How Unusual Is Your Betting Record? A Probability Calculator
Forecast Record's simplified Luck or skill calculator shows how unusual a betting record would be if no pick had an edge: for example, 100 picks at average odds 2.00 returning +10.0% or better has probability 18.4% with no bookmaker margin and 3.5% with a 10.0% margin; the model, its limits and the margin sensitivity are explained below; page as of .
Source: simplified binomial model on the stated inputs
How unusual is this result?
Compare a recorded result with a specified baseline.
This simplified model assumes independent bets with the same win probability. It does not support half wins, half losses or correlated selections.
Check a record
The model and its assumptions
The model asks one question: how often would a record like this appear if no pick had an edge at all? A pick without an edge wins exactly as often as its price implies: at decimal odds 2.00 that is one time in two, at 3.00 one time in three. A win pays the odds minus the stake, a loss costs the stake.
With N picks at average odds o, the number of wins follows a binomial distribution with probability p = 1 / (o × (1 + margin)). A return on stakes of r needs at least N × (1 + r) / o wins; the calculator adds up the chance of every such outcome. For a return below what the model gives on average, it gives the chance of doing this badly or worse.
With the margin at 0, prices are treated as fair, so the model averages a 0.0% return. Real prices carry a bookmaker margin, which makes a no-edge record drift below zero. The table below shows the same record under several margins.
Result
Under the simplified model, where every pick wins exactly as often as its price implies (no bookmaker margin), a record of 100 picks at average odds 2.00 returns +10.0% or better with probability 18.4% — about 1 in 5.4 such records. That needs at least 55 wins; the model averages 50.0 wins, and the return of a no-edge record typically lands within ±10.0 percentage points of 0.0%.
This is the probability of a result at least this extreme under the selected model.
How the answer depends on the margin
The margin is an assumption you choose: the same record looks very different under different margins.
| Margin assumed | Probability | Result counted | Wins needed |
|---|---|---|---|
| 0.0% | 18.4% (1 in 5.4) | this well or better | at least 55 |
| 5.0% | 8.4% (1 in 12) | this well or better | at least 55 |
| 10.0% | 3.5% (1 in 29) | this well or better | at least 55 |
| 20.0% | 0.5% (1 in 200) | this well or better | at least 55 |
What the number says
- It is a probability under a stated model. A small probability says the result would be unusual if no pick had an edge.
- Selection matters: among 100 records with no edge, about one would look as extreme as a 1-in-100 result by chance alone, so the most unusual record in a list is not, by itself, evidence of skill.
- One average price is an approximation. With fair prices, the spread of the total return depends only on the sum of the prices, so it is the same for a record of longshots and for the same average made of mixed prices. What differs is the shape: longshots make the distribution more lopsided, with rarer and bigger wins, so the tail probabilities differ from those of one average price.
- Picks on the same match are not independent, and some markets settle as half wins or half losses; this model handles neither. The intervals on our profiles use a bootstrap over matches instead (how uncertainty is estimated).
- The record itself can contain errors. A record kept before kick-off and a history restored later are different evidence; on our profiles they are never added together.
- Nothing you enter is sent anywhere: the calculation runs in your browser.
Method and example
The example above is 100 picks at average odds 2.00 returning +10.0%. Profiles link here with their own numbers filled in only when their live record fits this model: our own pre-match odds snapshot, at least 30 settled picks excluding voids and no half outcomes.
The same data, machine-readable
The same text as this page, from the same snapshot.
Check us