Back To The Right Mark
A great way to find new angles can be to use ChatGPT or similar AI tools. The initial question can be as open ended or specific as you like. For this example I simply asked ChatGPT to “give me an example of a profitable horse racing system based on filters”.
Of course, ChatGPT output is only as good as the input. And the input for this type of question is anything relevant on the internet. That means that some of what it suggests has some what questionable profitability. The only way to see if there is an angle is to enter the suggested filters in Selection Hunter. If it clearly doesn’t have an angle then simply ask ChatGPT to move on and suggest another selection system. In this example ChatGPT called this one – “Back to Winning Handicap Mark” System. The original rules were –
Filters:
Race Type:
- Handicap races only (Class 2-6).
- 8+ runners (ensures competitive markets & each-way potential).
- Distance: 5f to 1m4f (avoiding extreme distances).
Handicap Mark & Recent Form:
- The horse is running off the same OR a lower mark than its last winning rating.
- Must have run at least three times since its last win (prevents horses being overbet after a single poor run).
- Must have finished no worse than 6th in one of its last three runs (ensures some competitiveness).
Trainer & Stable Considerations:
- Trainer has had a winner in the last 14 days (stable in form).
- Trainer has a 15%+ strike rate at today’s course.
- Trainer has a history of success with handicappers dropping back to their last winning mark.
Course & Conditions Suitability:
- The horse has previously won at today’s course OR distance.
- The horse has shown ability on today’s ground conditions (no major ground concerns).
Market & Odds Filter:
- Odds between 5.00 and 15.00 (4/1 to 14/1) at post-time (focus on value selections).
- The horse must not be a major drifter in the market (indicates stable confidence).
In this system what I did was to recreate what I could in Selection Hunter. Using these filters didn’t really show an edge. By making a couple of adjustments I created this system which shows promise.
RaceClass – 2,3,4,5 or 6
Race Runners – Greater Than 8
Furlongs – Analysis showed this works very well in sprint races. 5, 6 and 7.
Position 1,2,3 races ago greater than 1. (Not Won).
Last Time Out Official Rating Change must be less than or equal to 0. This means the current OR rating must be the same or less than the OR in the previous race. We also add an additional filter here of LTO_OR (Last Time Out Official Rating) must be greater than 0. If we don’t know the LTO_OR then it will be given a 0. By adding this filter we can avoid including horses that shouldn’t be included.
T2YCourseStrikeRate – Last 2 Year Strike for the Trainer at this course. This filter seemed to work much better than the suggested filter of ‘a winner in the last 14 days’. (Use Trainer Win Ratio Greater than 0) to recreate that if you wanted to.
Age – Further analysis showed that Horses under the age of 6 showed a much greater egde.
You will notice we have missed out this filter
- The horse is running off the same OR a lower mark than its last winning rating.
I found that switching this filter produced better results. What we have instead is a Horse that hasn’t won in its last 3 races with an Official Rating that is either staying the same or decreasing since the last race. Hence it is called Back To The Right Mark.
We can see in the results above that the system shows promise. Using odds of 4 to 15 instead of 5 to 15 increased the strike rate slightly higher.
Variations to the above –
Remove Pos. 3 Races Ago as it has minimal effect on the results. Change it to Course Winner or Distance Winner of even H2YWins > 0 (Horse Must of Won at Least once in the last 2 years). All options will reduce the number of results and potentially increase the strike rate.
Disclaimer – There is no guarantee that the future results will mirror the historical results. It is simply an indication.
55 Percent V2
This is a really simple system that uses a combination of 2 ratings. Merlin Ratings and Predicted Rank. We are looking for horses that have number 1 rank in both and aged from 3 to 6. The odds have to be between 1.5 and 2.4.
This system really only works in Non Sprint Races. We can use the Stall = 0 to ensure that this excludes all sprint races.
In Selection Hunter the setup is as follows –
This produces a decent strike of over 55% and a consistent profit to level stakes. This opens up the potential to use a staking plan, subject of course to your own individual risk threshold.
Disclaimer – There is no guarantee that the future results will mirror the historical results. It is simply an indication.
MR2 Value
For any selection system to be profitable, it must have an edge. In this system we are looking for horses where the odds at the off are greater than the perceived odds when using Merlin Rating V2.
Merlin Rating V2 doesn’t discriminate between Race Class or Types of Racing so we will always have a Rating.
This doesn’t produce a huge number of bets but enough to say there is potentially an edge.
MR2 Odds – Must be less than 6.
T2YCourseStrikeRate – Greater than 5. This means that the Trainer has a greater 5% strike rate on this course over the last 2 years.
Race Class – Handicap Races over Class 2,3 and 4.
Age – Greater than 3. We are trying to pick out Horses that have some history behind them meaning the Ratings can be relied upon.
Odds – Between 6 and 10. (Can be higher). When the odds are greater than 6 we have a value bet.
Variations on this are to increase/decrease the Trainer Strike Rate. You can also increase/decrease the MR2 Odds and the odds at the off. This will have the effect of increasing/decreasing the number of bets and adjusting the strike rate.
If you don’t mind longer losing sequences then you can also increase the odds to 20 or 25.
Disclaimer – There is no guarantee that the future results will mirror the historical results. It is simply an indication.
60 Percent Place
This system looks for an edge in the Place Market. We are looking at Larger fields in the lower Class Races where the horses are generally less predictable. To ensure we have a Horse with a good history we use a Predited Ranking of Less than or Equal to 5 and a Racing Post Rating greater than 60.
We also want horses between the age of 4 and 6 which avoids horses with less experience and horses that are perhaps racing less than they once were.
Research shows that in the Place Market where Odds are between 1.4 and 3 AND they are Favourite at the off, there is a potential value in the bet.
We can see the Places Results and Strike Rate by pressing the Analyse Query Results Button.
Variations to this system are changing the odds ranges to increase/decrease strike rate.
Disclaimer – There is no guarantee that the future results will mirror the historical results. It is simply an indication.
Triple Top Rank
This system uses the Racing Post Rating, Official Rating and Top Speed Rating to try and find consistent winners.
Additionaly we want Class to be 2,3,4 or 5.
We also want to avoid Maiden Races (Horse has never won !).
Odds – To avoid over backed favourites use odds of over 2. Favourite Ranking should be less than or equal to 5.
This system doesn’t provide a selection every day so maybe one to combine with another system of a similar strike rate.
Disclaimer – There is no guarantee that the future results will mirror the historical results. It is simply an indication.
5 Furlong Specialist
This system looks for horses that are in form at 5 Furlongs. Designed with fairly open filters in order increase selections. You can easily add additional filters to reduce the number of selections if desired.
The Horse must of won over 5 Furlongs previously. At least 1 of the last 3 runs must of been placed. Days Since Last Run should be less than 60 days but again this can be extended to increase selections. The Trainer should also have previously won at this course.
Additional Filters could be Stalls 1-5, Race Types (Not Apprentice, Amateur etc).
This produces a good number of bets with an odds range of 2 and 20.
One option to increase the strike rate slightly is to use the Place Market at odds of over 10. This increases the strike rate to over 25% but also reduces the profit.
Disclaimer – There is no guarantee that the future results will mirror the historical results. It is simply an indication.
Irish Flat
The basis of this system was based on a ChatGPT question. “Create a horse racing selection system based on filters for Irish Horse Racing”. It offered 2 systems. One for flat and one for jumps. We will look at the Flat offering here.
Step 1: Race Selection Filters
Race Type Focus
- Flat: Focus on Handicaps, Group races, and good Maiden races (avoid Claimers and low-class races).
Class of Race
- Avoid low-class handicaps (Rated 45-65 on the Flat) unless a strong angle is found.
- Focus on Premier Handicaps, Listed, and Graded races.
Field Size Considerations
- Flat: Prefer 6–14 runners (avoids unpredictable large fields).
Course Bias & Specialists
- Certain Irish tracks favor front-runners (e.g., Galway, Limerick) or strong finishers (e.g., Leopardstown, Punchestown).
- Look for horses with good records at the track.
Going Conditions (Very Important in Ireland!)
- Many Irish tracks get very testing in winter—horses proven on soft/heavy ground have an edge.
- In summer, fast-ground specialists are key.
Step 2: Horse Form Filters
Recent Form
- Horse must have finished in the top 4 in at least one of its last 3 runs.
- Ignore horses with pulled up (PU), fell (F), or refused (R) in the last two outings unless there’s a strong excuse.
Days Since Last Run
- Flat: Prefer 7–42 days (fit but not over-raced).
Trainer Form
- Trainer must have at least a 10% strike rate in the past 14 days.
- Look for small trainers hitting form, as they often offer value.
Jockey Booking
- Top jockeys matter more in Ireland (e.g., Colin Keane on the Flat, Paul Townend over Jumps).
- If a top jockey is booked for a small yard, it’s a positive sign.
Handicap Mark Movement
- Avoid horses heavily penalized by the handicapper.
- Look for horses dropped 3-7 lbs from their last few runs.
Step 3: Market
Ideal Price Range
- Flat: 3/1 to 12/1 (shorter-priced horses dominate good races).
Step 4: Statistical & Performance Metrics
Speed Ratings & Sectionals
- Irish racing relies less on speed figures, but horses with the top 3 speed ratings in a race still hold an edge.
Course and Distance Winners
- C&D winners have a strong edge, especially at unique tracks like Galway, Bellewstown, and Laytown.
On this occasion Chat GPT coughed up some great info. If we were to add every filter there would be almost zero selections. By widening the paremeters we can come up with the following –
This produces enough selections to just about say this has an angle worth following.
We can fine tune this further by narrowing the filters or by adding the H_L3_PlaceWins but then we start to lose selections and are in danger of ‘back fitting’.
Another option here is to increase the odds range and use Place Bets when the Win Odds are 10 or over. The screenshot below shows the odds range increased to 20. The number of selections almost doubles. However our profit decreases. Our edge has decreased.
By having a cut off set at 10 so that any selections with Win Odds over 10 are Placed instead, we can get our strike rate at over 25% and profit back up over £200.
Lots of options to play around with depending on your style of betting and staking plan preferences.
Disclaimer – There is no guarantee that the future results will mirror the historical results. It is simply an indication.
