Mythbusters (Revisited)

In November 2017 Matt published an article that looked at five well established horseracing adages, writes Dave Renham. He tested them by looking at data going back to 2011 and he split his findings into two comparing an earlier data set to a more recent one. You can find the link here.

In this piece I will revisit his ideas whilst making a few little tweaks here and there. I will go back to 2011 like Matt did, but we now have an extra six years of data to add into the mix. Hence the research covers the time frame of 1^st Jan 2011 to 31^st December 2023.

Matt looked at UK racing only, which I will also do, and, in his piece, he lumped National Hunt, turf flat and all-weather racing together. I will do the same but will additionally focus on individual race codes when appropriate.

So let's get started with...

“Back the outsider of three”

This saying is a popular one, but what do the stats tell us? I have used Betfair Starting Prices (BSP) in three horse races to order the three runners in terms of market position. We'll begin by sharing the win percentages for all three market ranks:

 

 

These percentages are as one would expect with respectful gaps between each. Indeed, the favourite scores roughly twice as often as the second choice, who in turn wins approximately twice as frequently as the outsider. All well and good, but what about the bottom line? Here are those findings with both Industry SP and BSP figures shared:

 

 

As we can see, to Industry SP the returns are virtually the same. However, to BSP the outsider of the party has turned in a return of just under 6p in the £. This is an excellent illustration of the fact that we should not be lured in by high win strike rates. Of course, higher strike rates can turn a profit, any strike rate can. However, as punters we need to look for value because, ultimately, we are much more interested in profit than winners, right?

Now if we break the outsider of three data down by individual year big fluctuations in profit/losses can be seen, due mainly to sample sizes and standard variances. Hence, I am going to look at the annual data in a slightly different way using a method I first saw in Nick Mordin’s excellent book, ‘Winning Without Thinking’. He looked at data in five-year batches, which is a good way to try and compare things more effectively due to more reliable sample sizes. You can also see patterns changing more easily – if indeed they do change. Here is the breakdown including the Betfair profit and loss figures for these five-year groupings:

 

 

The strike rate has been consistent as one would expect given the bigger yearly groups. The majority of the five-year batches have seen a profit (six of the nine); and the three losing years showed only modest losses, with one of these losers (2019 to 2023) effectively breaking even.

Next, I thought it would be interesting to see how the outsider of three fared in different race codes so here are those figures:

 

 

A BSP profit for all three – it is interesting that the vast majority of three runner races occur over the jumps. So, while we're about it, let’s split the NH data into chases, hurdles, and NH Flat races:

 

 

Wow! That was worth doing. What a difference. Three runner chase races look to be the way to go. I am not sure why this is the case: I guess jumping mistakes become more significant in smaller fields so that could be part of it. Whatever the reason it certainly gives food for thought. I looked at the yearly breakdown for the outsider of three in chases and there were a couple of poor years, but nine of the 13 turned a BSP profit.

All in all, ‘back the outsider of three’ is an adage that seems to be TRUE.

There are plenty of worse betting approaches I can tell you!

 

“Never bet odds-on in a novice chase”

I heard this one in a betting shop when I was around 19 and just starting to dabble in the sport. You would think this one may be true given that novice chases are for horses with limited chasing experience and jumping mistakes are probably more likely. So, let’s look at the 12-year data:

 

 

It looks like the old adage is true given this initial data. Losses are quite small, but you would need a huge uptick in win percentage to get anywhere near a profit due to the short prices. If we look at the 5-year batch results, we see that all groupings produced a loss to BSP:

 

 

The losses range from just over 1p in the £ to just under 6p. I did also look at splitting the SPs up into groups to see if that would show us anything. Here are the findings:

 

 

The bigger the odds price the poorer the returns; it certainly seems generally worth swerving novice chasers priced between 4/6 and 10/11.

It seems for the second time in this piece we have a TRUE adage. At least it's fair to say we're edging towards TRUE over FALSE, especially at the odds-on prices closest to evens.

 

“Back the longest traveller”

This one does appear to have some logic behind it: why send a horse a huge distance unless you strongly fancy it, right?

To start with I looked simply at the ‘longest traveller’ – this includes joint-longest travellers, too. That's because distances are measured not only from individual stables but from training centres also (like Lambourn or Newmarket, for example). Therefore, we see a good number of joint-longest travellers. Here are the overall findings by race code. I have not included horses from overseas:

 

 

The strike rate for the 'all qualifiers' group (14.2%) is above the average for all horses in all races (average SR% is around 11%). However, despite this, losses are broadly in line with the average, both in terms of Industry SP and BSP returns. (Average ROI% for all horses in all races using Industry SP is –24%, BSP stands at –6%).

Before moving on, these 'ALL race' figures shared give you a baseline to judge any set of racing data / stats. They cover of 130,000 races in the UK since 2011 so we can be sure these figures are accurate.

Back to the longest traveller table and we can see that the turf flat group have fared slightly better than the rest in terms of returns, but those losses still rack up over a long series of qualifiers. Let's now examine the actual distance travelled by these longest travellers. Below is a graph showing the different win strike rates for different distance bands:

 

 

I have mentioned previously that strike rates do not tell the full picture, but it is noteworthy that the very longest travellers (300 miles+) have been by far the most successful group. Breaking the data down further we can see the profit/loss and returns for each travelling distance band:

 

 

The profit/loss figures make far more comfortable reading when we get to 250 miles or more. Both the 250–299 and 300+ groups have performed much better to Industry SP compared with the other groups and are close to breaking even to BSP.

It will come as no surprise that the 250-299 and 300+ groups had a few big-priced winners which of course will skew the figures, but all the other groups had similar high SP winners. In fact, the 300 or more group had fewer big-priced winners in proportion to the number of qualifiers than any other group. The longest travellers had just one winner priced over 100/1 (BSP 110.96) and one priced between 50/1 and 100/1 (BSP 53.85). Compare this to the 100-149 mile group which had ten winners returned above 100/1 including a BSP price of 880.09, and 15 winners between 50/1 and 100/1.

I would also like to share that horses which have travelled 300+ miles and were priced 7.0 or less on Betfair were not far from break even (loss of only 1.5% from 2497 runners).

When Matt looked at this in his article the adage seemed to be a strong FAIL – in that shorter time span at least. However, these longer 12-year stats are not as bad, especially if focusing on horses that have travelled 300+ miles. Using the 5-year results grouping technique, we can see that the figures have improved since the first piece was penned:

 

 

In conclusion, the adage ‘back the longest traveller’ looks still to be a FALSE one, but I suspect that adding a couple of extra filters, assuming they are logical and not back-fitted, may offer a chance of parity or even a small profit in the future.

  

“Follow a filly in form”

This is another extremely well-known saying. While the first three maxims we looked at were clear cut and obvious in terms of meaning and how to test them, this one is less so because it is harder to quantify the term “in form”.

In Matt’s article he focused initially on last time out winners including all female runners. That makes sense and I’ll start there as well. Hence, here are the overall stats for LTO female winners and splitting them by race code:

 

 

This paints quite a bleak picture for LTO winners despite decent looking strike rates. The all-weather returns are by far the worst of the three codes which should come as no surprise if you have read previous all–weather articles I have written where I've referenced female runners. In those I have shared data showing that females under–perform on the sand compared to the turf. To provide some numbers, let me compare the win strike rates of females on both surfaces – this is for ALL runners, not solely LTO winners:

 

 

There is a significant difference of 1.5% and this is a fair test because the average field sizes in both codes have been the same over the past twelve seasons. Not only that, the A/E indices are in favour of the turf runners too (0.87 v 0.82). Meanwhile, losses have been more than 7p in the £ worse on the sand (–11.9% v –4.7%).

Switching back to last time out female winners now, and it should be noted that fillies are female runners aged 3 or 4 so let me split the fillies’ data out from that for older mares (5yo and older). Mares have won slightly more often when attempting a repeat win (19.1% v 17.8%), and Industry SP and BSP returns have been similar too with a 1% difference for Industry SP and 0.3% for BSP.

With the age of these female runners not really making any difference to the stats, for the remainder of this section I will continue to look at both fillies and mares combined. There seems no obvious reason not to do this.

Onto to looking at female horses who have won their last two starts. Does that improve matters?

 

 

Once again, the AW bottom line is bad. However, back-to-back wins have certainly improved matters overall.

Indeed, females racing in flat races on turf have snuck into BSP profit. Looking in more detail at the turf flat data for these hat-trick seekers we find that figures are, perhaps unsurprisingly, not skewed by big-priced winners. There were 117 horses that started at a BSP of 35.0 or more and only one won – backing all 117 runners in this price bracket would have lost you £58.45 to £1 level stakes (ROI –50.0%).

Sticking with the hat–trick seekers racing on the turf, I decided to look at their two previous wins in terms of the surface they raced on. My hypothesis was that if one or both had come on the all-weather, then those wins on the sand would potentially be more significant and perhaps these horses performed even better when attempting the hat-trick. So, of the 2872 hat-trick seeking females racing on the turf, 748 of them had notched up either one or two of their back-to-back wins on the all-weather. Here is the full breeakdown for these runners:

 

 

It's nice when you are vindicated about a theory! Females who were able to win one or both of their last two runs on an all-weather surface have shown the profitable upgrade in performance I was hoping for whilst all but maintaining the turf win strike rate.

Before ending this section, I did quickly look at female runners who had finished in the first three on their last two starts and I’ll share them below:

 

 

What we can again take from this is that the bottom line is again much better on the turf flat compared to the all-weather (6% difference to BSP).

Taking all the data shared in this section the term “follow a filly in form” or to be precise “follow a female runner in form” is generally FALSE. The caveat is that hat-trick-seeking females racing on turf flat are possibly worth following if one or both of their previous two wins came on an all-weather track.

 

“The bigger the field the bigger the certainty”

Onto to our last adage. Again, it is not totally clear how this should be tested, but considering the word ‘certainty’, we should be looking at the clear favourite. How big though are we looking at in terms of the size of the field? Matt in his article chose 16 which is completely logical. For me, to begin with, I am going to look at it slightly differently and consider the favourite across all field sizes. This will hopefully offer some context. I have elected to group field sizes thus: 2 to 5 runners, 6 to 10 runners, 11 to 15 runners, and 16 or more runners. Here are the results for outright favourites:

 

 

The strike rates differ drastically but you would expect that given the number of runners. The ROI percentages for BSP are all within 0.6% of each other, but sadly the big field group (16+ runners) has produced marginally the worst returns.

Sticking with the 16 runners or more group because the adage states “the bigger the field”, let me split the favourite results by Race Code:

 

 

The turf flat and National Hunt have similar bottom lines, whilst there are very few AW races that qualified. How about splitting now by handicap and non-handicaps? Here is what I found:

 

 

A difference can be seen here with non-handicap favourites losing notably less. Indeed, to BSP they are within 1% of breaking even. On the flat non-handicap favourites with 16+ runners have just edged into SP profit.

Let me now break it down by the actual price of the favourite. One would guess / hope that the shorter priced ones would perform better, being more of a ‘certainty’ – well according to the price anyway! Here goes:

 

 

We seem to be getting somewhere here with the 9/4 or shorter jollies hitting a BSP profit and, obviously, no big-priced winners skewing the stats. So maybe this maxim has some veracity.

The penultimate test is to look at these 9/4 or shorter favs in 16+ runner fields where they were well clear of the second placed horse in the betting market. I chose an arbitrary measure of 3 points or more to see how that particular cohort did (e.g. if the fav is 2/1 then second fav must be 5/1 or bigger, etc). Here are the results:

 

 

A further improvement with Industry SP nicking a profit now, too, though we've wittled the sample size by a fair amount.

Finally, I want to take this cohort (9/4 or shorter, 16+ runners, 3 points clear of second fav) and split non-handicap vs handicap. When Matt looked at something similar to this, he noted the handicap results had proved profitable. Let’s see what these figures tell us:

 

 

The handicap stats do produce a positive outcome, hitting nearly 12p in the £ profit to BSP. For the record, virtually all the profit came from National Hunt favourites. Sadly, the sample size for these clear favourite handicappers is quite small, but there will be plenty of worse bets one could strike in future.

It seemed this adage would have been false until we considered shorter prices and differences in price between the favourite and market second choice. Hence, we can give this maxim a tentative TRUE if focusing on the 9/4 or shorter group.

*

This exercise has been enjoyable to work through, as well as hopefully building on previous research by Matt. Of course, you can interpret any of the adages in a slightly different way to me, and perhaps use slightly different research points. However, from the data I have shared in this piece I think there are plenty of interesting takeaways.

Until next time...

- DR

 

Other Recent Posts by This Author:

• Looking at Past Cheltenham Festival Trends
• Specific Course/Distance Analysis: Lingfield 1m2f
• More on Price Movement in NH Markets, Part 2
• More on Price Movement in NH Markets, Part 1
• ‘SR’ Ratings on the All-Weather