CalculatorsUpdated 5 Sept 20262 min read
Skill or Luck: Is That Record Real?
A season of bets ends with a number. The hard part is knowing whether that number came from judgement or from variance, and the answer is almost never visible by looking at it.
Over 120 bets at +8.3%, a bettor with no edge at all would do this well about 1 in 4 times. That is common enough that luck is the simplest explanation, and the record is not yet evidence of anything. To get there you would need roughly 585 bets at this rate — about 465 more than you have.
How it reads#
- The verdict is the output. The statistics under it are there so you can check the arithmetic, not so you can quote them.
- ROI is the flat one-unit convention: a winner returns the odds minus one, a loser returns minus one, over the bets that had a price. It is the same convention we use for our own published record, so the two are comparable.
- Standard error is how much a record like yours bounces around from luck alone. If it is larger than your ROI, the ROI is not yet telling you anything.
- Bets for 95% is how long the same rate would have to keep up before luck stops being a comfortable explanation.
Why the odds matter as much as the count#
The swing on one flat-staked bet is set by the price: back at 2.00 and a single result moves you by one unit, back at 6.00 and it moves you by five. So a record of short-priced bets settles down quickly and a record of long shots barely settles at all. Two people can both show "+10% over 200 bets" and one of them has said something and the other has not.
This is also why "I hit 60% of my bets" is not a claim on its own — at what price is the whole question, which is what the break-even rate puts a number on.
What it does not tell you#
It cannot prove an edge. The best it can say is that luck is an unlikely explanation for what already happened. That is a real and useful statement, and it is a weaker one than "this will continue".
It assumes your bets are independent and honestly recorded. Six correlated legs on one afternoon are not six bets. A record that quietly drops the bad month is not a record.
It says nothing about why. A method can be genuinely ahead of the closing price and still lose money after margin, and it can beat the market on one league and give it all back on another. Closing line value asks a different and often earlier question: not whether you won, but whether you were getting the better of the price.
For the underlying reasoning — including why seven straight losses is routine rather than a sign anything is broken — read how many bets it takes to prove anything.
Related
Frequently asked questions
How many bets do I need before my ROI means anything?
It depends on the prices, not just the count. The swing on a single flat-staked bet grows with the odds, so a record built on 1.5-shots settles down far sooner than one built on 5.0-shots. Enter your own numbers above and the calculator returns the sample size your particular ROI and price level would need.
I'm up 8% over 120 bets. Is that good?
It is encouraging and it is not yet evidence. At around 2.5 average odds, a bettor with no edge whatsoever would match that roughly one time in four. That is not a rare enough event to rule luck out, which is why the honest answer at that sample is 'we don't know yet'.
What does the p-value here actually mean?
It is the chance that someone with no edge at all — betting entirely at fair prices — would have finished this far ahead or further, purely by chance. A small number means luck is a poor explanation. It does not measure how large your edge is, and it never proves one exists.
Should I count voided bets?
No, and this calculator leaves them out of both halves. A returned stake is neither a win nor a loss; putting voids in the denominator drags every record toward zero for reasons that have nothing to do with how well you picked.
Is my record uploaded anywhere?
No. Everything on this page runs in your browser. Nothing you type or paste is sent to us or stored, and the page works the same whether you have an account or not.