How many bets does it take to evaluate a performance?

The short answer

No number of bets is enough, on its own, to make a record reliable. How much data you need to tell a method from plain variance depends on parameters you can identify. And that dependency can be calculated.

Why there isn't a single answer

"How many bets?" asks for a number, but the right answer is a list of parameters. The same number of decisions can be plenty in one case and far too few in another, depending on what's inside it.

This page doesn't ask which questions to put to a record, or how to judge a tipster. Those have their own pages. Here we look at what the required amount of data depends on, mathematically.

What the answer depends on

Six factors, all measurable on an existing record.

• The average odds you take your decisions at. They set the size of the variance directly.
• The associated win probability. The more extreme it is in either direction, the more concentrated or spread out the results are.
• The size of the assumed edge. A 1% edge and an 8% edge don't need anything like the same volume.
• The variance that follows from the two points above. Odds and probability together set how wide the possible swings around the expected value are.
• The spread of your stakes. Uneven stakes make an ROI harder to read than the same volume at fixed stakes, because a few big-stake bets can dominate the result.
• The metric you track. Strike rate, ROI and CLV don't settle on a stable value at the same speed.

A seventh factor is easier to miss: correlation between bets. Several decisions tied to the same event, the same competition or the same reasoning count for less than their raw number suggests. They add less independent information than the same number of genuinely distinct decisions.

Two profiles, two volume needs

A small edge at short odds and a clear edge at long odds don't need the same number of decisions to become readable. A 1% edge at average odds of 1.50 stays buried for a long time. The variance is modest, but it's enough to hide a signal that small. An 8% edge at average odds of 3.00 stands out much sooner. The variance per bet is wider, but the gap to detect is far bigger still.

The volume you need follows the gap between the signal you're looking for and the noise around it, not variance on its own. A clear signal breaks through faster than a faint one, even through more noise.

Two cases: 10 bets at +30%, 2,000 bets at +3%

Two records already compared on the ROI page show this mechanism at work. An ROI of +30% on 10 bets fits almost entirely inside the range that variance produces on a sample that small. A few different results would have been enough to reverse it. An ROI of +3% on 2,000 bets has far less room to be explained by chance alone. That doesn't guarantee it will repeat.

The second figure is smaller, and yet it carries more information. The volume behind it shrinks the share that variance can take up.

When I read a "+30%", I look for the number of bets behind it first. On 10 bets, that figure tells me almost nothing. On 2,000, the +3% holds my attention far more.

What the sample doesn't fix

A larger sample makes an existing gap easier to read, and it never makes an estimate more accurate. If the process producing the decisions is off from the start, more data just shows that bias more clearly. It doesn't correct it.

How to build up usable data

None of this can be calculated without a record kept properly, decision by decision. The tracking method (which columns to note, how to avoid gaps) has its own page.

What to record for every bet →

Applying this to someone else's record

The same volume requirements apply to a tipster or a forecaster. An example worked out elsewhere on the site shows why a high strike rate on a sample of very short favourites proves nothing. We don't redo it here: it's covered on the page about assessing a tipster.

How to assess a tipster's record →

The questions to ask any record, your own included →

No magic threshold appears on this page, or anywhere else on this site. Nobody should promise one: the number of decisions needed is calculated from the parameters above.

Put it into practice

Apply this guide to real odds.

Dig into the market

Odds movements are only part of the story. Here are the next topics to read.

Frequently asked questions

Sources and methodology