What we learn building AniQuant: honest backtesting, validation and the mistakes that cost money. With our own numbers and no hype.
Doubling your bet after a loss looks foolproof: most of the time you end up ahead. We measured it with 100,000 series of a fair coin. The martingale ends up ahead 61% of the time… and is ruined 27.5% of the time. No money management creates an edge that does not exist.
Fear, greed, hope, frustration. A discretionary trader lives through them on every trade. A quantitative trader doesn't make them vanish by magic: they take them off the screen and lock them inside a plan that is decided before a single dollar is at risk.
We let our strategy generator loose on a randomly built ES where, by construction, there is nothing to win. It found a hundred spectacular strategies that passed every filter. Only one test was able to unmask them.
We tracked the 7,532 stocks listed on Wall Street on 3 January 2000. Only 1,436 are still around. Forget the rest, and your backtest looks three times better than reality.