Asset Management:

5. Investment Strategies

Felix Wilke
Nova School of Business and Economics
Spring 2024
Topics

• Backtesting.
• Components. Trading rules.
• Data mining and biases.
• Relationship between portfolio sorts and regressions.

• Equity strategies.
• Discretionary long-short equity.
• Dedicated short biased.
• Quant equity.

1/28
Backtesting investment strategies
Backtest

• Backtest: Simulation of a trading strategy.

• Would the strategy have worked in the past?
• Backtest can teach you about a strategy’s risk, and give improvement ideas.

• Cliff Asness: “If your mechanic used the word ’work’ to mean that your car might
work 6-7 years out of 10, then you would fire your mechanic, but this is how asset
management tends to ’work’.”
• Components of a backtest:
• Universe: The set of securities to be traded.
• Signals: Data used as input, the source, and the analysis method.
• Trading Rule: How signals are used to trade, including review frequency, position
rebalancing, and position sizes.
• Time Lags: Ensuring the trading rule is implementable.

2/28
Trading rules

• Portfolio rebalance rule.

• Starts from a macro view of the portfolio.
• Looks at the entire portfolio of securities and defines how it is rebalanced.
• This type of trading rule is backtested as follows. For each time period,
• Determine the optimal portfolio of securities.
• Make a (paper) trade to rebalance to this portfolio.

• Enter-exit trading rule.

• Build up, trade by trade, from micro level.
• Think in terms of discrete trades:
• For each asset, determine when to enter a new trade and how to size the initial position.
• Determine how the position is resized over time, depending on the circumstances.
• Determine when to exit the trade.

3/28
Data mining and biases

• Backtests typically look a lot better than the real-world trading performance:
• The world is changing.
• All backtests suffer from data mining biases.
• Avoidable biases:
• Biased universe of securities.
• Biased trading signals and rules.
• In-sample tests (you should do an out-of-sample test).
• Unavoidable Biases:
• You gravitate toward a version of the trade that has worked well in the past.
• You tried the backtest because you heard someone made money on this trade.
• Discounting backtests.
• Goal:
• A strategy that performs well in future trading.
• Not simply to have the best possible backtest.

4/28
Adjusting backtests for trading costs

• Transaction costs reduce the returns of a trading strategy.

• Taking transaction costs into account in backtest:

• Compute the return on the portfolio.
• Compute the new security positions and the implied trades.
• Compute the expected trading costs for every security and add them up.
• Subtract the total expected trading cost from the portfolio return.

• T-costs are more important:

• The higher the turnover of the trading strategy.
• The larger the fund.

• Transaction costs change the optimal trading rule.

5/28
Backtests and regressions

Metatheorem. Any regression can be expressed as a portfolio sort, and any portfolio sort
can be expressed as a regression:

• A time series regression corresponds to a market timing strategy.

• A cross-sectional regression corresponds to a security selection strategy.

• A univariate regression corresponds to sorting securities by one signal; a bivariate

regression corresponds to double-sorting securities by two signals, allowing you to
determine whether one signal adds value beyond the other; multivariate regression
corresponds to sorting by multiple signals.

6/28
Market timing and time series regressions

• Time series regression of the excess return re of one security, say, the overall stock
market, on a forecasting variable F, say, the dividend-to-price ratio:

ret+1 = a + bFt + ε t+1

• The estimate of the regression coefficient b is given by:

∑Tt=1 (Ft − F̄)ret+1 T

b̂ =
∑Tt=1 (Ft − F̄)2
= ∑ xt ret+1
t=1

• which is the cumulative return on a long/short timing strategy, where the trading
position x is given by
xt = k(Ft − F̄)
where
1
k=
∑Tt=1 (Ft − F̄)2

7/28
Security selection strategies and cross-sectional regressions (1/2)

• Cross-sectional regression with forecasting variable Fit for every security i:

rit+1 = a + bFit + εit+1

• We can run this regression across securities at any time t:

∑N i i
i=1 (Ft − F̄t )rt+1
N
b̂t =
∑N ( Fi − F̄ )2
= ∑ xit rit+1
i=1 t t i=1

• Long/short security selection strategy with position in security i is:

xit = kt (Fit − F̄t )

• with
1
kt = .
∑N i
i=1 (Ft − F̄t )2

8/28
Security selection strategies and cross-sectional regressions (2/2)

• The overall estimate of the regression coefficient b̂ using Fama and MacBeth (1973):
1 T
T t∑
b̂ = b̂t
=1
• This is the average profit of the strategy over time.
• The risk of the strategy: the volatility of the regression coefficients:
v
u
u 1 T
σ̂ = t
T − 1 t∑
(b̂t − b̂)2
=1

• Sharpe ratio of the security selection strategy is

b̂
SR =
σ̂
• Corresponds closely to the t-statistic of the regression estimate:
√ b̂
t-statistic = T
σ̂ 9/28
Several trading signals

• We can regress returns on several trading signals, say, F and G:

rit+1 = a + bF Fit + bG Git + εit+1

• What does bF mean here?

• Similar to double-sorting securities by both F and G each period.

• One advantage of regressions is that it is easy to add many variables on the

right-hand side, while it becomes impracticable to quadruple-sort securities.

• Which strategies, market timing or security selection are more susceptible to biases?

10/28
Equity strategies
1. Discretionary long-short equity

• Discretionary trading.
• Possibly screen the universe of securities using some criteria, e.g. B/M.
• Make a tailored (fundamental) analysis of the selected securities.
• Relies on both hard information (data) and soft information (discussion with executives,
customers, other traders).
• Make long/short investment in analyzed securities based on an analyst’s overall
assessment.
• Strategies applied.
• Goes long and short, often more long than short.
• Security selection: Value, growth; earnings quality; management quality; sector
specialists; catalysts; flows in the market.
• Industry rotations.
• Market timing.
• Activist.

11/28
Value Investing

• Warren Buffett: “Intrinsic value is an all-important concept that offers the only
logical approach to evaluating the relative attractiveness of investments and
businesses. Intrinsic value can be defined simply: It is the discounted value of the
cash that can be taken out of a business during its remaining life.”
• Trade:
• Buy low (stocks with high intrinsic value / market value)
• Sell high (stocks with low intrinsic value / market value)
• Example:
• Buy a company with more cash than the equity value and no debt – if you can find it.
• How to find the intrinsic value more generally:
• Valuation! Fundamental analysis.
• Talk to the firm and everyone involved in its “value chain”:
• Management, employees, unions,
• Customers,
• Suppliers,
• Competitors.
12/28
Prices and fundamentals: a value investor’s perspective

13/28
Prices and fundamentals: margin of safety

14/28
2. Dedicated short biased

• Goes more short than long, the opposite tilt vs. equity long short managers.
• Dedicated short bias is a tiny subset of hedge funds.
• But shorting more broadly is important.
• Looking at short sellers is a great way to learn about shorting.
• Similar in techniques and process to equity long short.
• Focus on short ideas: managers like James Chanos look for companies:
• With materially over-stated earnings.
• Aggressive accounting methods.
• Incomprehensible statements in SEC filings.
• Engaged in outright fraud.
• With flawed business plan.
• No sustainable way to make profits (e.g. many internet companies).
• Based on technology becoming obsolete.
• Booms gone bust, relying on excessive use of credit.
• Examples:
• Hotels where all the rooms are empty.
• Pharmaceuticals with drugs no doctors prescribe (or with new risks).
15/28
Short sale frictions mean that companies can be overvalued

• Some people may have positive information/opinions about a company.

• Other people have negative information/opinions.
• If it is costly/difficult to sell short.
• Negative information may not be reflected in the price.
• The price could be too high.
• Future returns would be low.
• A large speculative premium can arise when people forecast the forecast of others:
• People may anticipate that the future price will be driven up by
• a greater fool, plus
• short-sale frictions.
• Example:
• Some people think that recessions are shallow and booms moderate:
Value in recession 80, value in boom 120, 50/50 probability so current value 100.
• Other people think that recessions are painful but booms great:
Value in recession 60, value in boom 140, 50/50 probability so current value 100.
• What if price in recession/boom is set by the most optimistic investor?
Then what is the current price?
16/28
James Chanos and the Enron bet

17/28
3. Quant equity: three types of quant strategies

1. Fundamental quant.
• Turnover measured in weeks or months.
• Often trading on factors, styles.

2. Statistical arbitrage.
• Turnover measured in hours or days.
• Often trading on temporary mispricing between pairs of securities.

3. High frequency trading.

• Turnover measured in seconds.
• Often trading on temporary “true” arbitrages, temporary mispricings, and liquidity
provision.

18/28
Fundamental Quant

• Trade on several quantitative factors, such as

• Value,
• Momentum,
• Quality,
• Size,
• Many others.
• Use information similar to that used by fundamental long/short equity manager.
• Try to “teach” computer what a good equity analysts does.
• Apply this across thousands of stocks around the world.
• In a systematic manner,
• Go long and short to eliminate:
• Market risk (dollar vs. beta neutral).
• Sector risk, unless you want to bet on sectors.
• Most of the idiosyncratic risk.
• What risk is left?
19/28
Value: the HML factor

• The Value Factor as defined by Fama and French (1993)

1 1
HML = (Big value + Small value) − (Big growth + Small growth)
2 2
1 1
= (Big value − Big growth) + (Small value − Small growth)
2 2

• Breakpoints based on NYSE percentiles to limit reliance on small stocks.

• Portfolios are value-weighted within each of the 4 groups.
• Rebalancing: portfolios constructed at the end of June based on past-year BE and ME.
• Accounting data lagged to be sure it was known at the time of portfolio formation.
• Could use most recent price (clearly known) as in HMLdevil , Asness and Frazzini (2013). 20/28
Factor construction

• Many value metrics exist.

• HML: book to market.
• Even simpler: look at past price moves (De Bondt and Thaler, 1985).
• Earnings-to-price, dividend-to-price, cashflow-to-price, etc.
• More sophisticated and model-dependent versions.
• DDM-to-price or RIM-to-price (dividend discount model, residual income model).
• Many ways to trade on each value metric.
• Portfolio construction.
• Fama-French.
• Value-weighted vs. equal-weighted.
• Signal-weighted (Asness, Moskowitz, and Pedersen, 2013).
• Dollar long-dollar short vs. volatility scaled.
• Peer group.
• All stocks vs. intra-industry vs. inter-industry.
• Rebalancing rule.
• E.g., annual (Fama-French) or monthly or daily.

21/28
Statistical arbitrage

• Also quantitative:
• But not necessarily based on economic fundamentals.
• Based on arbitrage relations and statistical relations.

• Arbitrage relations:
• Siamese twin stocks.
• Index arbitrage.

• Statistical relations.
• Pairs trading.
• Reversals.

22/28
Statistical arbitrage: twin stocks

• Siamese twin stocks:

• Based on a dual-listed company.
• Two companies incorporated in different countries contractually agreeing to operate their
businesses as if they were a single enterprise, while retaining their separate legal identity
and existing stock exchange listings.
• In integrated and efficient financial markets, stock prices of the twin pair should move in
lockstep.
• In practice, there are deviations from parity. Each stock moves partly with its own market.
• Trade: buy low, sell high.

• Multiple share classes.

• Gives rise to similar trade.
• Often named “A” and “B” shares.
• Usually differ in control rights (votes).

23/28
Pairs trading and index arbitrage

• Pairs trading.
• Similar to trading on twin stocks.
• But here stocks are not based on the same company.
• Not based an arbitrage relation, but a statistical relation.
• Find stocks that tend to move in lockstep.
• Identify a time when they don’t.
• Then put a trade betting they will move back together.

• Index arbitrage.
• Trade index against constituents.

24/28
High frequency trading

• “Arms race”
• Being fast is not important per se.
• Being faster is very important.
• Enhancing trading speed.
• Co-location of computers on the electronic exchange.
• Engineering.
• Placing and updating limit orders continually.
• Determine the equilibrium price.
• Submit limit orders around it.
• Cancel orders when equilibrium price changes, and submit new ones.
• Manage inventory risk.
• Continually hedge market and industry exposures.
• While often a liquidity-providing strategy, may also take liquidity.
• Hit stale limit orders.
• Take liquidity to hedge.

25/28
Sources of HFT profits

• Market making.
• HFT are today’s market makers in many electronic markets and provide liquidity.
• Profits from market making (i.e. price of liquidity service): bid-ask spread, market impact.
• Arbitrage between different trading venues.
• The early bird gets the worm.
• Analysis of the order-flow.
• Is somebody trying to trade a large block?
• News.
• Be the first to interpret and trade on e.g. earnings announcements.
• Statistical arbitrage strategies.
• E.g. index arbitrage, short term reversal, and pairs trading.
• And others.

26/28
Differences between HFT and traditional hedge funds

• HFT do not need much capital.

• Typically positions are not held over-night.
• External investors may not be needed, and, if so, no need for sales and marketing.

• HFT is highly technology intensive.

• HFT are secretive.
• Probably many proprietary strategies which works only as long as they are proprietary.
• Much of what we know is anecdotal.

• HFT strategies are difficult to back-test.

• You need real-time data.
• Your own market impact may be large.

• Strategies need not have an economical motivation as long as they work.

27/28
References

• Pederson, Efficiently Inefficient, Ch. 3, 6-9

• Articles
• Asness and Frazzini, 2013, The Devil in HML’s Details, Journal of Portfolio Management
• Asness, Moskowitz, and Pedersen, 2013, Value and Momentum Everywhere, Journal of Finance
• DeBondt and Thaler, 1985, Does the Stock Market Overreact?, Journal of Finance
• Fama and French, 1993, Common Risk Factors in the Returns on Stocks and Bonds, Journal of
Financial Economics

28/28