Backtesting and Simulation Flashcards
Steps of Backtesting
- Strategy design
- Historical investment simulation
- Analysis of backtesting outputs
Strategy design in backtesting
Step 1
Specify the investment hypothesis and goal
Determine investment rules and process
Decide key parameters
Return definition
Rebalancing / reconstitution frequency
Start and end dates
Historical investment simulation in backtesting
Step 2
Construct a portfolio to be tested
Strategy
Portfolio securities
Investment hypothesis
Make sure it is rebalanced on a predetermined frequency
Analysis of back-testing outputs in backtesting
Step 3.
Calculate portfolio statistics
Compute key metrics
What do we do in Historical investment simulation for Backtesting multifactor models?
Backtesting a multifactor strategy is similar to the method introduced earlier, but the rolling-window procedure is implemented twice, once at each portfolio “layer.”
Rolling window
Once at the factor level
-Again at the factor portfolio level
Risk parity portfolio
A portfolio allocation scheme that weights stocks or factors based on a equal risk contribution.
High volatility factor: Lower weights
Low volatility factor: High weights
The sum of the total standard deviation of each factor / Number of factors
They usually perform better than the benchmark, hence why they’re leveraged.
Requires a complete variance-covariance matrix at each rebalacing date.
Objective of backtesting
To understand the risk–return tradeoff of an investment strategy by approximating the real-life investment process.
Historical Scenario Analysis
Type of backtesting that explores the performance of an investment strategy in different structural regimes and breaks.
NOT THE SAME AS Historical simulation
Bootstrapping
Refers to random sampling with replacement, often used in historical simulation.
Random sampling with replacement, also known as bootstrapping, is often used in historical simulations because the number of simulations needed is often larger than the size of the historical dataset
What are the two types of analysis in Simulation analysis
Historical simulation and Monte Carlo simulation
Historical simulation differs from Monte Carlo Simulation
It assumes that sampling the returns from the actual data provides sufficient guidance about future asset returns.
What is another name for Random sampling with replacement
Bootstraping
What is Data snooping?
A form of statistical bias manipulating data or analysis to artificially get statistically significant results.
Also known as P-Hacking
What is reverse stress testing?
Identifying a set of exposures and then determining what would stress thos risk factors.
