Break-even win rate by risk-reward ratio
Break-even win rate = risk ÷ (risk + reward): 50% at 1:1, 33.3% at 1:2, 25.0% at 1:3, before costs.
| Risk : reward | Break-even win rate | With costs of 5% of the risk |
|---|---|---|
| 1:1 | 50.0% | 52.5% |
| 1:1.5 | 40.0% | 42.0% |
| 1:2 | 33.3% | 35.0% |
| 1:3 | 25.0% | 26.3% |
| 2:1 (risk 2 to make 1) | 66.7% | 70.0% |
| 3:1 (risk 3 to make 1) | 75.0% | 78.8% |
Why high win rates can still lose
A setup with a tight target and a wide stop needs to win most of the time just to stand still: risk 30 points to make 20, pay 1 point of costs, and break-even is 62.0%. So “it hits 65% of the time” says nothing until you set it beside the stop and target you would actually trade. Our own opening range breakout backtest is an example: its verdicts are after costs, and the win rate alone didn't decide the result.
Is the win rate enough?
Only when the whole range clears break-even. 82 hits in 126 sessions shows 65.1%, but the true rate is likely anywhere from 56.4% to 72.8%.
What the range means
A count is a sample. If the setup's true long-run rate were 60%, about one run of 126 sessions in seven would still show 65% or more; the Wilson interval is the set of true rates a count like yours is consistent with, at the confidence you pick. Against a 62.0% break-even, 56.4% to 72.8% straddles the line, so the calculator answers “can't tell yet”, and says the same rate would need about 955 sessions before its low end cleared it. A bigger gap needs far fewer: 68 hits in 100 at a 1:1 target and stop gives 58.3% to 76.3%, clear of 50%.
How much a 3-month, 6-month or 1-year count can say
About 63, 126 and 252 sessions. For a daily setup near 65%, that is a range of roughly ±11, ±8 and ±6 points at 95% confidence.
Short windows, long windows
A short window reacts to what the market has done lately and carries a wide range; a long one narrows the range and may mix different market conditions. Neither is wrong. A rate that holds across several windows deserves more trust than one that shows on a single window, and a setup that doesn't trigger every day has fewer sessions than the calendar suggests: count its occurrences, never the days. The interval also assumes sessions are independent, and consecutive days often aren't, so treat the range as the least uncertainty you have.
Weekday and other filters: is the better number real?
Often not yet. 20 of 25 Mondays (80.0%) against 62 of 101 other days (61.4%): the gap could be −2.5 to +33.4 points.
The best of five weekdays
A one-weekday filter cuts six months to about 25 sessions, where 80.0% observed fits anything from 60.9% to 91.1%. And if you looked at all five weekdays and kept the best, one of them was always going to look best. The second tab holds each of the five comparisons to 99% so the set keeps 95% (Bonferroni): the same Monday gap then runs from −9.5 to +36.9 points. Zero inside the range means the filter hasn't earned its place yet.
Where the counts come from
Your journal, a backtest, or a statistics service. We publish our own counts for the opening range and the first hour on gold, Bitcoin and Ether.
Our counts, and their limits
Initial balance vs opening range counts how often each broke, on which side and how far, on 164 gold days and 1,477 days each of Bitcoin and Ether. What we measured tests one opening range rule after costs on the same markets. Both say what they leave out: the Nasdaq 100 and S&P 500 aren't counted yet, and neither is updated daily. To count your own trades, the backtest sample size calculator takes results in R.
edgeful or ProEA's own tools?
For probabilities across many setups and tickers, kept current, without building anything: edgeful. For an opening range rule on your own chart, with its source: ours. We sell ours, and edgeful may pay us.
What each one is for
We're an edgeful affiliate and haven't subscribed to it or tested it.
What follows is from edgeful's own pages, read 11 November 2026.
edgeful is a subscription statistics platform: its pricing page lists 150+ reports across stocks, futures, forex and crypto, each filterable by weekday and date range, and its help centre says report data goes back to May 2018 (history and lookback limits). A report library is a job we don't do: our opening range and first-hour counts cover three markets and are a snapshot. What we sell is different in kind. Opening Range Desk and Opening Range Desk Pro run an opening range rule on your TradingView chart, with readable Pine source, and record that chart's own results; both are paid. Edge Hunt is our paid research room on how gold, the Nasdaq 100, the S&P 500 and Bitcoin behaved in the past, which claims no profitable system. None of the three is a report library you can filter. The edgeful plan picker has its plans, prices and no-trial, no-refund terms, and our comparison sets Opening Range Desk Pro beside it.