How to read a tipster's results before you automate them

Last checked 13 October 2026

Automating a tipster means their next losing run lands on your Betfair account while you're at work. So the record has to do the arguing before you switch anything on. Most records are written to be glanced at. Here's how to read one properly, in about ten minutes.

Profit in points tells you almost nothing on its own

"+50 points" sounds like a result. It isn't, until you know how many bets it took. Fifty points from 100 bets is a 50% return. Fifty points from 2,000 bets is 2.5%, and once commission comes off, close to nothing. The number you want is ROI: profit divided by total stakes, as a percentage. Every serious record quotes it. If one doesn't, work it out yourself and ask why they didn't.

Strike rate only means something next to the price

A 20% strike rate is excellent at average odds of 8.0 and a slow death at 4.0. The quick check is whether the strike rate is above one divided by the average price: at 5.0 you need to win more than one in five just to stand still before commission. So 20% at 5.0 is break-even, 25% at 5.0 is a 25% return, and 25% at 4.0 is nothing at all.

For lay records it's the other way round, and more dangerous: a 90% strike rate looks wonderful until you see the average price was 11.0 and the losing 10% cost more than the winning 90% made. The sums are in lay betting liability explained.

Sample size: how many bets before the number is real

This is the one nobody wants to hear. Luck moves a record a long way over a short run. For a backing system at prices around 5.0, chance alone moves the ROI by about 20 points either way over 100 bets, about 7 points over 1,000, and about 5 points over 2,000. So a system showing 10% over 100 bets could easily be a losing system having a good month. The same 10% over 1,500 bets is very hard to produce by luck.

Rough guide: under 200 bets, you're looking at weather. Between 500 and 1,000, you're looking at something, but hold it loosely. Past 1,000 at level stakes with the losers included, the record has earned a proper look.

Drawdown: the number that sizes your bank

Drawdown is the fall from the record's highest point to its lowest point afterwards, in points. It's the worst stretch anyone following the system has lived through. If the worst drawdown is 60 points and you stake £10, you'd have been £600 down from the peak at some stage, and you'd have needed to keep betting through it to get to the profit on the other side.

Size the bank from this number, not from the profit, and assume the next drawdown will be worse than the last one, because over a long enough record it always is. A bank that can't survive the worst run in the record can't survive the system.

What price was the record kept at?

This is the question that separates records a bot can follow from records nobody can.

• Betfair SP. Checkable against the official number after every race, and a bot placing at SP gets exactly it. A record at BSP transfers to your account one for one. It's why our own systems are kept at it; Betfair SP explained covers the detail.
• Advised price. The price available when the tip went out, often at 9am. Real at the time, and usually gone by the afternoon. A record built on advised prices can be honest and still be unreachable, because every follower chasing the same price moves it.
• Bookmaker best price or best odds guaranteed. Not available on the Exchange, so not available to a bot. Ignore it for automation purposes.

Then the follow-up: is it before or after commission? A 6% record at BSP before commission is a different proposition from 6% after it. Ask. A tipster who knows the answer will tell you in one line.

Are the losers in?

A list of winners isn't a record. The record is every bet, including the ones that lost and the months that went backwards, updated on a schedule the tipster doesn't control. Look for results that are added daily, that show losing bets with the same prominence as winning ones, and that you can trace back to a price somebody else published. That's the standard the systems on PuntHub are held to: every bet goes on the record, the losing runs stay there, and for the Gap systems the results come straight out of xCloudBot rather than from anyone's notebook.

Lifetime figures hide the shape

A record that made 300 points, 280 of them in one month two years ago, has been flat or losing ever since. The lifetime ROI won't tell you that. The month-by-month table will. Look at how many months were profitable, how long the longest losing stretch was in months rather than bets, and whether the recent year looks like the whole record or like a different system.

A checklist

• ROI, not points. What's the return on total stakes?
• How many bets, and over how long?
• Strike rate alongside average price. Does it clear one-over-the-odds?
• Worst drawdown in points. Can your bank take it, and then some?
• What price is the record at? Can a bot actually get that price?
• Before or after commission?
• Are the losing bets shown, and are results added daily?
• Does the last twelve months look like the lifetime record?
• For lays: return on stake or on liability, and the biggest price ever laid?

Then watch it on paper first

Once a record passes the checklist, you still don't know how it feels to follow. Copy the tipster into xCloudBot, set the system to simulation mode and level stakes, and let it run for a few weeks. You'll see the bets, the prices the bot actually got, and the first losing run, without any money on. Compare what the simulation did with what the tipster published for the same weeks. If they match, switch it on small. If they don't, you've learned something cheaply. Staking plans for a betting bot covers what to do with stakes after that.

Questions people ask