Most people asking how to find horse racing winners start by asking which rating to use. Which system, which factor, which number wins the most. The more useful question comes one step earlier: which kind of race am I looking at, and which rating works best there?
By the end of this guide you will be able to run a simple four-step process on any card. You will choose a slice of racing to focus on, test which ratings earn their place inside it, combine the best ones into a short contender list, and decide whether the price is worth taking. Each step does one job, and the whole thing is built around how races are actually run.
How do you find horse racing winners? Pick a tight set of race conditions, measure which ratings work inside them using the A/E ratio (actual winners divided by the winners the odds implied, where 1.00 means the market has it about right), combine the ones that stand out, then choose bets from the result. In an earlier Race Advisor analysis of over 1.5 million UK runners, the strike rate of the same rating moved from 28% to 49% depending on the conditions it was applied to. Conditions first, ratings second.
In this guide
- Step 1: Which race conditions should you pick first?
- Step 2: How do you measure a rating inside those conditions?
- Step 3: How do you combine the ratings that survive?
- Step 4: How do you turn the score into bets?
- A race-day routine you can run in minutes
- Why does the order matter?
- Common mistakes and how to avoid them
- How do you know the method is working?
- Frequently asked questions
- Do this today: a short checklist
Step 1: Which race conditions should you pick first?
Most people open a racecard and ask which horse their rating likes. A stronger start is asking whether this is the kind of race the rating suits.
In that earlier analysis, Race Advisor’s PR Rank 1 (the top horse on our proprietary power rating) won 32.3% of races across over 159,000 runners, with an A/E to Betfair SP (the exchange’s closing price) of 0.99. That is a healthy, fairly priced rating. Slice it by race type and the strike rate shifts:
- Flat Turf: 30.3% win rate across over 58,000 runners.
- Flat All Weather: 31.7% across over 36,000 runners.
- Chase Turf: 31.9% across over 22,000 runners.
- Hurdle Turf: 35.4% across over 35,000 runners.
- NH Flat: 36.6% across over 5,700 runners.
The A/E stays between 0.98 and 1.00 throughout, so the market prices the rating consistently. The strike rate is what changes, by six percentage points between Flat Turf and NH Flat. Same rating, different conditions. Those figures describe the past, so read them as a map of where the rating has been strongest.
Class and race type move it too. PR Rank 1 won 28.6% of Class 6 races and 35.0% of Class 1 races. The largest gap of all is between handicaps and non-handicaps, where it won 26.6% of the UK handicaps (over 98,000 runners) and 41.4% of non-handicaps (over 61,000 runners). In handicaps the weights are set to bring the field together, so the top-rated horse has less to spare.
How to choose your slice
Choose a slice you will actually bet. It should be narrow enough that ratings behave consistently, and wide enough to give you plenty of races. All Weather handicap sprints up to six furlongs are a slice. Class 4 to 5 Flat Turf handicaps from seven furlongs to a mile are another. National Hunt Flat is a slice on its own.
A simple rule for picking: choose one code (Flat or National Hunt), one surface or ground type, one broad class band and one distance band. That is four decisions, and together they define a slice. If a slice gives you only a handful of races a week, widen one of the four. If it gives you more races than you can study, narrow one.
If you are unsure where to begin, start with the slice you already watch most. Familiarity helps you spot when something looks wrong. Our guide to fast winner scoring without overthinking shows how a narrow focus keeps the workload manageable.
Step 2: How do you measure a rating inside those conditions?
Once you have your conditions, the next question is which ratings deserve a place. The measure to use is the A/E ratio. An A/E above 1.05 on a meaningful sample points to a rating worth following. Around 1.00 says the rating matches the market.
To measure a rating, split it into buckets. For ranked ratings the buckets are simple: rank 1, rank 2, rank 3 and so on. For raw ratings, split the population into four or five roughly equal-sized groups so each has a decent sample.
Across the full UK dataset, the PR Rank gradient looks like this:
- Rank 1: 32.3% win rate, A/E 0.99, over 159,000 runners.
- Rank 2: 19.6%, A/E 0.99, over 160,000 runners.
- Rank 3: 13.8%, A/E 1.01, over 160,000 runners.
- Rank 4: 10.1%, A/E 1.01, over 157,000 runners.
- Rank 5: 7.8%, A/E 1.02, over 151,000 runners.
- Rank 6 and below: 3.9%, A/E 1.00, over 726,000 runners.
That is a well-behaved rating. The win rate falls steadily from 33% at the top to 4% at the bottom, and the A/E stays between 0.99 and 1.02 all the way down. A rating where the win rates barely separate across the buckets is telling you little beyond the price, so set it aside. Keep the ones where the win rate moves cleanly through the gradient, or where one bucket lifts above 1.05 on a good sample.
How to work out an A/E ratio
The calculation is short. For every runner in a bucket, take the price and convert it to the chance it implies (a price of 4.0 implies 25%). Add those chances up to get the number of winners the market expected. Then divide the winners that actually happened by that expected figure.
Worked example (illustrative, not real results): a bucket holds 200 runners, and their prices imply 40 winners. If 44 of them won, the A/E is 44 divided by 40, which is 1.10. If 36 won, it is 0.90. Above 1.00 means the bucket won more often than its prices suggested. Below 1.00 means it won less.
Check the sample before you trust the number
Small buckets produce wild A/E figures. A bucket of a few dozen runners can show 1.30 or 0.70 through luck alone. Before you trust any A/E, ask how many runners sit behind it and whether the result holds across different seasons. If it only works in one year, treat it as a lead to investigate, not a rule. Be most suspicious of results that look too good.
Step 3: How do you combine the ratings that survive?
When two ratings that measure different things point at the same horse, the shortlist gets shorter and the strike rate rises. The examples below show how that looks.
Race Advisor’s two flagship ratings are PR Rank, weighted toward fitness for today’s conditions, and VDW Rank, a separate method weighted toward consistency and class. They look at the race in different ways, so when they agree you have two separate views pointing at the same horse. In Hurdle Turf non-handicaps, the combined signal looked like this:
- PR Rank 1 only: 45.6% win rate, A/E 1.00, over 16,800 runners.
- VDW Rank 1 only: 32.3%, A/E 0.98, over 16,800 runners.
- PR Rank 1 and VDW Rank 1 (the same horse): 49.3%, A/E 0.99, over 7,800 runners.
When the two agree, the strike rate climbs to nearly one race in two, and the contender list gets much shorter. The same pattern holds in tighter slices. In Flat All Weather Class 5 handicaps, PR Rank 1 alone won 28.9% (A/E 1.01) across over 6,300 runners. With VDW Rank 1 agreeing, it won 31.9% (A/E 1.01) across over 2,700 runners.
What agreement gives you is a smaller pool with higher confidence, which saves time on a busy card. To rank your contenders, multiply each horse’s bucket A/E values together. A horse in a PR Rank 1 bucket worth 1.01 and a VDW Rank 1 bucket worth 1.00 scores 1.01. Do that for each horse and the highest score is your strongest contender. It is a way of ordering the horses, not a price. Five or six ratings that measure different things will serve you better than ten near-duplicates.
Worked example: scoring three horses
Here is the multiplication on a small scale. The numbers are illustrative, not real results.
| Horse | Rating A bucket A/E | Rating B bucket A/E | Combined score |
|---|---|---|---|
| Horse 1 | 1.04 | 1.02 | 1.06 |
| Horse 2 | 1.00 | 1.08 | 1.08 |
| Horse 3 | 0.96 | 1.01 | 0.97 |
Horse 2 ranks first on the combined score, even though Horse 1 looks better on Rating A. Horse 3 sits below 1.00, so it drops down your list. This is a ranking tool. It tells you where to look first, not what price to accept.
Pick ratings that measure different things
Choose ratings that look at different parts of the picture, such as an overall power rating, a class or consistency measure, and a pace or fitness measure. Two ratings built from the same raw inputs will agree for the wrong reason, and you gain nothing.
Step 4: How do you turn the score into bets?
If the first three steps are done carefully, this one is straightforward. You have a ranked list of contenders within your chosen conditions.
You can use the list directly: the top contender to win, the top two for a dutch (splitting your stake so each returns about the same if either wins), or the top three for a place play. If you like to think in prices, use the ranking as one input to your own price for each horse and compare that with the available Betfair SP. Where your own price says a horse should be 4.0 and the market offers 6.0, you have a candidate worth a closer look.
Not every race needs a bet. The score ranks every runner, but only some races will give you a contender at a workable price. If your top contender is 2.0 in the market and your own price says 2.2, there is little to gain, so move on to the next race. Choosing your races is one of the easiest advantages to use.
A simple decision rule for the price
Before every bet, ask one question: is the price on offer bigger than the price I think the horse should be? If yes, the bet is a candidate. If no, leave it. Write your own price down before you look at the market, so the market cannot talk you into it.
Keep your stake the same size for each qualifying bet while you are learning the method. Our guide to betting on horse racing simply covers a straightforward way to structure your bets around a short list of runners.
A race-day routine you can run in minutes
The four steps describe the thinking. On the day, you want something quick. Here is a routine that applies the method to your chosen slice without turning into hours of work.
- Open the card and filter to your slice. Ignore every race outside it.
- For each race in the slice, read the top contenders on your chosen ratings. Note where two ratings agree.
- Write your own price for the top one or two contenders.
- Compare with the market. Keep only the races where the price is bigger than yours.
- Place the bets and record them straight away.
If you want a quicker way to sort a race, our guide on analysing a race quickly uses the same logic in a five-point check. For finding the runners that matter most, see how to find the most competitive runners in a race. And if you are short on time, the two-minute fast selections method shows how far a short checklist can take you.
Once you are comfortable, you can widen the view. Our guide to building the 360 picture of a race shows how to fold in extra factors without drowning in detail.
Why does the order matter?
Most “best rating” answers go wrong by running the steps in reverse. Someone finds a rating, enjoys its headline strike rate, and applies it to every race. The strike rate then swings between conditions and it is hard to tell why.
The four-step order keeps it tidy. Conditions come first, because they decide what the rating is being asked to do. Measurement comes second, because A/E shows whether the rating earns its place. Combination comes third, because two independent confirmations beat one strong opinion. Bets come last, once there is a price worth taking.
It is also the thinking behind our guide to analysing a race quickly. The day-to-day routine there (cut the field by price, look for rating agreement, check Race Predictor, check days since last run, read the pace) is this method applied to your chosen slice.
Common mistakes and how to avoid them
- Starting with the rating. Start with the slice of racing you will bet. Then test the rating inside it.
- Judging by strike rate alone. A high strike rate can come with short prices. Check the A/E so you know whether the price paid for it.
- Using tiny samples. Wait until a bucket holds plenty of runners before you trust it, and check that it holds across more than one season.
- Stacking similar ratings. Pick ratings that measure different things, otherwise agreement means nothing.
- Betting every race. Pass on races where the price is not bigger than your own.
- Changing the rules after a bad week. Decide your slice and your rules in advance, and review them on a set schedule instead.
How do you know the method is working?
A repeatable process still has losing runs, so judge it over a meaningful run of bets rather than one bad day. Keeping a record of your results shows you which slices work best for you. The bet tracking tool is a simple way to log each bet, its price and the result.
At the end of each month, look at three things. First, how many bets did you place in each slice? Second, what was the average price you took against your own price? Third, did you follow your rules on every bet? Honest answers to those three questions tell you more than a single profit figure, because they show whether the process is being followed. Results will vary, and no method removes the risk of loss, so only stake money you can afford to lose.
Frequently asked questions
What is the best rating for finding winners?
There is no single best rating. The strike rate of the same rating moves a long way depending on the race conditions, so the better question is which rating works best in the slice you bet.
What is a good A/E ratio?
An A/E above 1.05 on a meaningful sample points to a rating worth following. Around 1.00 means the rating matches the market. Always check how many runners sit behind the figure.
How many ratings should I combine?
Five or six ratings that measure different things will serve you better than ten near-duplicates. Start with two, check that they add something, and build up slowly.
Do I need to bet on every race in my slice?
No. Bet only where the market price is bigger than the price you think the horse should be. Passing on a race is a valid decision.
Can this method guarantee a profit?
No. The method helps you make clearer, more consistent decisions, but every bet carries risk and losing runs are normal.
Do this today: a short checklist
- Choose one slice of racing: one code, one surface or ground type, one class band and one distance band.
- Pick two or three ratings that measure different things.
- Note the A/E for each rating inside your slice, and how many runners sit behind it.
- Keep the ratings that separate cleanly or lift above 1.05 on a good sample.
- Score today’s runners in your slice by multiplying the bucket A/E values together.
- Write your own price for the top one or two contenders before you look at the market.
- Bet only where the market price is bigger than yours, and pass on the rest.
- Record every bet so you can review the slice after a meaningful run.
To take the shortcut, open today’s race cards. PR Rank, VDW Rank and the Race Predictor (our model’s win chance for each horse) are already on every UK race, so you can choose the conditions you trust and read the contender list inside them.
