An all weather horse racing system only works if its filters survive contact with real data. Most AW guides repeat the same ten or twelve rules that have been floating around since the late noughties. Some of those rules still hold up. Some have probably been priced in by the exchange market. And one or two were never really true to begin with.
By the end of this guide you will be able to turn ten filters into a written all weather system, run it on paper, record it properly and decide whether it deserves small live stakes, needs more evidence, or should be dropped. Each step is short and you can start today.
Some of the filters come with numbers from our data. Others are sensible ideas for you to test yourself, and each one says which it is. In an earlier Race Advisor analysis of more than 32,000 UK all weather races between 2013 and 2026, a few things stood out clearly, and they shape what follows.
In short: all weather (AW) racing has smaller fields than turf (9.9 runners on average against 11.0), favourites that are priced about right, and a handful of angles around distance form and trainer and jockey form that are worth testing. The blanket “lay the favourite” rule, which used to be the headline AW strategy, did not come out ahead at Betfair SP once commission is taken off.
Contents
- What an all weather horse racing system actually needs
- The ten filters to test
- How to build the system from the filters
- How to record your selections
- How to test the system
- Go live, keep testing or drop it
- Common mistakes to avoid
- Frequently asked questions
- Do this today
What an All Weather Horse Racing System Actually Needs
AW racing is not turf with floodlights. The six UK courses (Wolverhampton, Kempton, Lingfield, Newcastle, Chelmsford and Southwell) use different artificial surfaces, such as Polytrack and Tapeta, and form does not always transfer cleanly between them. Fields are smaller. And the same horses turn up repeatedly under the lights, so there is plenty of form to study.
An AW system that holds up needs filters that respect those realities. Below are ten, with the numbers where we have them. Betfair SP means Betfair Starting Price, the exchange’s closing price, and a level stake means the same amount on every selection.
A system, in this guide, is simply a written set of rules that picks the same kind of horse the same way every time. If you cannot write the rules on one page, you do not have a system yet, you have a hunch. The point of writing it down is that you can then test it, and a tested rule is worth far more than a remembered one. For the wider market picture, see all weather horse racing statistics.
The Ten Filters to Test
Read these as ingredients, not as a recipe. The next chapter shows you how to combine them.
1. Use AW Form, Not Turf Form
Turf form often tells you little about what a horse will do on AW. Even within AW, form may not carry over between surfaces or courses. A sensible starting point is to weight a horse’s last three runs on the type of surface in front of you more heavily than anything else.
This is filter one because it is easy to ignore. A horse arriving with two strong turf runs at, say, Goodwood can look attractive on a Wolverhampton card. Make AW form the first cut, not an afterthought, and test whether it works for you. The same caution runs the other way: does all-weather form transfer to turf looks at how far AW form travels once a horse switches to grass.
2. Stop Laying AW Favourites Blind
The old AW playbook said to lay favourites on artificial surfaces. In an earlier analysis covering 2013 to 2026, backing every UK AW favourite at bookmaker SP at level stakes lost 7.9% across 35,333 selections. That is the number that gave the rule its reputation.
At Betfair SP, with the standard 2% commission applied, the same favourites lost 1.4% across 32,972 bets, with a win rate of 34.1%. Their A/E was 1.00. A/E is the ratio of actual winners to the number the prices implied, so 1.00 means the market priced them about right. When backing the favourites loses little and the favourites win as often as the prices say, laying them blind leaves no obvious gap to exploit.
These figures describe that stretch of racing and will not repeat exactly.
For comparison, across Great Britain and Ireland, all codes, from 1 July 2013 to 14 October 2026, the Betfair SP favourite won 34.4% of 167,257 races, with an A/E of 1.00 and a return of -1.4% at level stakes after 2% commission. The all weather figures look very similar.
The lesson is broader than AW favourites. Any system built on the idea that the market consistently overprices a popular angle should be tested at Betfair SP after commission before any money is staked. Many “lay the obvious” systems struggle there. There is more on this in all weather favourites.
3. Filter on Trainer Form to Narrow the Field
Hot trainers do win more often on AW. In the same earlier analysis, runners from a trainer with a 14-day win rate above 15% won 14.1% of AW races, compared with 8.5% for a trainer who had not had a winner in the previous fortnight. That is a gap of 5.6 percentage points, across 87,698 in-form runners and 118,312 cold-trainer runners.
A higher strike rate (the percentage of runners that win) is not the same as value, though, because runners from in-form yards may well go off at shorter prices. So use trainer form to narrow a field and lift your strike rate, then check the price before you back anything.
4. Use Jockey Form the Same Way
Jockeys showed a similar pattern. Riders whose winners-to-rides percentage over the previous 14 days was above 15% won 13.5% of their AW mounts. Riders who had not ridden a winner in that fortnight won 8.4%. The samples were 84,726 and 77,943 runners respectively.
Pairing an in-form jockey with an in-form trainer may lift your strike rate further. Treat it as a way to shorten your shortlist, not as a reason to back a horse at any price.
5. Favour Distance Winners Over Course Winners
This one is an idea to test, not a rule. The thinking is that a horse that has already shown it can run the trip is a better guide than one that merely likes the track. Many racecards flag “course winner” and “distance winner” separately, and it is worth checking in your own records which label has been the more useful for you.
If you have to pick one, many people take the horse that has already proved it can run the trip. Keep a note of how each does for a few months before you lean on it.
6. Be Careful With Sex Rules
You will hear plenty of folklore about fillies and mares on the all weather, from “lay the fillies” to sweeping claims that female horses underperform. Treat these with care. Broad rules about sex tend to be too blunt to bet on.
If the idea appeals, test it on your own records. Look at short-priced fillies and mid-priced mares separately, over several courses and several seasons, and see whether anything holds up before you change how you bet.
7. Read Pace by Field Size
Pace shape matters in small AW fields. A horse that suits a slow pace, or has a clear front-running profile in a small field, can be well placed on tight tracks. This is a judgement for you to make from the replays and the form, and it works best as a tiebreaker between two otherwise comparable horses rather than a primary signal.
Try writing down, for a couple of dozen races, which horse you expect to lead and whether it made a difference. That gives you your own evidence.
8. Use Smaller Fields as a Structural Advantage
In the earlier analysis, the average UK AW field between 2013 and 2026 was 9.9 runners, compared with 11.0 on turf. About 56% of UK AW runners came from fields of 11 or fewer. That gives you fewer runners to assess in each race.
In that work, the horse ranked first by Betfair SP (the market favourite) won 34.1% of races, the second-ranked won 20.4%, then 14.1%, 10.0% and 7.6% for third to fifth. The first three in the market account for roughly two-thirds of winners, so a shortlist of two or three horses per race covers a lot of ground.
One idea to test is to restrict your AW play to fields of, say, 12 or fewer. Skip the 16-runner sprints on a Friday night, and see whether the cleaner races suit you better.
9. Refuse to Be Drawn Into Surface Mythology
Polytrack, Tapeta and the old Fibresand each have their own folklore. Tapeta favours stamina. Fibresand rewards front-runners. Polytrack rides fast. Some of that may be true. A lot of it is repeated without being checked.
We have not split our figures by surface type, so there are no hard numbers here. The honest approach is to treat each course as its own profile rather than ascribing an edge to a surface name. A course is also a cleaner filter than a surface label, because drainage, weather and the regular trainers who target it all differ from track to track.
10. Build the Filter Stack, Not the Trigger
Every filter above is a narrowing step, not a betting signal. A serious AW system uses them in sequence: cut to AW form only, restrict to smaller fields, keep the in-form trainer and jockey combinations, take the distance winner over the course winner, weight the pace fit, then ask whether the surviving horse is actually a value bet at Betfair SP.
The mistake is to use any single filter as a buy trigger. None of them stands alone. The aim is a shortlist of a few horses per race. Treat the stack as a workflow to try out rather than a proven method. The next chapter shows how.
How to Build the System From the Filters
Building the system means turning ten loose ideas into one written rule set. Do it in this order. It takes about an hour the first time.
Step 1: Choose a small number of filters
Do not use all ten. Pick three or four. A system with eight rules will match so few horses that you will never collect enough results to judge it, and every extra rule gives you another chance to fit the rules to past luck. Our wider write-up on why betting systems fail through backfitting explains the trap in detail.
A sensible first stack, to be tested and not assumed to work, is: AW form first (filter 1), fields of 12 or fewer (filter 8), an in-form trainer (filter 3) and the distance winner label (filter 5). Filters 4, 6, 7 and 9 can wait until the base version has a track record.
Step 2: Split filters into cuts and tiebreaks
A cut removes a horse from consideration. A tiebreak only chooses between horses that survive the cuts. Making this split stops you drifting into judging each race on a whim.
| Filter | Role in the system | Why |
|---|---|---|
| 1. AW form | Cut | Easy to check, removes horses with no relevant surface evidence |
| 8. Field size (12 or fewer) | Cut | Easy to check and keeps the races you assess manageable |
| 3. Trainer in form | Cut | Narrows the shortlist and lifts the strike rate |
| 5. Distance winner | Tiebreak | An idea to test, so it should not remove horses on day one |
| 7. Pace | Tiebreak | Needs judgement, so keep it as a late step |
Step 3: Write the rules as yes or no questions
Each rule must have a clear answer that anyone could check from the racecard. “Is the trainer in form?” is vague. “Has the trainer won with 15% or more of runners in the last 14 days?” is clear. You can borrow the 15% line from filter 3 or set your own, but write it down before you look at any results, and do not move it afterwards.
Here is an illustrative rule sheet. The wording is a template, not a recommendation of these exact lines.
- The race is on a UK all weather course.
- The field has 12 runners or fewer.
- The horse has run on the all weather within its last three starts.
- The trainer’s 14-day win rate meets your written threshold.
- If more than one horse passes, take the one that has already won over the trip. If still tied, take the one whose pace suits the race.
- Only back it if the Betfair SP is at or above your written minimum price (see the next step).
Step 4: Add a price rule
A shortlist is not a bet. The last rule is a price check. Work out the shortest odds at which you would still take the horse, then skip it below that price. Our guide to working out the minimum odds you should take walks through the method, and how to calculate your edge shows how to judge whether the price is worth taking.
Using Betfair SP as the reference price keeps the test clean. It is a price anyone can look up afterwards, so you cannot flatter your results by remembering the best price you saw.
Step 5: Decide your stake before you start
Use a level stake for the test. One unit on every selection keeps the maths simple and stops a hot streak from hiding a weak rule. If you want to think about real stake size later, start with horse racing stake sizing, but not until the test says the system deserves it.
How to Record Your Selections
A test is only as good as its records. Use a spreadsheet or a bet tracker. The bet tracking tool does the arithmetic for you, and how to track bets the right way explains what to log.
Record these columns for every selection:
- Date, course, race time and field size.
- Horse, trainer and jockey.
- Which filters it passed (a simple yes or no for each).
- The Betfair SP and the result.
- Profit or loss in units after 2% commission on winnings.
- A short note on why you took or skipped it.
Record the selection before the race, not after. Take a screenshot or save the list of selections each morning. Every skipped horse counts too: if the rules said back it and you chose not to, log it as a selection anyway. Otherwise you are testing your mood, not your system.
Be honest about what you could really have taken. If the horse was only available at a price you could not get, the record should reflect that. Betfair SP is the cleanest reference because it is public and checkable.
How to Test the System
Paper trade first
Paper trading means recording bets as if you had placed them, without staking any money. It is free, and it shows you whether the rules give a sensible number of selections each week. If your stack produces two selections a month, it is too tight to test. If it produces forty a day, it is too loose to follow. Adjust the number of filters (not the thresholds) until the volume feels workable, then lock the rules.
Collect enough selections
Strike rates mean little on small samples. A handful of winners can come from luck alone, and even a sound approach has long losing runs. Our test guide, how to test a horse racing system before you bet it live, suggests recording at least 100 selections at the price you could really have taken, and treating any answer under 1,000 bets as a hint, not proof. That is a sensible floor here too.
Work out the chance it is just luck
Turn each bet into a percentage profit or loss, using Betfair SP with 2% commission taken off winnings. Average them, work out the standard error, and read off the estimated chance that the true return is below zero. The linked guide has the four steps and the Excel formula. Its rule of thumb is that under 10% is a starting point for small live stakes, and over 10% means you do not have enough evidence yet. That is a rule of thumb, not a proven cut-off.
Here is a worked example with invented numbers, labelled illustrative, not real results. Suppose a test of 120 selections shows an average return of +4% per bet with a standard error of 5 percentage points. The average is under one standard error from zero, so the chance the true return is below zero is high, well over 10%. The honest conclusion is “not proven”, even though the profit looks pleasant. Now suppose the same average came from 600 selections, so the standard error fell. The same +4% would then be much stronger evidence. Sample size does the work, not the headline return.
Test one version, then freeze it
If you change a rule halfway through, you have started a new test. Keep a version number on your rule sheet (v1, v2) and restart the count whenever you change anything. Our post on why you should stop tweaking your system explains why constant adjustment erases the evidence you are trying to build.
Check the split
Once you have a decent sample, split the records into halves, for example first half and second half, or by course. If the system only works in one half or at one course, treat it with suspicion. A real edge should show up, even if faintly, in most slices. A result that depends on a few big-priced winners is fragile, so look at the results with and without the biggest winner.
Go Live, Keep Testing or Drop It
Use a simple decision rule agreed in advance. One decisive recommendation: only go live when you have at least 100 recorded selections and the estimated chance of a negative return is under 10%, and even then start with small stakes.
| What the test shows | What to do |
|---|---|
| Fewer than 100 selections | Keep paper trading. Do not stake money. |
| 100 or more, chance of negative return above 10% | Keep collecting. The result could easily be luck. |
| 100 or more, chance of negative return under 10% | Try small live stakes, and keep recording exactly as before. |
| Clearly negative after a large sample | Drop the system, or write a new version and restart the count. |
When you go live, keep the record going and compare live results with the paper test. Live results are usually a little worse than paper, because prices move and discipline slips. A gap is normal. A big gap is a warning. Plan for losing runs before they happen: horse racing drawdowns and bankroll sizing shows how to work out how deep a bad patch might go.
No test can promise future profit. Racing changes, and a system that worked over one stretch can fade. Review your records every month and be willing to stop.
Common Mistakes to Avoid
- Using every filter at once. Pick three or four. More rules mean fewer selections and a higher chance of fitting past luck.
- Treating a filter as a buy signal. A filter narrows the list. Only the price check tells you whether to bet.
- Judging on bookmaker prices. Test at Betfair SP after commission so the numbers can be checked by anyone.
- Moving thresholds after seeing results. Write the rules first, then freeze them.
- Stopping the test at a lucky high. Decide the sample size in advance and stick to it.
- Skipping losing selections. Log every selection the rules produce, winners and losers alike.
- Assuming turf knowledge carries over. AW racing has its own rhythms, so keep AW form at the front of the process.
- Trusting surface folklore. Check any claim about Polytrack or Tapeta against your own records before you use it.
Frequently Asked Questions
Does an all weather horse racing system work?
It can only be judged by testing. The earlier analysis in this guide found that all weather favourites were priced about right (A/E of 1.00), so a system needs more than the favourite label. Test your own rules on paper first and judge them on Betfair SP after commission.
How many filters should I use?
Start with three or four. Each extra filter shrinks your sample and adds another chance of fitting the rules to past results.
How many selections do I need before I trust the result?
Aim for at least 100 recorded selections as a starting point, and treat anything under 1,000 as a hint, not proof. These are rules of thumb, not guarantees.
Should I back or lay on the all weather?
This guide tests backing selections. The old blanket rule of laying favourites did not come out ahead at Betfair SP after commission in the earlier analysis, so do not lay on that idea alone.
Is in-form trainer data worth using?
It can help shorten your shortlist, because in-form trainers won more often in the earlier analysis. It does not prove value, so always check the price.
Can I use the same system on turf?
Not without testing it. AW and turf behave differently, and AW form does not always carry over to grass.
Do This Today
- Choose three or four filters from the ten and write them as yes or no rules.
- Mark each one as a cut or a tiebreak.
- Write your minimum price rule and the level stake you will use for the test.
- Set up your record sheet or open the bet tracking tool.
- Open today’s race cards, find the all weather meetings and see which races survive your first cuts.
- Log every selection before the race, with the Betfair SP and result afterwards.
- Set a date to review, once you reach 100 selections, and stick to it.
Take it one step at a time and let the records decide. That is how a list of filters turns into a system you can trust, or one you can safely drop.
