Win rate, betting ROI and sample size: reading a research record
Understand what win rate and flat-stake ROI measure, why stakes matter and why backtests need separate reporting.
Win rate is only one measure
Win rate is settled wins divided by settled wins plus losses. Voids and pending selections should be reported separately. Ten wins from twenty settled bets produce a 50% win rate, but the return depends on the odds, stake sizes and settlement rules.
A transparent flat-stake calculation
Consider twenty hypothetical bets of one unit each at decimal odds of 2.20. Ten wins earn 12 units of profit, and ten losses cost 10 units. Net profit is 2 units on 20 units staked: 10% betting ROI before costs. If all those prices were 1.80 instead, net profit would be −2 units and ROI −10%, despite the same 50% win rate. These are constructed illustrations, not Game Point Data results.
Keep fees visible
Betting ROI usually measures net betting profit divided by settled stakes; it does not automatically include the research subscription. For example, £1,000 monthly turnover at a hypothetical 5% betting ROI produces £50 before the subscription. A £49 fee would leave £1 before other costs. At −5%, the betting loss and fee together would be £99. No such ROI is promised, and more wagering is not a remedy.
Why a segment average is not a pooled return
An equal-weight average of strategy-segment returns gives each segment the same weight. A pooled return requires the underlying net profit and stakes, with overlapping matches counted once. Without those records, averaging percentages cannot establish the return of a combined service. Historical forecast counts may include overlapping entries.
Separate research from live evidence
Retrospectively chosen backtest segments benefit from knowledge of past outcomes. They are useful research leads but need prospective testing under frozen rules. A small winning sample can reflect chance; a losing period alone does not identify the cause. Review sample size, price evidence, withdrawals, costs and drawdowns. Monthly summaries should retain a complete internal record from which the figures can be checked.
Read the free execution guide · All research notes
18+ and subject to local requirements. Betting involves risk of loss. Past performance does not guarantee future results. Examples are hypothetical and exclude costs unless stated. No investment or asset-outperformance claim is made.