Horse Racing Stake Sizing: How to Set a Bank That Survives the Losing Runs

Last updated: 21 October 2026

The bet lost, and you already knew whose fault it was. The jockey gave it too much to do. The trainer ran it over the wrong trip. That tipster you follow has lost the plot. Anyone but you, and certainly not the size of the stake you had on. I know the feeling because I lived in it for years. When I started out I bet off the Postdata section of the Racing Post and nothing else, and when a selection won I was a genius, and when it lost it was the fault of that bloody newspaper.

The truth I avoided for far too long is duller and more useful. Losing runs are not a sign that your method is broken. They are a mathematical certainty, and horse racing stake sizing is simply the craft of making sure they cannot wipe you out before the winners arrive. Get the bank right and a bad patch is an inconvenience. Get it wrong and the same bad patch ends your betting altogether.

By the end of this guide you will be able to do four things: work out the losing run your own method should expect, set a bank big enough to survive it, turn that bank into a stake you can bet every day, and know when (and when not) to change it. It takes a calculator or a spreadsheet and about twenty minutes.

How big should my betting bank be for horse racing?

Work out your expected losing run from your strike rate, then hold a bank at least three times that long in points. At a 20% strike rate the maths points to a losing run of around 21 bets, so 60 points is a sensible floor, with each bet staked as a single point. That keeps two thirds of your bank intact even if the worst run lands the moment you start.

In this guide

• Why stake sizing is the thing backers blame last
• Step one: get a sample you can trust
• Step two: work out your expected losing run
• Step three: the three-times rule for a bank that survives
• Step four: set your point size and stay level
• What real strike rates and losing runs look like
• A worked example from start to finish
• When to resize your stake, and when to leave it alone
• Beyond flat points: percentage staking and Kelly
• Where stake sizing ends and edge begins
• Common stake sizing mistakes
• Frequently asked questions
• Do this today

This is one chapter of a bigger subject. For the full picture of how stakes, prices and volatility fit together, see the main horse racing staking plan guide. This page stays focused on the first job: sizing the bank and the stake.

Why stake sizing is the thing backers blame last

Stake sizing is the last thing most backers examine because it is the least exciting and the most personal. It is far easier to question the jockey, the going, or the source of your selections than to admit your bank was simply too small to survive a normal run of outs. I still feel the pull to blame the ride even now, after years of stripping the emotion out of my betting. It is human nature. When things go wrong we want it to be anyone’s fault but our own.

Accepting that losing streaks would be a permanent part of my life was the hardest part of turning betting into a full-time job. There are so many variables feeding into a selection that when it wins you tell yourself the right ones mattered, and when it loses you tell yourself the wrong ones did. The honest position is that nobody knows in advance which variables will prove relevant in a given race. You cannot remove the variance. You can only build a bank that absorbs it, which is exactly why understanding losing runs matters more than agonising over any single result.

Two different jobs: the bank and the stake

It helps to keep two ideas apart. The bank is the pot of money set aside for betting and nothing else. The stake is the slice of it that goes on one bet. Sizing is the link between them.

Make the bank too small for the stake and an ordinary bad patch finishes you. Make the stake too small for the bank and you survive, but the results crawl. The aim is a stake large enough to matter and small enough that the worst realistic run leaves you standing. The four steps below find that balance using your own numbers rather than a rule copied from someone else.

Step one: get a sample you can trust

Everything that follows rests on one number, your strike rate. A strike rate is the share of your bets that win. If it is wrong, every later figure is wrong with it, so this step matters more than it looks.

Paper trade your method until you have at least 100 results showing a profit. Without a real sample you are sizing a bank around a guess, and a guess is what got most of us into trouble in the first place. More is better. A strike rate from 30 bets can swing wildly from one batch to the next, while one from several hundred settles down.

Measure it the way you will actually bet

Record every selection, not just the ones you liked afterwards. Record the price you could really have taken. In practice that means Betfair SP (the exchange’s starting price) or a bookmaker price you can genuinely get, not an optimistic number from the morning. A strike rate built on prices you cannot obtain tells you nothing about the bank you need.

If you do not yet keep a clean record, start now. The free bet tracking tool will hold every selection, price and result in one place, which is all this step requires.

Step two: work out your expected losing run

Your expected losing run comes straight out of your strike rate, and there is a simple formula for it. Say those 100 paper bets returned 20 winners. That is a 20% strike rate, which as a probability is simply 0.20. The formula for the losing run you should expect across that many bets is LOG(100) divided by minus LOG(1 minus 0.20). In a spreadsheet you would type =LOG(100)/-LOG(1-0.2) and it returns 20.63. In plain English, over a run of 100 bets at a 20% strike rate you can reasonably expect to hit a losing sequence of around 21 in a row at some point.

The two pieces are easy to adjust. The first number, inside the first LOG, is how many bets you are planning across. The second, after the minus LOG, is one minus your strike rate as a decimal. A 30% strike rate makes it 1 minus 0.30. A 25% strike rate makes it 1 minus 0.25. The higher your strike rate, the shorter the losing run the formula predicts, which is precisely why backing more likely winners buys you a calmer ride even before you think about price.

What changes the answer

• Your strike rate. Lower strike rates mean longer runs, and the effect is steep. Halving the strike rate more than doubles the expected run.
• How many bets you plan across. The more bets you place, the longer the worst run you will meet. Plan across the number of bets you will realistically place in a year, not just the first 100.
• Your tolerance. The formula gives a typical worst case, not a ceiling. Exceptional runs can and do arrive, which is why the next step builds in a margin.

A quick reference table

The table below applies the formula to a plan of 100 bets at different strike rates. These are illustrative worked examples, not real results. Work out your own with the formula above.

Notice how fast the bank requirement falls as the strike rate rises. A method that wins one time in ten needs more than four times the breathing room of one that wins one time in three. If your method sits at the lower end, you are not doing anything wrong, but you must size accordingly. It also helps to know whether your strike rate comes with a genuine pricing edge, which the section on where stake sizing ends and edge begins covers.

Plan across more bets and the run grows slowly rather than dramatically. As an illustration only, the same 20% strike rate planned across 1,000 bets gives an expected run of about 31 rather than 21. That is why the bank should be built with margin, not to the exact number.

Step three: the three-times rule for a bank that survives

Once you know your expected losing run, multiply it by three to set your bank. That is the rule I have used for years and it is built on a hard lesson. If the formula says expect a losing run of around 21 and you walk in with only 20 points, a bad start can clean you out entirely. With £20 in your pocket betting £1 a time, the worst run wipes you before the method ever gets the chance to prove itself.

So at a 20% strike rate, with an expected losing run near 21, I would start with a bank of roughly 60 points. If you had £600 to commit, that makes a point worth £10. Hit the worst losing run straight out of the gate and you still have two thirds of your bank in reserve, ready for the winning sequence that the same maths says is coming. The multiplier is not magic. It is breathing room, and breathing room is the difference between a temporary drawdown and a closed account.

It is worth being honest about what the three-times rule does and does not do. It does not guarantee survival, because no bank can rule out an exceptional run of variance. What it does is push the odds of ruin down to a level a disciplined bettor can live with, and that is the most any staking plan can promise.

What if I do not have that much money?

Then shrink the point, never the margin. If the sum says you need 60 points and you only hold £300, a point is £5, not £10. Cutting the bank to fit a stake you would like to bet is the single most common way to end up with a bank that cannot take a normal bad patch. The smaller point keeps the logic intact, and you can grow into a larger one later as the bank grows.

A drawdown is not the same as a losing run

One caution. A losing run counts consecutive losers. A drawdown measures the fall from your highest point to the lowest point that follows, and it can include winners scattered through the slide that never quite recover the old peak. So a bank sized only for the losing run can still feel pressure during a long patchy spell. The next step towards checking that is covered in how to calculate drawdowns and size your bankroll, which walks through the three spreadsheet columns that show your worst fall.

Step four: set your point size and stay level

Your point size is just your bank divided by the number of points you decided to hold. A 60-point bank of £600 means one point is £10, and every selection goes on at one point until you have a reason grounded in fresh results to change it. Level stakes are not glamorous, but they keep the maths above honest and they stop the single most common way backers detonate a healthy bank.

That way out is chasing. After a few losers the temptation is to double up to win it all back in one go, which quietly tears up every assumption your losing-run calculation was built on. A method that wins at 20% does not owe you a winner because you are due one. Each race starts fresh. Staking flat keeps you in step with the strike rate you actually measured rather than the one your frustration wishes you had.

Write the stake down before the racing starts

Decide the point value once, write it where you will see it, and place it on every selection. The goal is to take the decision away from the moment, when your mood is at its least reliable. If you want to stake more on a selection, that is a sign you believe its edge is bigger, and that belief belongs in your records first, tested over many bets, rather than in the stake box on a Saturday afternoon.

The same logic at the percentage level

A bank of 60 points is a stake of about 1.7% of the bank per bet. If that feels too big for your temperament, go lower, which means a bigger bank in points and a smaller point. My own simple starting rule in the bankroll management guide is 0.5% of the bank on each bet, which is the same as holding 200 points. It is a gentler start, and a convention rather than a tested formula. Both routes give the same answer in spirit: small, fixed, and sized so the longest realistic dry spell leaves most of the bank intact.

What real strike rates and losing runs look like

One caution on the strike rate itself. It has to be measured on real results at the prices you can actually get, which in practice means Betfair SP rather than an optimistic assumption. Race Advisor’s data covers more than 1.5 million UK runners across over 163,000 races between 2013 and 2026, and it sets the expectation cleanly. Back the favourite in every race and you win 34.4% of the time. The top-rated horse on our PR Ratings wins 32.3%. Drop down the market and the strike rate falls away fast: the second favourite wins 20.4%, the third 14.0%, the fourth 10.1%, and everything from the fifth in the betting down wins just 3.9% between them. A disciplined method that lands in the 15% to 35% band the formula above assumes is realistic, not wishful.

Now look at what that does to the losing-run sum. Across those same bets, level-backing the favourite produced a worst losing run of 27 in a row, and the top-rated runner 30, which is almost exactly what the formula predicts once you feed in the real number of bets you are placing. The point is not that favourites are a good bet. The A/E sits at 1.00 on the favourite and 0.99 on the top-rated horse, so the market has already priced them fairly and there is no edge in simply backing the obvious. The point is that even the highest strike rates in the game still bury you in long losing runs, so the bank has to be built for them.

What this means for your own method

• Expect long runs even from good methods. If the most popular selections in racing can lose 27 or more in a row, yours will too at some point.
• Do not read a long run as proof of failure. Compare it with what the formula predicted before you change anything.
• Do not borrow someone else’s bank number. A bank sized for a 35% method is far too small for a 15% one.

For a deeper look at how real streaks behave across the data, see horse racing losing runs.

A worked example from start to finish

Here is the whole process on one made-up method. Every number in this example is illustrative, not real results.

1. Sample. You paper trade 100 selections at prices you could genuinely take. 22 win. Your strike rate is 22%, or 0.22.
2. Expected losing run. In a spreadsheet, =LOG(100)/-LOG(1-0.22) returns about 18.5. Expect a losing run of around 18 or 19 at some point.
3. Bank in points. Three times 18.5 is about 55 points.
4. Point size. You have £900 set aside for betting. £900 divided by 55 is about £16 a point.
5. Stress test. The expected losing run of about 18.5 bets at £16 costs roughly £300, about a third of the bank. Painful, not fatal.
6. Stay level. Every selection goes on at £16 until your next scheduled review.

The test at step five is worth doing every time. If a third of the bank gone feels unbearable, do not change the maths. Choose a smaller point, so the same run costs less, and bet that instead. You are sizing for the person who has to sit through the run, and that is you.

When to resize your stake, and when to leave it alone

Stakes should change on a schedule, not on a feeling. My rule is to review on a fixed date, such as the first of each month, and never in the middle of a run in either direction.

Leave it alone when

• You are in the middle of a losing run that the formula said was possible.
• You have just had a big winner and the bank feels easy.
• A tipster, friend or forum post has made a persuasive case for betting bigger.

Review and resize when

• It is the scheduled review date. Divide the current bank by your number of points to get the new point size.
• You have collected enough fresh results to re-measure your strike rate. If it has moved, redo the formula and the bank target.
• You are planning across a much larger number of bets than before, which lengthens the expected run.

Resizing on a schedule has a nice side effect. The point gets smaller after a poor stretch and larger after a good one without any decision from you, and always in step with the bank. It is a calm way to be cautious in bad times and to grow in good ones.

Beyond flat points: percentage staking and Kelly

Level points are the honest default because they keep testing clean. Two natural next steps are staking a fixed percentage of the current bank and sizing by edge. Neither is needed to get started, and neither fixes a method with no edge.

Percentage staking

A fixed percentage of the current bank means the stake falls after losses and rises after wins, so a run of losers can never take the bank to zero in one go. It is the same idea as the schedule above, applied after every bet instead of every month. If you want it, the simplest version is in the bankroll management guide.

Kelly and proportional staking

The Kelly criterion sizes each stake by the size of your edge. It is powerful and easy to misuse, because it takes your estimate of the winning chance as gospel, so an optimistic estimate produces an oversized stake. If you want to use it, treat full Kelly as a ceiling and bet a small fraction of it. The full explanation, including what a proven edge looks like, is in the Kelly criterion betting guide.

Laying is a slightly different problem, because the amount you can lose is the liability rather than the stake. If you lay horses, the sizing choices are covered in fixed stake versus fixed liability laying.

Where stake sizing ends and edge begins

Stake sizing keeps you in the game. It does not, on its own, make you a winner. A perfectly sized bank attached to a method with no edge just loses your money more slowly and more comfortably. The bank is the thing that buys your edge enough time to show up in the results, which is why I treat it as the foundation rather than the strategy.

The real work begins when you ask whether your method has an edge at all. The only honest test is to track every selection and measure it on actual versus expected at Betfair SP across a real run of bets, because a profitable-looking strike rate can still be losing money once price is taken into account. Size the bank to survive, then spend your energy proving the edge is genuine.

A short way to hold the two ideas together: the bank decides whether you are still betting in a year, and the edge decides whether being there is worth it. You need both, and you size the first before you chase the second.

Common stake sizing mistakes

These are the habits that quietly undo a well-built bank. Each one has an easy fix.

• Sizing from a hunch. Fix: derive the bank from your measured strike rate using the formula, every time.
• Using too small a sample. Fix: wait for at least 100 results, and more if you can, before trusting the strike rate.
• Assuming prices you cannot get. Fix: record the price you could genuinely take, such as Betfair SP.
• Raising the stake after a win streak. Fix: resize only on your scheduled review date.
• Doubling up to chase. Fix: write the point value down and bet it on every selection.
• Shrinking the bank to fit a bigger stake. Fix: keep the margin and shrink the point instead.
• Copying someone else’s bank number. Fix: their strike rate is not yours, so their bank is not yours.
• Never measuring the drawdown. Fix: add the three spreadsheet columns from the drawdowns guide to your records.

Frequently asked questions

How many points should a horse racing bank be?

Work out your expected losing run from your measured strike rate, then hold three times that many points. At a 20% strike rate the expected run is around 21, so about 60 points is a sensible floor.

What percentage of my bank should I stake on each bet?

A bank of 60 points is a stake of about 1.7% per bet, and a 200-point bank is 0.5%. A smaller percentage is gentler on a bad run. Pick the one you can sit through, and size it from your own numbers rather than copying a rule.

Should I increase my stakes when I am winning?

Not mid-run. Review on a fixed date, divide the current bank by your points, and use the new point size until the next review. That grows stakes with the bank without letting a hot streak talk you into overbetting.

What if my strike rate is below 15%?

Then your expected losing runs are long and you need a bigger bank in points, or a smaller point. The formula handles it. Do the sum with your own strike rate and size accordingly.

Does the three-times rule guarantee I will not lose my bank?

No. No rule can. It makes ruin much less likely from an ordinary bad patch, but an exceptional run is always possible, and a method with no edge will still lose money.

Is Kelly better than level stakes?

Kelly sizes by edge, but it relies on an accurate estimate of your winning chances, which is hard to know. Level points are the safer starting point. If you use Kelly, treat it as a ceiling and bet only a small fraction of it.

Do this today

Twenty minutes is enough to size your bank properly. Work through this checklist.

1. Count your results. If you have fewer than 100 selections at prices you could genuinely take, keep paper trading, using the bet tracking tool.
2. Work out your strike rate as a decimal: winners divided by bets.
3. Type =LOG(100)/-LOG(1-your strike rate) into a spreadsheet, swapping 100 for the number of bets you plan to place.
4. Multiply the answer by three. That is your bank in points.
5. Divide the money you have set aside by that number of points. That is your point size. If it is smaller than you hoped, keep it that way.
6. Write the point value down and stake it on every selection.
7. Pick a monthly review date. Do not change your stakes before then.
8. Add the drawdown columns to your records, using the drawdowns guide.
9. Put it to work on today’s racecards and track every result.

Then read the main horse racing staking plan guide to see how stakes, prices and volatility work together once your bank is sized. A sized bank and an honest record are what turn betting from a hope into a process.