The expected-value formula
For one unit staked, EV = p × decimal odds − 1. At a 55% probability and odds of 2.00, EV is 0.55 × 2 − 1 = 0.10, or a mathematical +10% per unit staked.
If the probability is 48%, the same price gives −4%. High odds alone do not create value.
Break-even hit rate
The minimum hit rate is 1 / odds. At 1.80 you need to win more than 55.56%; at 2.50, more than 40%. This is a useful threshold, but it does not show whether you estimate probability better than the market.
Actual hit rate should only be compared with the threshold using a consistent sample with recorded prices and matching settlement rules.
The cost of probability error
Suppose a model estimates 55%, but the true probability is closer to 50%. At odds of 2.00 the claimed +10% edge disappears. Small estimated edges are easily erased by model error, line movement or execution constraints.
Calibration is therefore valuable: outcomes assessed near 60% should occur about 60% of the time over a large sample. oddmind.app publishes this check on its Accuracy page.
Why high accuracy is not the same as profit
You can correctly pick many 1.20 favourites and still lose after several upsets. Conversely, a strategy below 50% may have positive expectation at sufficiently high prices.
Verified ROI requires executable odds recorded with each signal. With hit history alone, accuracy can be reported honestly, but return cannot.
Quick answers
Does positive EV guarantee profit?
No. It describes an average expectation if the probability is correct and the sample is sufficiently large.
How is value different from accuracy?
Accuracy counts correct signals; value also accounts for the price and estimated probability of each outcome.
Can historical ROI be calculated without recorded odds?
No. Without publication-time prices, the result would be an assumption rather than verifiable return.
