Martingale is borrowed from 18th-century gambling. The original logic: if I keep doubling my bet after each loss, the eventual win covers all prior losses plus one unit of profit. So as long as my bankroll is "big enough," I'm guaranteed to make money. The math is mathematically true. The math is also fatal.
The catch is that bankroll is never actually unlimited and a long-enough losing streak will ALWAYS happen, given enough trades. After 8 consecutive losses your stake is 256x the original. After 12 it's 4,096x. Markets routinely deliver streaks at that length — fat tails are the rule, not the exception. When the streak that ends it lands, you don't lose 1 unit. You lose everything.
Variants and how they end: - Strict martingale (1, 2, 4, 8, 16…) — guaranteed account-killer. The expected value is negative once you account for finite capital. - Anti-martingale (size up after WINS) — actually has a defensible logic. Letting winners ride extracts edge from streaks. - "Martingale is fine, my system never has 5 losses in a row" — every backtested system says that. Live markets don't read backtests. - Soft martingale (1.5x or 1.3x stake after loss) — slower account-killer. Same direction.
Martingale isn't usually deliberate. It sneaks in. A trader has a 1R loss, then revenge-sizes 1.5R on the next trade to "make it back." That's a martingale tilt. Two losses in, they're at 2.25R. By the time they recognize the pattern, the account math has shifted into territory where one more bad trade flushes a quarter of equity.
In TradeOnyx your Risk-per-trade setting is the single biggest defense against this. The dashboard flags every trade that exceeds your declared risk — and if you skim the Journal tab after a losing day, those flags are usually clustered. That's the martingale signature.
The deeper signal lives in the Trades tab: filter for the last 5 trades after a losing trade vs the last 5 trades after a winning trade. If the post-loss group is consistently bigger in size, that's tilt — not strategy. TradeOnyx surfaces both groups so the comparison takes one glance, not a spreadsheet evening.