Staking Strategies

Three ways to size a bet, backtested honestly.

These are position-sizing schemes, not signals. The same qualifying picks, replayed under three different stake-sizing rules. Bankroll growth here compounds weekly stakes; it’s a different metric from the flat, per-signal ROI shown on the Track Record page.

Live performance

Same real account, replayed under all three.

Real settlement history, real compounding bankroll, not the backtest below. This is what the live account has actually done under each strategy’s stake sizing, since inception.

StrategyTotal returnMax drawdown

Advanced

+1661.3%

38.7%

Conservative

+2418.2%

47.8%

Risky

+3824.9%

67.3%

Higher-return strategies carry real added drawdown risk. Risky is named Risky on purpose. Historical results are not a guarantee of future performance.

Backtested comparison

Two ways to beat flat staking, one default.

The Advanced track sizes every qualifying signal the same weekly stake by default. We also researched sizing each bet by its own odds bucket instead: more on historically stronger odds ranges, less on weaker ones. Pro subscribers can pick any of the three as a suggested stake per signal on the Live Signals dashboard.

Strategyvs. Advanced stakingMax drawdownWorst week

Advanced

Default, Pro subscribers

—

58.1%

−36.1%

Conservative

Pro

+17.3%

58.1%

−38.0%

Risky

Pro

+80.0%

59.7%

−47.7%

Bankroll growth, compounding weekly stakes. A one-time, out-of-sample backtest result, not a live-updating figure.

Read this honestly, not as a sure thing: both figures come from one out-of-sample test: bucket weights derived from the 2024-25 season only, then applied to 2025-26 results that weren’t used to derive them. That’s real evidence, not a guarantee: it’s a single validation on two seasons of data. Risky in particular carries measurably higher drawdown risk for its larger backtested return. It’s named Risky on purpose. Conservative was chosen specifically because it showed no meaningful increase in drawdown over flat staking.