This post is a follow on from my last one regarding the Prediction Cards you can find each day at Inform Racing. Just go to the Download link once logged as a subscriber. You can take a look at the previous post by clicking here.
I did come up with some proof that the higher total scores on these cards did in fact win more and that the top end just about broke even at SP odds.
I have done a fair bit more delving now and have come up with some more useful and interesting results, that almost turn upside down how we expect the horses with a clear and obvious chance to win all the time and make us money.
If you are unfamiliar with the prediction cards, you can follow this post by taking a look at today's prediction cards by clicking here.
The results from this post should be taken in context at this point, as I am only going on two months of results at the moment, all weather flat racing and jumps racing from between mid november 2025 to mid Jan 2026.
So only two months of results but in that time there still have been a decent number of races and runners to give the results some validity in these early stages.
What I hope it will do is show you how the not so obvious is often the best way to go if you want to be profitable.
Also to give you some ideas about your own betting and how to look at a horse race from a different angle. So you can use it with any type of betting that you do, and not only with what I am doing on these prediction cards.
Most horse racing betting content starts in the same place. Trying to picking winners.
This project deliberately did not.
Instead of asking “which horse is going to win?”, I asked a different question:
“What does a typical winner actually look like – and how does the market price that profile?”
That difference in mindset shaped everything that followed and is something you can start doing yourself. As we all generally use the same basic ideas when looking at a race and probably most of the time come up with the same horses that 'have the best chance'.
Just backing the obvious however is counter intuitive and will leave us backing horses at short odds and failing to make a profit.
The prediction cards I started posting a few months ago, were the start of my challenge to try and look outside the box, when it comes to finding winners and betting success.
A different way of seeing the real chance of a horse and not just taking it's form figures, age, days since run, previous distance wins, blah, blah, blah.
These cards also contain the Master speed ratings and the + or - to a horses last handicap win in the same race code on the cards. A draw and pace visual for each race, along with the jockey, the jockeys wins to runs ratio, their Actual to Expected (A/E) value and their Impact values (IV).
It was only recently that I was thinking of ditching the jockey stats but it turns out that one of these jockey values has become a base rule in what we have uncovered so far.
When you think about it, there aren't many - maybe no other websites - that provide you with the A/E and I/V values for a jockey, so this in itself is a new angle you can incorporate into your form reading.
A/E (Actual vs Expected) tells us whether a jockey's winners odds outperfom market expectation, while IV (Impact Value) measures how often a jockey wins compared to the average rider.
In both cases a score of 1 is the middle number and anything below or above is worse or better, with a score of 2 for example, meaning they are riding twice as well as expected or compared to the average jockey.
The next logical step for me will be to mark up the same values for the Trainer and see how that works out.
What I currently have is far from perfect and far from guarantees winners but it is a start and I want to continue testing and adding more useable data until we have something that works better. Something that gives a different outlook on a race that can build better profits by finding better priced winners.
What we can't get away from is that finding biggere priced winners isn't as easy as finding ones from the sharp end of the market and that betting this way must lead to longer losing runs - unfortunately.
This post explains how the prediction race cards were analysed, how two very different betting 'systems' emerged from the same framework, and why runners that look ‘worse’ on paper are often exactly where long-term profit lives.
Why we started with data and not opinion
Racing markets are very good at rewarding confidence. Horses with clean profiles, strong narratives, and reassuring form tend to attract money. The problem is that confidence is expensive.
Our approach was to strip emotion and storytelling out of the process and let the data speak first.
Each day’s prediction cards were uploaded as normal. Results were later added using Betfair Starting Price (BSP), which is critical as this removes any hindsight bias. We are not cherry picking prices or results after the event. Every profit and loss figure is settled at the same price the market offered at the off.
The methodology in simple terms
The process followed a strict order:
1. Look at winners only.
2. Identify the shape of typical winners using medians and interquartile ranges. IQRs give us a range to work within that is 25% lower and 25% higher than the median.
3. Only then test profitability.
4. Avoid adding rules unless the data clearly earns them.
Rather than guessing filters, we let distributions guide us. If most winners cluster in a certain range, that range is worth testing. If they don’t, it’s probably noise.
This approach was applied separately to flat all weather hanicaps and jump handicaps only. No non handicaps at this point as the variations in these races range from maiden and novice events to full blown grade 1 races and would give some very conflicting results.
Sticking to handicaps gives us the same basic structure of a race and crucially, we didn't assume these two codes would behave the same and they don't.
I find that, even when running queries on the system builder, all weather flat and turf flat races give some very different results and that can also be the same very often with chase and hurdle races.
Getting started.
First of all we ran a query against every race to see what stood out where the winners were concerned.
These were run against every column on the prediction card from the speed rating to the jockey impact value. Again you can look at today's prediction card by clicking here.
This gave us the initial set of outputs for both flat and jumps handicaps and non handicaps - filtering just for the winners data.
At this point there are over 1600 races - so no small test - but this did include Irish races too. Later I decided to stick to UK racing only as the data here pointed to far more reliability.
This gave us the winning percentage for all values on the cards including the category labels of:
No issues, Probably ok, Probably not and Uncertain. On the cards these score 2pts, 1 pt, -1pt and 0pts in this order.
You can download this initial file below and take at how you would identify the better values to consider. This is the initial test so it also includes non handicap races.
There are also the top lists for actual jockey and runs/wins for the jockey which can be ignored at this point, although it does give the top eight jockeys over this period for each code.
I am using Python to read all past cards and to collate this data and all results you will find on this post. Python is a general-purpose programming language known for its emphasis on code readability and simplicity.
You can open this text file in a new window by clicking here >
From here we took what stood out and made further tests, cross referencing the data against all winners and started working on flat and jumps separately.
All weather flat handicaps - where messy is good.
The all weather flat system ended up being the most counter-intuitive.
Key findings: (Note to self - "expect the opposite of what you were expecting!")
- Lower Total Scores (Less or equal 2) outperformed higher scores.
- A jockey impact value of 1.4 or bigger was meaningful§
- A mid range betting forecast between 100/30 and 9/1 was far more valuable as a range.
- BSP prices between 6 and 25 contained nearly all the value.
- Horses labelled No issues by form analysis were consistently unprofitable.
- Horses labelled Probably ok, Uncertain, or even Probably not, were highly profitable.
Just to put the Total Scores of less or equal 2 into a little context and to explain why it may not seem that strange.
A horse that rates Uncertain in four of the categories and No issues in the other. Or Uncertain in three categories and Probably ok in the otehr two. Both would result in a total score of 2.
Uncertain could be no previous runs on the going or at the course for example or maybe finished close up a couple of times at the distance but lost three or four other attempts, so Uncertain isn't such a negative, it is just saying there is no real evidence either way.
This is where we start to see how those horses with higher scores are being overbet due to their 'obvious' chances and are mostly found at shorter odds.
Out of all of the five conditions factors of going, distance, course, general and form, form seemed to give the most positive outcomes.
The Form column calculates a combination of data from the average of last 7 SP Decimal odds, average of last 7 distances beaten, average of last 7 speed ratings, number of runs since a win, number of days since a win.
Below are the results of a split test using our set rules, ran against the form column and labels.
So from 57 race dates and a total of 276 qualifiers, horses considered to have No issues regarding the factors mentioned above, lost over 50% ROI, had just a 6% strike rate and a 28.59 loss to £1 stakes at BSP from 53 qualifiers. All BSP totals include 2% commission on all winning bets.
If you look at the other three labels you get 17%, 18% and 19% strike rates, excellent ROI percentages and very decent profit of over 200pts in total.
Fot those who like to see the better form and better scores perform better overall, this is uncomfortable data – but that’s the point.
On the All-Weather, the market heavily overpays for reassurance and obvious form. Clean profiles are popular, but popularity is already baked into the price. Whilst those with slight doubts, unclear narratives, or imperfect recent form are 'punished' by their odds far more than they should be.
The edge here is contrarian by nature but value comes from runners the market would rather avoid.
To clarify the rules we currently have in place, they are the jockey impact value of 1.4 or greater, betting forecast odds of between 100/30 and 9/1, a total score of just 2 or less and a Bsp of between 6.0 and 25.0.
I haven't discounted the No issues labels from the form category from the calculations, however, this may change going forward if for example No issues in the form column continues poorly.
You don't have to follow these, or even use the prediction cards. But when you next look at a race, try and consider where the hidden winners could be found and think about where your value may lie.
Perhaps one factor you could start using is the jockey impact value as this is one of the positive values, one of the values where is a good score is profitable.
Jockey IV tells us whether a jockey tends to out perform or under perform the other riders. An IV of 1.4 means the jockey wins 40% more often than the average.
You can download a full set of results for all qualifiers from the all weather handicaps by clicking here.
It's worth noting the draw number on this sheet above, shown in brackets after the horse name. There is only one winner from a double figure stall and about thirty odd losers. Bearing in mind these are all weather tracks, so stall one has the inside rail around a turn - apart from 5f at Southwell and up to 1 mile at Newcastle - this may be something you want to bear in mind.
Jump handicaps where competence and price matter.
Jumps racing told a very different story.
Here, we found that winners were more likely to share:
- Mid-range Total Scores (2 - 5, not extremes)
- Stronger jockey impact (IV greater or equal 1.6)
- Non-negative handicap difference, so on the same mark as last handicap win or higher.
- Bigger prices (BSP 10–25)
Unlike all weather racing, jumps do reward competence. But they still punish uncertainty too heavily at the right prices.
When we split jumps handicaps further, another insight appeared:
Chase races and hurdle races had similar strike rates.
Chase races delivered far higher ROI.
The explanation is structural as chase races involve more jumping errors, pace collapses, and attrition. When things go wrong, favourites are eliminated. Competent, well-priced runners benefit disproportionately.
Hurdles are tidier, the market is better calibrated and the edge still exists, but it’s thinner
Why losing runs don't invalidate the edge.
At prices between 10 and 25 BSP, longer losing runs are inevitable and in testing, losing runs of up to 40 bets appeared – and were entirely expected.
This is not a flaw, it's the cost of operating in mispriced parts of the market.
The important point is that:
- Drawdowns were finite.
- Profits were not concentrated in a single day or race.
- Performance remained stable across weeks.
If you expect frequent winners at big prices, you are misunderstanding how value works.
You can download a full set of results for all qualifiers from the jump racing handicaps by clicking here.
The big picture
Two sets of rules emerged from the same framework here.
- All weather flat handicaps reward contrarian thinking. Messy profiles outperform clean ones.
- Jumps handicaps reward competence combined with mispricing, particularly in chases.
Neither system relies on prediction, both rely on price and this is the central lesson.
Betting is not about finding the best horse, it's about finding the wrong price. The market is excellent at spotting what looks right but much worse at pricing uncertainty correctly and this is where long term profit lives.
Summary
I reiterate that I have been working with only two months of results - but a decent enough amount of bets to give some value. Of course none of what I have written guarantees future success as the racing markets evolve and edges decay but the process matters more than the outcome.
If you start with data, respect price, and stay uncomfortable long enough, the numbers have a habit of telling you things intuition never will.
So start looking outside the box.
