Why your backtest lies: survivorship bias

The most common and most expensive mistake in a retail backtest is survivorship bias: using today's S&P 500 to simulate the past. Companies that went bankrupt or left the index vanish from your sample… and with them, their losses.

The result is an inflated return you'd never have captured live, because at the time those companies WERE in your universe and WOULD have hurt you.

We reconstruct the index point-in-time: at each historical date we use exactly the companies that were members then, including those that later fell, with a delisting-aware exit. It's more work and uglier numbers — but real.

That's why, when you see a CAGR on Soberquant, it's survivorship-free and net of costs. Not the prettiest; the one you'd actually have lived.