Backtesting ยท Lesson 3.6
Survivorship bias
If a backtest only includes stocks that still exist today, it implicitly assumes that every disastrous failure never happened. That makes the results look better than they really would have been.
Why this matters
Bankruptcies, takeunders, and delistings happen. A strategy back-tested against today's survivors is an optimistic version of the strategy you would have actually run.
How FactorSage handles it
When you backtest against the S&P 500 universe, FactorSage uses the point-in-time membership and keeps historical data for stocks that have since left the index or been delisted. Removals and replacements are part of the simulation.
Related
- Point-in-time universesBacktesting
- Survivorship BiasGlossary