Home › Betting Strategy › Betting Systems
By The Guv'nor · Last updated October 2026
Are there horse racing betting systems that work? Not the popular ones. I back-tested ten on every UK and Irish race over two seasons (25,955 races, October 2024 to September 2026) at Betfair SP (BSP). After an illustrative 5% commission, eight lost money, one broke even and one made a small profit from a sample too small to trust.
How I tested them
Look, anyone can show you a system that made money. The test is whether it made money on races it hadn't already seen. I took the horse racing betting systems UK punters get pitched most often and tested them. Here's what I did, so you can check it.
Data: Betfair's free BSP price files, UK and Irish win markets.
Period: 1 Oct 2024 to 30 Sep 2025 (12,964 races) and 1 Oct 2025 to 30 Sep 2026 (12,991 races, 119,341 runners; two 'To Be Placed' place markets that sit in Betfair's UK win files are left out).
Stakes: level 1pt on every qualifier, settled at Betfair SP (BSP). Dead heats settled as dead heats. My other pages pay dead heats in full, so a few 2025/26 returns here sit a touch lower: every favourite -1.7% (-1.6% paid in full), second favourite +0.1% (+0.2%), non-handicap favourites -1.0% (-0.9%), BSP 20.0-49.9 +6.0% (+6.2%) and Gowran Park favourites -7.6% (-6.7%).
"Favourite" means the shortest BSP in the race, which isn't always the bookmakers' SP favourite.
Returns are before commission unless I say otherwise. Where I show "after commission" I've knocked 5% off winnings as an illustration; check your own rate.
BSP makes a fair test because, in Betfair's own words, there's "no margin for profit built in". If a system can't win at BSP, I wouldn't expect it to win at bookmaker SP either, where the bookie's margin is built in.
The scorecard: horse racing betting systems that work (or don't)
Same rules, two separate seasons, side by side. Nobody selling a system will show you this.
| System | 2024/25: bets, return | 2025/26: bets, return | Both seasons, after 5% commission | Verdict |
|---|---|---|---|---|
| Back every favourite | 13,001, +0.8% | 13,048, -1.7% | -3.6% | Loser |
| Back every second favourite | 12,964, -1.5% | 12,991, +0.1% | -4.7% | Loser |
| Favourites in 2-4 runner races | 710, -3.4% | 823, -1.9% | -4.9% | Loser |
| Favourites in 16+ runner races | 682, +15.0% | 711, +3.2% | +4.9% | Small sample, unproven |
| Favourites in non-handicaps | 4,641, -0.4% | 4,668, -1.0% | -3.5% | Loser |
| Second favourites, handicaps of 12+ runners | 1,965, -3.4% | 2,053, -0.9% | -6.3% | Loser |
| Outsiders at BSP 20.0-49.9 | 23,812, -1.8% | 23,141, +6.0% | -2.9% | One good season |
| Morning favourite, backed at BSP | 12,963, +0.1% | 12,991, +2.3% | -2.2% | Small loser after commission |
| Steamers (BSP 25%+ shorter than the morning price), backed at BSP | 15,393, -3.4% | 16,516, -6.0% | -8.8% | Loser |
| Lay every odds-on favourite (return on liability) | 1,966, +2.2% | 2,112, +4.0% | +0.1% | Break-even |
Two things jump out. Almost everything sits within a few per cent of zero before commission, because the exchange market is pretty efficient (more on that in value betting in horse racing). And commission is what turns "nearly" into "losing".
Why most systems are back-fitted noise
Most systems you'll see advertised were built backwards. Someone takes a pile of past results, tries filter after filter (course, trainer, distance, days since last run, price band) and stops when the numbers look lovely. Then they sell you the filters.
With enough filters you will always find something that made money in the past, purely by luck. Results are noisy: a few hundred bets on 4/1 shots can swing your return 15 or 20 points either way with nothing clever going on.
To put a number on it, "favourites in 16+ runner races" made +9.0% before commission on 1,393 bets. The likely range from luck alone is roughly -2% to +20% (a 95% range from how much the returns bounced around), and zero is inside it. I tested ten systems, so one looking decent by chance is exactly what you'd expect.
A system that only works on the data it was built on is overfitted. There's a proper test of that further down, and if I am honest it's the most useful part of this page.
Backing favourites and second favourites
The oldest favourite betting system there is: back the favourite in every race. Favourites win about a third of the time, and I've broken that down by race type, field size and course on how often favourites win, so I won't repeat it all here.
What matters for a system is the return, not the strike rate. Backing every BSP favourite made +0.8% in 2024/25 and -1.7% in 2025/26, before commission (-1.6% on that page, which pays dead heats in full). Pooled and after 5% commission, that's -3.6% on 26,049 bets. You'd collect plenty of winners and slowly lose money.
The second favourite won about one race in five (2,661 of 12,964 in 2024/25; 2,735 of 12,991 in 2025/26) and returned -1.5% then +0.1% (+0.2% with dead heats paid in full). "The favourite is overbet, so back the second favourite" sounds clever. On exchange prices, the data doesn't back it up.
This is why I bang on about profit rather than win rate. Anyone can talk about high win rates when they tip odds-on horses all day. Odds-on favourites won 60.9% of 1,966 races in 2024/25 and 60.3% of 2,112 in 2025/26, and still lost money to back: -1.4% and -2.6% before commission.
Field-size and race-type systems
Next, the filters people bolt on to "improve" favourites.
- Small fields (2-4 runners). Favourites win more than half of these, at short prices: -3.4% and -1.9% before commission. A high strike rate, a steady loss.
- Non-handicaps. Favourites win getting on for half of these, and the market prices that in: -0.4% and -1.0%.
- Second favourites in big handicaps (12+ runners). -3.4% and -0.9%. No hidden value in "look past the jolly" either.
- Favourites in 16+ runner races. The only one in the green both seasons: +15.0% on 682 bets, then +3.2% on 711. Look at the drop, though. After commission the second season was slightly negative (-0.7%), and the whole thing sits within the range luck alone could produce.
Would I bet that big-field angle blind? No. I'd want several more seasons first, and I'd expect it to shrink further.
Following morning steamers
A steamer is a horse whose price shortens sharply as money comes in. "Follow the steamers" is one of the most popular systems going, and it's built on a real effect: smart money does move markets. The question is whether you can make it pay.
I defined a steamer as a runner whose BSP ended at least 25% shorter than its morning price on the exchange (the average matched price before 11am, from Betfair's files). Here's what that gave:
| Price taken | 2024/25 (15,393 bets) | 2025/26 (16,516 bets) | Can you actually do it? |
|---|---|---|---|
| Morning price (hindsight) | +64.9% | +67.6% | No: hindsight |
| BSP | -3.4% | -6.0% | Yes |
That first row is the one system sellers love to show you, and it's hindsight. You only know a horse steamed once the off has come and gone. Backing the same horses at BSP, the price you'd really get after spotting the move, lost money in both seasons.
A fairer version, backing at BSP any runner whose average pre-off price was 20% or more shorter than its morning price, returned -5.4% (25,796 bets) and -1.3% (26,003 bets). By the time the move is visible, the value has gone.
So the money is made by getting on before the move, which means doing the work early. More on morning prices against BSP in starting price explained.
Laying odds-on favourites
The lay system horse racing forums love: odds-on favourites get beaten about 40% of the time, so lay them all on the exchange. When you lay, you're acting as the bookie. If the horse loses you keep the backer's stake (less commission). If it wins you pay out the liability.
You lay a favourite at 1.80 (4/5) for a £10 stake. Your liability is £10 x 0.80 = £8.
Horse loses: you win £10, less 5% commission = £9.50.
Horse wins: you lose £8.
To break even you need it to lose more than 8 / (8 + 9.50) = 45.7% of the time. At 1.80 the market says it loses 44.4% of the time.
That gap is tiny, and it's the whole problem. I laid every outright BSP favourite under 2.0 (odds-on) at BSP, 1pt lay stakes:
- 2024/25: 1,966 lays, 39.1% of the favourites lost. +2.2% on liability before commission, -0.8% after.
- 2025/26: 2,112 lays, 39.7% lost. +4.0% before commission, +1.0% after.
- Both seasons after commission: +0.1% on liability. Call it break-even.
You're also risking more than you stand to win on every bet, so a bad week hurts. "Favourites get beaten" isn't an edge, because the price already knows it.
Place betting systems
The usual place betting system: back the favourite to place instead of to win, and enjoy far more winning bets. It feels better. Does it pay?
Betfair runs a separate "To Be Placed" market with its own BSP. Its download site limits heavy use, so this is a sample: every fifth day of 2025/26, giving 71 race days from 5 October 2025 to 25 September 2026 and 2,478 races. I only kept races where every runner could be matched to its place market.
| System | Bets | Placed | Return before commission | After 5% commission |
|---|---|---|---|---|
| Favourite to place | 2,478 | 63.6% | -1.4% | -3.1% |
| Odds-on favourite to place | 357 | 83.8% | -2.1% | -2.8% |
| Second favourite to place | 2,478 | 53.9% | +1.8% | -0.6% |
| Runners at win BSP 10.0-19.9, to place | 4,964 | 28.5% | +3.7% | -0.1% |
| Every runner to place | 22,904 | 29.9% | -1.1% | -4.5% |
The favourite placed nearly two times in three, but its median place price was only 1.59, so a typical winning bet paid about 59p profit per £1. It still lost. The more often a system wins, the less each win pays.
Bookmaker each-way bets are different, because the place part pays a fixed fraction of the win odds, and that's where value sometimes hides. See each-way betting explained.
Overfitting: how to test a system honestly
Here's the test I promised: find the "systems" that made the most in 2024/25, then run exactly the same rules on 2025/26.
Test 1: back the favourite at your best course
Plenty of people swear by a course where "the favourites always win". I ranked every course with 100+ races in 2024/25 (55 of them) on the return from backing every BSP favourite there, then checked the top five the following season.
| Course | 2024/25 favourite bets | 2024/25 return | 2025/26 favourite bets | 2025/26 return |
|---|---|---|---|---|
| Pontefract | 113 | +34.6% | 115 | +33.0% |
| Newton Abbot | 109 | +32.2% | 116 | +3.2% |
| Gowran Park | 135 | +29.8% | 142 | -7.6% |
| Navan | 137 | +29.0% | 144 | -11.6% |
| Goodwood | 124 | +26.0% | 132 | +20.2% |
Pooled, the top five courses went from +30.2% (618 bets) to +6.3% (649 bets). The five worst courses of 2024/25 (Naas, Worcester, Fairyhouse, Newbury and Fontwell) went from -18.1% to -5.6%. Both ends drifted back towards the middle, which is what luck does. A couple held up, but nothing in the 2024/25 numbers told you which ones would.
Test 2: let the computer go fishing
Then I did what system builders do, only on a bigger scale. I split every race by course, Flat or Jumps, handicap or not, and backed the favourite, second favourite or third favourite. That gave 765 rules. 471 of them had at least 30 bets in 2024/25.
- 218 of those 471 made a profit in 2024/25, and 105 made +20% or more.
- The top 20 averaged +71.5% (875 bets). The next season, the same 20 rules made -3.4% (892 bets).
- All 218 profitable rules together: +20.2% in 2024/25, -0.4% in 2025/26.
- Only 105 of the 218 (48%) were still in profit the next year. That's a coin toss.
Crazy right? Every one of those 218 rules looked like a winning system on 2024/25 data. Run the other way (pick the best on 2025/26, test on 2024/25) and the top 20 fell from +64.2% to +16.2%, so it's not a one-off.
Test 3: the small-sample trap
Going through 2025/26 in date order, after the first 500 races (to 12 October 2025) runners at BSP 50.0-99.9 showed +57.1% from 385 bets. That was 8 winners. By the end of the season, on 9,873 bets, the same band was at -1.6%. At big prices, one or two winners make or break the numbers.
My checklist for testing a system
- Split the data before you look. Build on one period, test on one you haven't touched.
- Count the winners, not just the bets. 300 bets on outsiders might be 10 winners. That proves nothing.
- Use a price you could really get. BSP, or an early price you could have taken at the time, never one that needs hindsight.
- Take off commission or the bookie's margin. Most "edges" are smaller than that.
- Remember how many things you tried. Test 50 ideas and a few will look brilliant by luck.
- Check each season on its own. All the money from one month or one season is a story, not an edge.
Even a sound system will go through long losing runs, so it's worth reading my staking plans tested by simulation before you put real money behind one.
What my own approach is built on (without the secret sauce)
So, what do I actually do? I was a Quantitative Researcher at a Tier 1 investment bank, and I'm essentially a statistics/maths geek. I use those skills to analyse horse racing, and it's gone well enough that most of your favourite bookies won't take a bet from me.
What every system above misses is a reason the price should be wrong. My newsletter usually has 'Banker' bets, which are the shorter-priced, more solid ones, plus outsiders from a new but proven system that picks out horses that have won or come close at a weight or mark higher than they're running at today. When I've got inside information, such as stable information passed down from the trainers of horses I own, or smart money moving in the market, that goes in too.
I won't publish my filters, but you can see how my approach to well handicapped horses works. The rest is discipline, which I've set out in my horse racing betting strategy rules:
- I bet on every horse I tip.
- Strict points staking: divide your bank into points, say 1pt = £5 on a £100 bank (more on that in what 1pt means in betting).
- No chasing, and no going all in on a feeling.
- If you're unsure about a bet, skip it. Some days there's nothing worth backing at all, and that's a stinky day for the ponies.
- I judge it on profit month to month, not win rate. I like outsiders and each-way value, because even when they place we make more profit than an odds-on shot winning.
Losers are part of it. Understanding that losers are a thing is the key to being a winner longer term. Trust the process. If you'd rather I did the legwork, that's what my tips service is for.
- The Guv'nor
Betting systems: your questions answered
Do horse racing betting systems work?
The simple, mechanical ones don't. Backing favourites, second favourites or steamers, or laying odds-on shots, came out between about -9% and break-even over two seasons after a 5% commission. Anything that lasts needs a reason the price is wrong, tested on data it wasn't built on.
What is the most successful horse racing betting system?
In my two-season test, backing BSP favourites in races of 16 or more runners did best, at +4.9% after commission. But it's 1,393 bets, the return fell from +15.0% to +3.2% in year two, and luck alone could explain it. Keep testing it; don't bet it.
Is laying favourites a good system?
Not on its own. Laying every odds-on BSP favourite returned +0.1% on liability after commission over 4,078 lays. The favourites that get beaten are already priced in, and you risk more than you win on each bet.
Are paid betting systems worth it?
Ask the seller for results on races after the system was built, at prices you could have taken, after commission. If all they've got is a back-test, you're looking at the 2024/25 column of my overfitting test. My take on judging tipsters is in are horse racing tipsters worth it.
More in this guide
- Horse racing betting strategy: 10 rules from an ex-quant
- Value betting: finding overpriced horses
- How often do favourites win?
- AI horse racing predictions: can AI pick winners?
- How to bet on horse racing
I run Bolts Up Daily. I was a Quantitative Researcher at a Tier 1 Investment Bank (essentially a statistics and maths geek), and I now use those skills to analyse UK and Irish racing, profitably enough that most of your favourite bookies won't take a bet from me. Every horse I tip, I'm betting on myself, and every bet comes with a stake in points. I own racehorses too. Look, I can't guarantee every tip is going to win (anyone that can is a fraudster), but I care about profit, not strike rate.
More from The Guv'nor · How my tips service works