Betting your whole bankroll: full compounding

Reinvesting everything — what you bet and what you won — into the next trade, especially on "almost certain" signals at favorite prices. Backtest curves are beautiful. The first time the market turns seriously, the account hits zero and there is no next trade.

The math that does not forgive

Imagine a signal that hits 97% of the time while paying little each time — the typical profile of the expensive favorite. With your whole bankroll on each trade, your probability of surviving 50 trades is 0.97⁵⁰ ≈ 22%. That is: almost 4 out of 5 times, ruin arrives before trade 50, even if the signal is as good as you think. And if your 97% estimate is actually 95% — nobody measures their own hit rate to the decimal —, survival falls to ≈ 8%.

What we saw in our systems

We tried the full reinvestment design per asset in one of our paper systems: it worked until the first adverse turn of an asset arrived, and that single turn took the whole bankroll assigned to that asset. It did not take a streak: it took one window. Along the same lines, in high-probability trades we have measured that a single complete adverse trade costs what was won in eight to eleven perfect trades: when everything is in, that cost is paid with everything.

Why the backtest lies to you here

The backtest averages: it shows the final result of thousands of chained trades as if the account could "continue" after hitting zero. In real life, zero is an absorbing state: nothing compounds from zero. Any strategy that hits zero along the way is broken even if its historical average is pretty, because you live the path, not the average.

What to do instead

• A fixed, small fraction of your bankroll per trade, decided cold: if a single window can truly hurt you, the size is wrong, not the signal.
• Think about survival first: growth comes from still being alive within a year, not from the best week.
• Increase sizes with history measured in DEMO and small, never with the emotion of a good streak.

The full series

This page is part of the Money-Losing Strategies series, where we cover the traps we have tested and measured in our own systems. And if you want to see how a system that avoids these traps performs, the full history — failures included — is in the public statistics.