1113-1136, 2020

matches played by ith team against all the teams other than A. In Eq. (4) and Eq. (5) the
contribution made by team B in the ranking of team A is calculated by
𝑅𝑅(𝑈𝑈𝑖𝑖 ) 𝐺𝐺𝐺𝐺(𝐴𝐴) 𝐺𝐺𝐺𝐺(𝑂𝑂𝑖𝑖 )
, where 𝑅𝑅(𝐴𝐴) = and 𝑅𝑅(𝑂𝑂𝑖𝑖 ) = .
𝑅𝑅(𝑂𝑂𝑖𝑖 ) 𝑇𝑇𝑇𝑇(𝐴𝐴) 𝑇𝑇𝑇𝑇(𝑂𝑂𝑖𝑖 )
While ranking A the contribution by team B is 1 while it is 5 by team C (Tab. 12). The
calculated contributions are logical and true as team C is a stronger opposition and winning
from a stronger opposition must reward higher than winning from a weaker opposition.

5.3 Dynamic damping factor

Static damping factor was used by Daud et al. [Daud, Muhammad, Dawood et al. (2015)]
but using static damping factor is not the optimum solution for the techniques using ratios
as the impact is the same for all the teams. The TR approach was only resulting in false
points for teams that win more, with teams that winless being “dampened”. The use of a
static damping factor is just a scaling factor and seems useless. The proposed techniques
CTR and WCTR use dynamic damping factors that help determine the regular and non-
regular teams. The overall use of dynamic damping factor assigns different weights to the
emerging and regular teams and thus proper strength of the teams is calculated. Given
below is an example data in tabular form, which would help to illustrate the importance
of dynamic damping factor.
Tab. 13 is showing the statistical records for six teams. The statistics for team A are not
shown as in this example. Its rank is being calculated. Suppose team A has played 65
matches in all. As discussed earlier, while calculating the rank of team A through
proposed CTR, WCTR, and Daud et al. [Daud, Muhammad, Dawood et al. (2015)] all the
teams contribute in it. The contributions made by B and for ranking team A through TR
[Daud, Muhammad, Dawood et al. (2015)] is 0.25. Even though the proposed CTR and
WCTR, the contributions made by team B and C are equal i.e., 1. This is not logical as B
is just an emerging team, while team C is an experienced and regular team. Winning from
B must not be the same as winning from team C but the tricky statistics with static
damping factor would make the proposed technique behave the same as it was through
TR. To solve such issues the concept of dynamic factor is introduced in this paper.
Table 13: Example data to understand the need for dynamic damping factor
Team Total Total Games Games Total R(Ui ) R(Oi ) Contribution Contribution
games games lost Played Matches GL(Ui ) GL(Oi ) made by made by
lost played against against Played = TG(Ui ) = TG(Oi ) team in team in
against against Teams Teams Ranking A Ranking A[]
Team A Team A other other R(Ui ) PR(Ti)/CTi
R(Oi)
(GL(Ui)) (TG(Ui)) than A than A
(GL(Oi)) (TG(Oi))
B 1 2 3 6 8 0.5 0.5 1 0.25
C 10 20 30 60 80 0.5 0.5 1 0.25
D 7 19 21 41 60 0.37 0.51 0.72 0.25
E 10 22 26 48 70 0.45 0.54 0.83 0.28
F 12 21 24 49 80 0.57 0.49 1.16 0.33
G 8 17 17 38 55 0.47 0.45 1.04 0.32
Context Based Adoption of Ranking and Indexing Measures 1133

To calculate the damping factor, di, for ith team following steps are performed.
i. Find the mean of the matches (mm) played by all the teams.
ii. (a) If the number of matches played by the ith team is greater than or equal to mm then
di =1, otherwise
(b) di = Number of Matches Played by Opposition
The total number of matches played by all the seven teams is 418, and the mean number
of matches (mm) is 59.71. The damping factor for team C, which played 80 matches is
greater than mm, is 1 while the damping factor for team B, which played 8 matches
would be calculated through the rule (i), i.e., dB=8/59.71=0.13. The contributions made
by team B and C would in ranking team A would be calculated as:
𝑅𝑅(𝐴𝐴 )
Contribution through team B=𝑑𝑑𝐵𝐵 ∗ 𝑅𝑅(𝑂𝑂𝐵𝐵 ) = 0.13 ∗ 1 = 0.13
𝐵𝐵
𝑅𝑅(𝐴𝐴𝐶𝐶 )
𝑑𝑑𝐶𝐶 ∗ =1∗1=1
𝑅𝑅(𝑂𝑂𝐶𝐶 )
𝑅𝑅(𝐴𝐴 )
Contribution through team C=𝑑𝑑𝐶𝐶 ∗ 𝑅𝑅(𝑂𝑂𝐶𝐶 ) = 1 ∗ 1 = 1
𝐶𝐶

The above two contributions for team B and C reflect that winning from regular team weighs
more than an emerging one and the strength of the teams are accurately determined.

5.4 Incorporating winning margins in team ranking

Daud et al. [Daud, Muhammad, Dawood et al. (2015)] proposed WTR, which is the
combination of TR and t-index. WTR inherits the problems of both techniques (as
discussed above) and has factual, Nomenclature and conceptual problems, discussed in
detail by Saqlain et al. [Saqlain and Usmani (2017)]. The winning margin is an important
factor that should be incorporated while ranking the teams. Suppose a team B lost its
match against team A by a margin of 200 runs, and team C lost its match against A by a
margin of 1 run. Although both the matches end with the same result but winning by 200
runs must be given more weightage than winning by 1 run. The proposed WCTR ranks
the teams by incorporating not only the strength of opposition and dynamic damping
factor but the winning margins are used as well. Suppose the rank of team A is to be
𝑀𝑀(𝐴𝐴𝑖𝑖 )
calculated. In Eq. (5) where WCTR score of team A is calculated, )
represents the
𝑀𝑀(𝑂𝑂𝑖𝑖
winning margins. This represents the ratios of losing margins against team A to all other
teams. This not only looks for one match-winning margin but the overall history between
A and ith team along with the overall history of winning/losing margins of team A against
all other teams is encountered. The use of historical winning/losing margins make the
proposed WCTR a robust ranking technique.

6 Conclusion
Adoptions of PageRank and h-index are presented for cricket team ranking. In this regard,
three ranking measurements are proposed i.e., ct-index, CTR and WCTR. The investigation
focuses on the importance of a margin in a win and quality of opposition. If a win is won by
a larger margin of runs and wickets, the impact is significant and affects the team’s overall
ranking. The use of a dynamic damping factor produces a significant difference from the
1134 CMC, vol.65, no.2, pp.1113-1136, 2020

use of a static damping factor. The weighting factor used comes into play most prominently
when two teams happen to win a similar number of games against competitors who are
ranked approximately the same. Adopting the h-index and PageRank produces an accurate
ranking of international cricket teams. The result is positive because the opposing team is
weighted as a strong or weak opponent, and the margin of the win (in terms of runs and
wickets) is taken into consideration for both the winning and losing team.

Funding Statement: This research did not receive any specific grant from funding
agencies in the public, commercial, or not-for-profit sectors.

Conflicts of Interest: The authors declare that they have no conflicts of interest to report
regarding the present study.

References
Amin, G. R.; Sharma, S. K. (2014): Measuring batting parameters in cricket: a two-
stage regression-OWA method. Measurement: Journal of the International Measurement
Confederation, vol. 53, pp. 56-61.
Borooah, V. K.; Mangan, J. E. (2010): The ‘Bradman class’: an exploration of some
issues in the evaluation of batsmen for test matches, 1877-2006. Journal of Quantitative
Analysis in Sports, vol. 6, no. 3, pp. 1-21.
Bracewell, P. J.; Ruggiero, K. (2009): A parametric control chart for monitoring
individual batting performances in cricket. Journal of Quantitative Analysis in Sports, vol.
5, no. 3.
Burrell, Q. L. (2007): Hirsch’s h-index: a stochastic model. Journal of Informetrics, vol.
1, no. 1, pp. 16-25.
Cerchiello, P.; Giudici, P. (2014): On a statistical h index. Scientometrics, vol. 99, no. 2,
pp. 299-312.
Dadelo, S.; Turskis, Z.; Zavadskas, E. K.; Dadeliene, R. (2014): Multi-criteria
assessment and ranking system of sport team formation based on objective-measured
values of criteria set. Expert Systems with Applications, vol. 41, no. 14, pp. 6106-6113.
Daud, A.; Muhammad, F.; Dawood, H.; Dawood, H. (2015): Ranking cricket teams.
Information Processing and Management, vol. 51, no. 2, pp. 62-73.
Duch, J.; Waitzman, J. S.; Amaral, L. A. N. (2010): Quantifying the performance of
individual players in a team activity. PLoS One, vol. 5, no. 6, e10937.
Egghe, L. (2006). Theory and practise of the g-index. Scientometrics, vol. 69, no. 1, pp.
131-152.
Ezadi, S.; Allahviranloo, T. (2018): New multi-layer method for Z-number ranking
using hyperbolic tangent function and convex combination. Intelligent Automation and
Soft Computing, vol. 24, no. 1, pp. 217-223.
Farooq, M.; Khan, H. U.; Malik, T. A.; Shah, S. M. S. (2016): A novel approach to
rank authors in an academic network. International Journal of Computer Science and
Information Security, vol. 14, no. 7, pp. 617.
Context Based Adoption of Ranking and Indexing Measures 1135

Gao, S.; Wang, Z.; Chen, L. C. (2019): SI bitmap index and optimization for Membership
query. Intelligent Automation and Soft Computing, vol. 25, no. 4, pp. 683-689.
Haley, M. R. (2016): On the inauspicious incentives of the scholar-level h-index: an
economist’s take on collusive and coercive citation. Applied Economics Letters, vol. 24,
no. 2, pp. 85-89.
Haveliwala, T. H. (2002): Topic-sensitive PageRank. Proceedings of the 11th
International Conference On World Wide Web, pp. 517-526.
Hirsch, J. E. (2005): An index to quantify an individual’s scientific research output.
Proceedings of the National Academy of Sciences of the United States of America, vol.
102, no. 46, pp. 16569-16572.
Iván, G.; Grolmusz, V. (2011): When the web meets the cell: using personalized PageRank
for analyzing protein interaction networks. Bioinformatics, vol. 27, no. 3, 405-407.
Manaskasemsak, B.; Rungsawang, A.; Yamana, H. (2011): Time-weighted web
authoritative ranking. Information Retrieval, vol. 14, no. 2, pp. 133-157.
Mihalcea, R.; Tarau, P. (2004): TextRank: Bringing order into texts. Proceedings of the
International Conference On Empirical Methods in Natural Language Processing, pp.
404-411.
Min, B.; Kim, J.; Choe, C.; Eom, H.; McKay, R. I. (2008): A compound framework
for sport prediction: the case study of football. Knowledge-Based Systems, vol. 21, no. 7,
pp. 551-562.
Mukherjee, S. (2012): Identifying the greatest team and captain-a complex network
approach to cricket matches. Physica A: Statistical Mechanics and Its Applications, vol.
391, no. 23, pp. 6066-6076.
Mukherjee, S. (2014): Quantifying individual performance in cricket-a network analysis
of batsmen and bowlers. Physica A: Statistical Mechanics and Its Applications, vol. 393,
pp. 624-637.
Nykl, M.; Campr, M.; Ježek, K. (2015): Author ranking based on personalized
PageRank. Journal of Informetrics, vol. 9, no. 4, pp. 777-799.
Page, L.; Brin, S.; Motwani, R.; Winograd, T. (1999): The PageRank citation ranking:
bringing order to the web. Technical Report. Stanford InfoLab.
Pérez-Rosés, H.; Sebé, F.; Ribó, J. M. (2016): Endorsement deduction and ranking in
social networks. Computer Communications, vol. 73, pp. 200-210.
Pradhan, D.; Paul, P. S.; Maheswari, U.; Nandi, S.; Chakraborty, T. (2017): C3-
index: a PageRank based multi-faceted metric for authors’ performance measurement.
Scientometrics, vol. 110, no. 1, pp. 253-273.
Qader, M. A.; Zaidan, B. B.; Zaidan, A. A.; Ali, S. K.; Kamaluddin, M. A. et al.
(2017): A methodology for football players selection problem based on multi-
measurements criteria analysis. Measurement: Journal of the International Measurement
Confederation, vol. 111, pp. 38-50.
Saqlain, S. M.; Usmani, R. S. A. (2017): Comment on “ranking cricket teams.”
Information Processing and Management, vol. 53, no. 2, pp. 450-453.
 1136 CMC, vol.65, no.2, pp.1113-1136, 2020

Springer, E. P. (2016). Using the h index to assess impact of DOE national laboratories.
Wang, Q.; Ren, J. D.; Davis Darryl, N.; Cheng, Y. Q. (2018): An algorithm for fast
mining top-rank-k frequent patterns based on node-list data structure. Intelligent
Automation and Soft Computing, vol. 24, no. 2, pp. 399-404.
Xiang, L.; Wu, W.; Li, X.; Yang, C. (2018): A linguistic steganography based on word
indexing compression and candidate selection. Multimedia Tools and Applications, vol.
77, no. 2, pp. 1-21.
Yeh, J. Y. (2018): Rank-order-correlation-based feature vector context transformation for
learning to rank for information retrieval. Computer Systems Science and
Engineering, vol. 33, no. 1, pp. 41-52.
Zhang, B.; Wang, Y.; Jin, Q.; Ma, J. (2015): A PageRank-inspired heuristic scheme for
influence maximization in social networks. International Journal of Web Services
Research, vol. 12, no. 4, pp. 48-62.

View publication stats