Variance: Why a Good Model Still Has Terrible Weeks

A strategy that wins 55% of bets at even money is genuinely profitable, and it will still produce a run of six consecutive losses roughly every hundred bets. That is not the system failing. That is arithmetic. Coin flips clump, and so do football results.

This is the single most expensive thing a bettor can misunderstand. The typical failure is not picking badly, it is abandoning a working approach three weeks into an ordinary downswing, or doubling stakes to recover, which converts a survivable run into a terminal one.

The practical defence is deciding in advance what a bad run looks like. If you know a six-loss streak is normal, it stops being evidence about your method. If you have not done that arithmetic, every losing week feels like proof, and you will keep switching systems forever.

Key points

55% at even money still produces six-loss streaks regularly
Judge a method over hundreds of bets, never over a week
Decide your acceptable drawdown before you start, not during
Chasing converts a normal downswing into a permanent loss

Frequently asked

How long can a losing run last if my method is sound?

Longer than feels reasonable. At a 55% strike rate, runs of six or seven losses appear regularly, and a bad month is entirely compatible with a profitable year.

How do I tell variance from a broken model?

Variance shows up in results while your process is unchanged and you are still beating the closing line. A broken model shows up as consistently losing to the closing line as well as losing money.

Should I reduce stakes during a downswing?

Reducing is defensible; increasing is not. Flat staking through the run is usually the right answer, and it is why flat staking exists.

Put it to use

Variance: Why a Good Model Still Has Terrible Weeks

Key points

Frequently asked

Put it to use

Read next