How to tell if a tipster is really profitable
A screenshot of a good run proves nothing. With yield, sample size and a little statistics you can better separate skill from luck.
What to ask for first
Before looking at any number, check that the data is complete and cannot be altered:
- All bets, including losers, with date, odds and stake.
- Published before the event starts, in a dated channel that cannot be edited.
- Odds that were really available. A price that lasted one minute is not the one you will get.
- Consistent stakes. If results are given in units, check that the unit does not change size when convenient.
If any of this is missing, the numbers below stop being useful.
Step 1: work out the real yield
Add up what was staked and the profit across the whole history, and apply the yield formula. Do it yourself from the raw data, without relying on the tipster’s own summary.
Look also at the odds you actually get. If you follow the tips late, your price will be worse and so will your yield. What matters to you is not the channel’s performance but yours.
Step 2: look at the sample size
This is where most conclusions fall apart. For flat-stake bets, a useful approximation of the noise in yield is:
Standard deviation of yield ≈ √(average odds − 1) ÷ √(number of bets) × 100
At average odds of 2.00 this gives:
| Bets | Standard deviation | Usual range from pure chance (±2 deviations) |
|---|---|---|
| 100 | 10.0 points | -20% to +20% |
| 500 | 4.5 points | -9% to +9% |
| 1,000 | 3.2 points | -6% to +6% |
| 5,000 | 1.4 points | -3% to +3% |
In plain terms: a bettor with no edge at all can show +15% over 100 bets without anything being strange about it. The range is centred on zero for simplicity; with the bookmaker’s margin, it would really sit a little lower.
Step 3: compare the yield with the noise
An example. A tipster with average odds of 2.00 and an 8% yield:
- Over 200 bets: the standard deviation is 7.1 points. That 8% is just over one deviation from zero. It is compatible with pure luck.
- Over 2,000 bets: the standard deviation is 2.2 points. That 8% is more than three deviations away. It is now hard to put down to chance alone.
This is an approximation, not a formal test, and higher odds need a bigger sample. At odds of 4.00 the noise is 1.7 times greater than at 2.00, so you would need about three times as many bets for the same confidence.
The bias that distorts everything
Imagine 1,000 tipster channels with no skill at all. After a while, a few will have brilliant results by pure chance, and those are the ones you will see advertised, because the others close or change name. This is called survivorship bias, and it means that finding someone with a great record is less surprising than it seems.
Warning signs
- Promises of winnings or “sure bets”.
- Only shows screenshots of wins.
- Deletes messages or resets the history from time to time.
- Gives no specific odds or stakes.
- Mentions fixed matches. This is a common scam.
- Pressures you to raise stakes or chase losses.
A better test than the result: CLV
Because results take so long to stop being noise, many analysts also compare a tipster’s prices with the closing ones. Consistently beating them is a quicker signal, though not an infallible one. We explain it in the guide to CLV.
How to apply this to the tipsters you follow
Record every bet you follow along with its tipster, and compare. BetTracker Pro separates profit, yield and win rate by tipster: just import your history and assign each bet. For a quick sum, use the yield calculator.
No past result guarantees the future. Even with a solid record, following a tipster carries a risk of losing money.
Frequently asked questions
How many bets does a tipster need to be reliable?
With odds around 2.00, fewer than 500 bets say little, and you should reach several thousand to speak with more confidence. High odds need even more.
Does a high win rate mean they are good?
Not on its own. A 70% hit rate at odds of 1.30 loses money. Win rate only makes sense alongside average odds, which is why yield is what you look at.
Is it worth paying for a tipster?
It depends on whether their real edge, applied to the odds you can get, covers what you pay. Measure it with your own data before deciding, and remember there is no guarantee.