Glossary
The terms behind an honest backtest
Backtesting has its own vocabulary, and the words that matter most are the ones that decide whether a strategy’s edge is real. Here they are, in plain English — the same definitions Thuztra’s engine lives by.
Survivorship bias
Survivorship bias is the error of backtesting only on assets that survived to today, which silently deletes past failures and flatters results.
Slippage
Slippage is the gap between the price a backtest assumes and the price a real order actually fills at, especially in size or illiquid names.
Transaction costs
Transaction costs are the brokerage, fees and spread paid on every trade — a drag a backtest must charge or it overstates returns.
Walk-forward analysis
Walk-forward analysis repeatedly tunes a strategy on one window and tests it on the next unseen window, exposing overfitting that a single backtest hides.
Luck test (Monte Carlo permutation test)
A luck test re-runs a strategy on hundreds of shuffled versions of the market to estimate the chance its result is just luck — reported as a p-value.
Maximum drawdown
Maximum drawdown is the largest peak-to-trough fall in a strategy's equity — the worst loss you would have had to sit through.
CAGR (compound annual growth rate)
CAGR is the constant annual rate that would grow a starting amount to its ending value over the period — a single, smoothed yearly return.
Alpha
Alpha is the return a strategy earns above its benchmark — the value added (or destroyed) versus simply holding the index.
Rebalancing
Rebalancing is periodically resetting a portfolio back to its target weights — selling what grew, buying what lagged — on a fixed schedule.
Lookback period
The lookback period is the trailing window of past data a strategy uses to compute a signal, such as a 6-month momentum rank.
Look-ahead bias
Look-ahead bias is using information in a backtest that would not have been known yet at the moment of the simulated decision.