Start With the Complete Sample
A useful record includes wins, losses, pushes, voids, odds, stake size, grading date, and the time window being measured. A cherry-picked streak or a screenshot of winning plays is not the same thing as a tracked production record.
Look for enough detail to understand what the service actually issued to customers. Research, backtests, and shadow testing should be separated from customer-facing performance.
Judge Results With Odds and Units
Win rate alone can mislead because a 55% record at heavy negative odds may perform worse than a lower hit rate at better prices. Units and ROI put the result in betting terms rather than just scoreboard terms.
A disciplined record should make losing periods visible. Drawdown, sample size, and recent form help set expectations without pretending any model can remove uncertainty.
Check the Rules Behind the Record
A transparent service should define how picks are graded, when odds are captured, whether pushes and voids are counted, and whether every issued pick is included.
The most important question is simple: could the same rules reproduce the public performance numbers from the underlying ledger?
A Five-Bet Audit You Can Reproduce
Consider five hypothetical one-unit-risk bets: wins at +120 and -110, two losses, and one push. Net profit is 1.20 + 0.9091 - 1 - 1 = 0.1091 units. The win rate among decisions is 2 / 4 = 50%; the push is not a win.
If a report excludes returned push stakes from its turnover denominator, ROI is 0.1091 / 4 = 2.73%. Including the original pushed stake gives 0.1091 / 5 = 2.18%. Neither percentage can be interpreted without the stated convention. These are illustrative grading choices, not a claim about an ESA field.
Key Takeaway
ESA reports customer-facing production results with sample size, units, ROI, and boundaries around what is and is not part of the public record.