Poroflio part 6 Flashcards
What is the purpose of backtesting?
To simulate an investment process using historical data and assess the risk and return properties of a strategy.
What does backtesting help investors do?
Reassure about strategy performance, refine strategies, and improve investment processes.
Why is backtesting intuitive?
It mimics the real-life process of formulating strategies, testing them with historical data, and evaluating outcomes.
Who uses backtesting?
Both systematic/quantitative managers and fundamental managers.
What key assumption does backtesting rely on?
That the future will resemble the past to some degree.
What is a major risk of backtesting?
A model that performs well in backtesting may fail in real-world performance.
Why might good real-world models fail in backtesting?
If they do not exhibit predictive power in historical simulations, they are unlikely to be implemented.
What technological advances have made backtesting easier?
Availability of large data sets and increased computing power.
What are the three fundamental steps in backtesting?
Strategy design, historical investment simulation, analysis of backtesting output.
What is the first step in backtesting?
Determine assumptions, investment objectives, and investment universe.
What complicates return definition when investing globally?
Currency denomination and whether investments are hedged.
Why is long historical data preferred in backtesting?
To maximize confidence in the results.
What are common factor categories used in equity strategies?
Defensive value, cyclical value, growth, momentum, analyst sentiment, profitability, leverage, earnings quality.
What is data mining trap in backtesting?
Selecting factors based only on good past performance without logical or economic rationale.
What is risk parity portfolio construction?
Combining factors so that each contributes equally to overall portfolio risk.
What is rolling window backtesting?
Walk-forward system that continually recalibrates and rebalances as new data arrives.
What is a key assumption in rolling window backtesting?
Past patterns of performance will repeat over time.
What is a potential drawback of rolling window backtesting?
It may not capture the dynamic nature of markets and extreme downside risks.
How is a multifactor strategy backtested?
First, create factor portfolios with walk-forward backtesting; then combine factors into BM or RP portfolios and rebalance monthly.
What is the Fama and French hedged portfolio methodology?
Long top quintile of stocks by factor scores, short bottom quintile, and rebalance monthly.
Why might backtesting methodologies produce different results?
Non-linear relationships between factors and returns, different methodologies capture different aspects.
What performance metrics are typically used to analyze backtesting output?
Sharpe ratio, maximum drawdown, Sortino ratio, average return, volatility.
What is the advantage of using multiple backtesting methods?
Ideally, similar results from multiple methods increase confidence in the strategy’s robustness.
What does VaR measure in backtesting?
Minimum expected loss over a time period at a given probability level.
What does Conditional VaR (CVaR) measure?
Average loss given that the loss exceeds VaR.
What is maximum drawdown?
Greatest loss from peak to trough in a portfolio’s cumulative returns.
Why use a log scale for cumulative return plots?
It makes equal percentage changes visually consistent and easier to spot structural breaks.
What is survivorship bias?
Error from only considering surviving securities, ignoring those that disappeared over time.
How can survivorship bias be avoided?
Using point-in-time data that captures securities active at each moment in history.
What is look-ahead bias?
Using information in backtesting that would not have been available at the decision time.
How can look-ahead bias be minimized?
Using point-in-time data or applying lag assumptions to reporting periods.
What is data snooping?
Testing many strategies and selecting the one with best performance, risking false positives.
How can data snooping be addressed?
Use higher critical t-values for significance and apply cross-validation techniques.
What is cross-validation?
Testing a model on a different data set than the one used to create it to avoid overfitting.
Why is rolling-window backtesting considered a form of cross-validation?
It uses in-sample periods to tune models and out-of-sample periods to test them sequentially.
What is historical scenario analysis?
Testing investment strategies over different historical regimes (e.g., recessions, volatility shifts).
What defines a high-volatility regime in historical scenario analysis?
Periods when the VIX is above its five-year moving average.
What is the main difference between historical simulation and Monte Carlo simulation?
Historical simulation resamples actual past returns; Monte Carlo draws from assumed probability distributions.
What advantage does Monte Carlo simulation have over historical simulation?
Monte Carlo can simulate never-before-seen events and account for non-normal distributions.
What is bootstrapping in historical simulation?
Sampling historical returns with replacement.
Why is fitting the correct distribution crucial in Monte Carlo simulation?
To accurately capture fat tails, skewness, and tail dependence in asset returns.
What is tail dependence?
Correlation between extreme outcomes in different assets.
What is a multivariate distribution in Monte Carlo simulation?
A distribution that models the joint behavior of multiple correlated variables.
How is sensitivity analysis used in simulation?
By changing the assumed distribution (e.g., from normal to skewed t) and rerunning the simulation.
Why conduct sensitivity analysis?
To assess how model misspecification affects risk and return estimates.
What is the multivariate skewed t-distribution?
A distribution allowing for fat tails and skewness, used to better model real-world asset returns.
What are key outputs to compare when interpreting simulation results?
Sharpe ratio, CVaR, and downside risk measures.
