SNOWFALL (Frankie Dettori) wins The Cazoo Oaks Epsom 4 Jun 2021 - Pic Steven Cargill / Racingfotos.com

Racing Systems: Flat Season Angles

With the turf flat season about to start, I thought now was a good time to start looking in detail at some potential flat racing systems, writes Dave Renham. As with all articles in this series I am going to look over the long term, studying UK data from Jan 1st 2009 to Dec 31st 2021 with all profits quoted to Betfair SP.

When sharing these racing systems I am not advocating that you use them 'as is' necessarily, but my plan is to give you all the facts so you can make an informed choice. As we know systems are not for everyone, but for many readers there will be something to take out of this research. See what you think...

The "Good Newmarket debut" system

For years I have noticed that horses that had their first ever flat run at Newmarket are worth a close look in their next race. Indeed, backing all Newmarket debutants in their next race would have yielded a BSP profit of £801.39 to £1 level stakes. This equates to a return of more than 17 pence in the £. All looks fine and dandy! However, digging deeper we can see that in 2010 a horse won at odds of 814.82, which even after 2% commission returned a profit of around £798 for a £1 bet. Take that winner out and profits end up at just £30 for the 13 seasons. Not quite so exciting!

With one winner effectively wiping out all the profits, it is not a system we can confidently use in its raw form, but it is still impressive to note how well this simple idea has done. Perhaps we can implement a sensible and logical system rule or two to it, and find a potential system. Clearly we need to eliminate that big priced winner and the most logical way to go is to stick to horses that impressed on their debut run at Newmarket. My first thought, then, was to look only at horses that had finished in the first three on their debut run. My system now reads:

  1. Flat racing (Turf / All weather)
  2. Second career start having made debut at Newmarket
  3. Finished in the first three on debut

Here are the results of this system (2009-2021):

 

 

The baseline figures look good – returns pushing 17p in the £ is the same as the raw system produced. The strike rate is really good as well, with horses winning at a rate of better than one race in every three. As I have mentioned before, we need to look for consistency and one of the best ways of doing this is to study the yearly results. The BSP returns are plotted yearly in the graph below:

 

 

There have been ten winning years out of 13, and the losses in 2012, 2013 and 2016 were not bank breaking in the least. What is pleasing is that the results match the system logic, this logic being that races at Newmarket which involve debutants are usually well above average in terms of standard for the specific race type. Maidens in particular see most of the top stables sending out what they perceive to be their best unraced horses. Hence, horses that finish in the first three on their Newmarket debut are not only well thought of pre-race, but they have produced the goods early on the track too.

More positive news is that the overall results are not skewed by any huge priced winners and this is illustrated if we implement a maximum price rule. For a price rule it is easier to use traditional SP and if we limit selections to horses that had an SP of 12/1 or less we get the following overall figures:

 

 

These results show that the vast majority of qualifiers started at 12/1 or less and, ultimately, this backs up the fact that the system has been consistent and not skewed by big priced winners. We still see the same ten winning years out of 13, and the losing years now produce smaller losses in each of the three years.

To me, this looks a playable system and one you might consider utilising in some way, be it following the rules blindly if that way inclined, or looking at system qualifiers and making a judgement call based on your own form/race reading.

The "Fit and in form" system

Going back to the 80's and early 90's there were successful systems based on horses that returned to the track quickly. In those days, with little or no computer data to crunch, punters and most of the betting fraternity were unaware that horses were often inclined to run well after a very short break. I suppose the misconception was that such horses would need more time to recover. However if we look at the strike rates of all runners in all flat races we can see that when we break down the data, the more recent the last run the better:

 

 

Horses returning to the track within four days have a far superior strike rate to any other group, scoring 15.4% of the time. Nowadays, the punters are aware of this quick returning bias, which is why blind profits cannot be made any more. Having said that, even now, backing all such runners returning to the track within four days would have lost you less than 1p for every £ bet – so almost a break even scenario. That is from nearly 18,000 runners!

So the system I want to share is based around quick returners, focusing on horses that appear in form. The system rules read:

  1. Flat racing (Turf / All weather)
  2. Days since last run 1 to 4 days
  3. Finished in the first five in both of their last two starts.

Note, for the last rule I could have chosen finishing in the first three in their last two runs, or I could have chosen a top six finish. In fact, I decided on top five before checking the results. For the record, the results would have even better if I had included a top 6 finish in their last two runs, but this is simply back-fitting and the increase in profitability would very likely be down to luck. Similarly, I could have chosen top three – they also made a profit when I checked and with a much better strike rate. However, although there is some logic in picking slightly better performances on their last two starts this again would potentially be back-fitting.

This angle's strike rate has fluctuated a little, as one might expect, ranging from a low of 10.4% in 2019 to a high of 23.7% in 2020. However, 2019 was the only year the strike rate dipped below 14.5%, and in eight years the strike rate has exceeded 20%. The overall bottom line reads as follows:

 

 

That's a good looking strike rate which most punters would be comfortable with and sound profits, too, showing a 7p in the £ return. Let us now look at the yearly breakdown by BSP Returns (ROI%):

 

 

There have been nine winning years and four losing ones with the worst year actually being last year. That said, losses were less than 10% so it wasn’t a disaster. In terms of big priced winners, there were a couple but almost all price brackets showed a profit which is comforting.

Is this one a system I would personally use? Not in its raw state, no, as it averages over 500 bets a year which is too many for my style. However, I do often look for quick returners, so this system for me is a case of note the qualifiers and do some further digging.

The "3yos versus older horses in long distance races" system

Not the catchiest title in the world for this one! However, I have long noted that 3yos seem to perform well in longer distance races when racing against older horses. So here is the system.

  1. Flat racing (Turf / All weather)
  2. Male 3yos
  3. Races for horses aged 3 and older
  4. Distance 1m6f or more

3yo males seem more robust than 3yo females when it comes to running longer distances. The male strike rate since 2009 is over 5.5% higher than the female one, which is why I've included the male rule. Perhaps the reason 3yos run well against their elders in long distance races relates to the sliding weight for age scale; I’m not sure, but the overall stats look decent:

 

 

As ever, we need to look at the annual breakdown by BSP ROI% to give us a fuller picture:

 

 

There have been nine winning years out of 13, which shows reasonable enough consistency but, as we can see from the graph, 2021 has really skewed the overall numbers. Of the £274.10 profit to £1 level stakes, 2021 contributed £212.76. That equates to 77% of the overall profit.

Moreover, if we look closely, 2009 to 2012 produced good results – profits of £142.98 to £1 level stakes. Since then, and taking out that 2021 outlier, there would have been a loss of £81.64. Now that is only a loss of 7p in the £ so not terminal, but this system looks quite risky in terms of backing every selection.

So what options do we have? Well, as I have mentioned before, you do not have to back every selection. If you are a punter who does their own research and race analysis then you can look at each system selection and deploy your usual race reading criteria to decide whether it is a possible bet or not.

It is worth noting that system qualifiers that drop back in distance from last time out have performed below the norm; this is probably what we should expect as the system logic is based on the fact that the younger horses are less exposed over longer distances. System qualifiers that dropped in trip from their last run would have lost you 22p in the £. However, despite the logic, I did not add this to the original system rules as one could argue this is back-fitting (and it is also possible to argue that it is not in this instance).

For me, this is not a system to use in its raw form, but again I would note qualifiers and see if there are other compelling reasons which would potentially produce a betting opportunity.

The "Ran in the Oaks LTO" system

This is a very simple system that has performed well for many years. It is much more a niche or micro system as you will only get a handful of qualifiers each year. For this system I have included runs in Ireland next time out along with the UK, as of course a good proportion of Irish runners take part in the Epsom race. For the record, runners that race in the UK next time make a profit of 80p in the £; those that raced in Ireland next time returned a more modest, but still not to be sniffed at, 12p in the £. Thus, I am adding the Irish course results despite the fact they lower the overall return percentage. For me I feel this system needs to include all runners.

Here is the breakdown:

 

 

As you'll have gathered from the ROI comments above, these are very strong looking figures and, even at traditional SP, would have seen returns of 31p in the £. Let’s look at the annual breakdown. As we are looking at small samples this time I have used profit figures to £1 level stakes rather than ROI%:

 

 

The 2020 race was delayed due to COVID and was run in July with a small field of eight. Overall, there have been nine winning years out of 13. Small losses in four years but considering the small sample size these yearly results look positive. I was also able to back check the previous five seasons to this (2004 to 2008); four of those five years made a profit, too. Profits in those five years would have equated to around 60p in the £; so very similar. So, going back to 2004, we have 13 winning years out of 18.

The system does not rely on a good run in the Oaks, in fact horses that finished fifth or worse have proved the best value, still scoring around 20% of the time and producing returns of approximately 100 pence in the £.

I like this system and it is clear horses that ran in the Oaks require very close scrutiny next time. For the record, horses that ran in the Derby last time out have also made a profit going back to 2009, but one big priced winner has skewed the overall profits somewhat. Taking that winner out would have left you with a very small overall loss. Going all the way back to 2004 backing all qualifiers would have seen you break even.

I hope you have enjoyed this latest probe into system betting. In the next article the focus with be flat racing once again and I will be analysing some trainer statistics.

- DR

Other Recent Posts by This Author:

Your first 30 days for just £1
10 replies
  1. 10 Things You Didn't Know about Geegeez Racecards
  2. Richard
    Richard says:

    Really enjoyed that article (as usual), thanks Dave.

    I have long thought about the Newmarket idea, but would be interested to see how Newbury performs too as many a decent Maiden/Novice is run there. Also, instead of looking at just the first three, perhaps include those horses finishing within a certain distance of the winner, say ten lengths?

    Reply
    • Dave Renham
      Dave Renham says:

      Hi there – I use a variety of tools including a personal database.
      There are at least 3 more articles to come in the series and in two of those (the 7th and 8th articles in the series) 6 of the 8 systems can be created using the Query Tool.
      Dave

      Reply
  3. pfebrey
    pfebrey says:

    Hi Dave, plenty of room for thought, I’m trying to create these systems in Query Tool but am struggling with the Days Since Last Run system, any ideas from anyone how to help me?

    Reply
      • Matt Bisogno
        Matt Bisogno says:

        Hi guys

        Dave’s research is not confined to Query Tool. We do not currently have some of these variables within the QT setup, but we will do later this year.

        I’m very much looking forward to sharing more details on this in a month or two’s time.

        Matt

        Reply
        • pfebrey
          pfebrey says:

          Thanks matt, looking forward to it, a case of the quick returns to scan the website each day for qualifiers.

          Reply
  4. chazzer66
    chazzer66 says:

    Really interesting research – many thanks for all the hard work! Is any of this reproducible in QT or do all these angles require other tools?

    Reply
    • Dave Renham
      Dave Renham says:

      I replied earlier to someone else so just in case you can’t see the reply I’ll copy and paste what I wrote There are at least 3 more articles to come in the series and in two of those (the 7th and 8th articles in the series) 6 of the 8 systems can be created using the Query Tool.
      Dave

      Reply

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply to MrForce Cancel reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.