# [CASE STUDY] The Best Horse Racing Predictions Using Reynolds Ranking

## Quick Steps To Predicting The Winner Of A Race

In this case study I’m going to be focusing on how you can use Reynolds Ranking ratings in your horse racing predictions.

I first came across the Reynolds Ranking scoring method many years ago, reading the content of Dave Schwarz. He’s a US handicapper and bettor who I greatly admire, and who’s come up with a lot of concepts and strategies I follow and believe in.

You can visit his website here, and whilst all of his content is focused on American racing, there’s a lot that can be translated over to the UK.

Like a lot of American racing, it’s very data-centric, which is how I personally bet, and it can take a bit of time to get your head around some of his concepts. But like most things, once you’ve done it, you’ll be pleased that you did.

The purpose of Reynolds Ranking scores are to provide a very quick overview, across a range of ratings, of a horses likely ability in the race. You can think of it as the snapshot guide to the race.

This guide is going to show you not only how to calculate the rating, but also how you can use it in your own betting to find winners and make profits.

## CHAPTER 1: What The Heck Is Reynolds Ranking?

Before we can get stuck into how you use Reynolds Ranking, we need to start by explaining exactly what it is and how we calculate it.

The concept of the Reynolds number first came about in 1851 by George Stokes, however it was popularised by Osborne Reynolds in 1881 and was named after him.

What has it got to do with horse racing?

Absolutely nothing!

According to Wikipedia…

The Reynolds number is the ratio of inertial forces to viscous forces within a fluid which is subjected to relative internal movement due to different fluid velocities, which is known as a boundary layer in the case of a bounding surface such as the interior of a pipe.

If you want to get your head around that, you can read more about it on Wikipedia here.

But you don’t have to in order to be able to use it in your horse racing predictions models and systems.

In simple terms, what Osborne Reynolds essentially did was to multiply together multiple variables. Doing this he was able to combine the variables into a single score that predicts how turbulent the flow through a pipe becomes.

But how does this help us?

Dave realised that this helped horse racing bettors because we have a similar problem. We have a lot of factors which we need to interpret. The more factors we have the harder it is to interpret them, as our brains can only cope with a certain amount of information.

We need to find a way to combine this information in order to simplify our understanding of what we are being told.

If you have 6 ratings for a horse and it’s a twelve horse race, then there are 2,556 possible combinations you need to consider.

Yes… that’s over two thousand combinations from just six ratings per horse!

In a 15 horse field that goes up to 4,005 combinations.

Very rapidly it becomes impossible for our brains to actually compare the data we are providing it. Which is why profitable horse racing can also be consider an art form.

By using our brains to summarise the information, combined with experience and instinct, we can pinpoint profitable bets with accuracy.

However, we still want to make it as simple as possible, and the aim of using Reynolds Ranking scores, is to be able to eliminate horses rapidly, and focus on the strongest runners in the race.

If we can reduce a field of 15 runners down to just 5, we reduce the number of possible comparisons in our six ratings for 4,005 to just 435. Something which is far easier for us to process, with the ultimate aim of making your horse racing predictions more accurate.

Before we move on to how we can use Reynolds Ranking’s to do this, it’s important to understand how we calculate the rating, and that’s what I’m going to explain next.

## CHAPTER 2: How To Calculate Reynolds Ranking Scores

You’ll be pleased to hear that calculating the Reynolds Ranking scores is significantly easier than it would seem based on the Wikipedia’s explanation of what it is.

In fact, it’s simply the case of multiplying numbers together.

Inside the Race Advisor Pro software we do this automatically for whatever race card you use, but it’s perfectly possible to do this in a spreadsheet or with paper and a pencil.

I’ve made a simple race card with just two ratings, these are the PFP and AeEClLGR. The PFP is a form based rating and the AeEClLGR is a prize money based rating for the horses last Good Race (hence the GR at the end of the name).

As you can see I’ve chosen a race which has only got five runners, and two of those are non-runners. I’ve done this intentionally so that it’s easier to see, later examples will include more runners.

In order to calculate the Reynolds Ranking for your ratings, you need to have the horses ranks for each factor.

As you can see above, I’ve also added the Ranking versions of those two ratings to the race cards. However, inside the RA Pro Members Club we automatically calculate Reynolds Ranking scores for any ratings that have a rank available, they don’t need to be added to the race card specifically.

All you need to do to get a Reynolds Ranking is multiply your two (or more) rankings together. For the horses on this card this means:

**Angel Of Harlem**

2 x 1 = 2

**Go Another One**

1 x 2 = 2

**You’re The Man**

3 = 3

Because we’re using rankings instead of raw ratings, the lower the score the better the horse can be considered. This is because a rank of 1 is the best horse in the race, a rank of 2 is the second best horse in the race etc.

In this race, based on these two ratings only, we would be able to see that the horses could be considered strongest in this order:

*Angel Of Harlem (joint 1st)*

*Go Another One (joint 1st)*

*You’re The Man*

Why are the scores inside our software different?

Because we have a lot of ratings. Some of our members have race cards with tens of ratings on them. When you’re multiplying scores together across tens of ratings, you can end up with huge numbers (into the billions) which aren’t useable. In order to prevent this happening we adjust our Reynolds Ranking scores to a logarithmic scale.

If you want to do this, you can simply use a scientific calculator, or a spreadsheet, and do the sum:

log(score,10)

This would change the scores to what you can see here…

Apart from the scale, it doesn’t change anything, the best horse is still the one with the lowest score.

In the above example, I’ve only chosen two ratings at random to show how these calculations are done, and they’re not going to give an overview of the horses’ performance.

This is so you can see how easy it is to calculate Reynolds Ranking’s for any factors that you want. Here’s an example of the scores for the same race using a bigger range of ratings.

### In Summary

Calculating the Reynolds Ranking is as simple as multiplying a horses rank across multiple factors together. Make sure that you always use the horses rank for a factor or rating, and not the raw rating.

**Remember that the lower the score the better the horse has performed.**

If you want to use a lot of ratings, then you can re-scale the numbers by changing them to a logarithmic scale using the sum log(score,10).

Now that you know how to calculate a Reynolds Ranking, I’m going to look at how you can use this information in your horse racing predictions to find a profit.

Going forwards I’ll be using the Race Advisor Pro Members Club software and ratings in the examples. If you’re not yet a member and would like the Reynolds Rankings calculated automatically for you, then **you can register for a trial here.**

## CHAPTER 3: Horse Racing Predictions With Reynolds Ranking

Now we know where the method of Reynolds Ranking came from, and how to calculate it, we can start to look at how we’re going to use it to make a profit from our betting.

In this chapter I’m going to be focusing on how you can eliminate horses rapidly and find crucial information quickly using Reynolds Ranking scores.

In the next chapter, I’m going to look at how you can take this to a more advanced level if you want to.

To do this I’m going to make four race cards, one for each of the following categories:

- Form
- Speed
- Class
- Connections

I’m going to choose a race with more runners for this example, and you can see the Form race card has been loaded first and sorted in order of RR scores.

### Form Race Card

Initially I’m going to be looking for those horses that fall into the top half+1 of the field. This race has ten runners, which means we’re looking for the top 6, and you can see I’ve marked them above.

I’ll do this for each of the race cards and, using the X on the left of the horses name, eliminate any that don’t meet this criteria across all the race cards.

Removing them doesn’t mean I won’t bring them back, but we’re wanting to reduce the field quickly.

#### Speed Race Card

#### Class Race Card

#### Connections Race Card

Once we’ve done this, the horses that appear in the top half of the field, plus one, on all race cards are:

- Geological
- Koybig

That’s it, *just two horses.*

At the time of writing we can see that these two horses, highlighted in green, are third and fifth in the betting market, and don’t have odds higher than 29/1. It’s just personal preference, but I never tend to bet any horses with odds higher than 29/1.

We’ve reduced a field of ten to just two horses in a couple of minutes.

But we don’t just go and bet on them. **We now need to do our due diligence…**

- Confirm that the horses are suited to the conditions and likely to run well
- Confirm which other horses are threats and how much of a threat

These two steps are both very important in the process of selection finding. The first is crucial because a horse may be strong, but completely unsuited to the conditions of the race. If they’re unsuited to the conditions of the race, then we don’t want to be betting on them!

The second step is also important. Only by identifying the possible other threats in the race can we make a judgement on whether the odds for the horses we want to bet on are good.

### Doing Our Due Dilligence

We start by noting that this is a flat, All Weather race, running over seven furlongs, with a winners prize of £6,655.

Clicking on a horse’s name we open up it’s racing history…

You can see that similar races to todays, shown in the red box above, have resulted in poor performances from this horse. Two of these had significantly more prize money, but even so it’s not strong.

Scrolling through the rest of the horse’s history we can see that despite having had a good race 41 days ago, it hasn’t won a race over similar conditions since 2017.

Apart from this being over eighteen months ago, the weight the horse was carrying was 9-13 and today it’s carrying 12-0.

This gives us significant reason to think the Koybig may struggle to perform today, and we can remove it from our selections.

Geological has more experience on the All Weather, but hasn’t raced on it in 2019. Here are just some of the later races in 2018.

While only having won one of these races on the 17th August, he performed well by finishing within three lengths of the winner in 64% of them.

The race he won, he was carrying much less weight than he is today where he’s the top-weighted, but the prize money was nearly 300% more than on offer today.

This tells us that he can definitely perform, the question will be whether the weight is too much for him.

As this is a sprint race, although at the long-end of sprints, I’ll also take a quick look at the pace.

I’ve expanded our pace type analysis for the race above the race card. This tells us it’s a fast race, and the most likely winners come from mid-pack and early pressers. A lot of the horses in this race are mid-pack horses, but importantly our selection is looking to not have a pace issue.

If he’d had a pace issue, alongside the potential weight problem, I would have most likely skipped the race.

In summary… **step one of our due diligence has shown that one of our selections looks strong enough to race in the current conditions, but there is a concern over whether the weight he’s carrying will be an issue.**

*On to step two of our due diligence!*

The quickest way to look at potential threats in a race is to sort the race card by Betfair SP.

The current favourite is Blacklooks. The second favourite is War Hero, but his odds aren’t that much shorter than our selection. We’ve already removed the third favourite, Koybig, as a selection, and Ken’s Sam’s is only marginally better in the odds than our selection.

Usually I’d look closer at any horse which has significantly lower odds than our selection. In this case, we’ll take a closer look at Blacklooks.

This horse has raced on All Weather more recently than our selection, however the performances have been very poor with him finishing 18.5, 8.95, 7.75 and similar lengths behind the winner, even in races with less prize money.

His best performance have come on the flat turf where the ground has some give in it.

We don’t need to look much further for me to be confident that his odds of 3.7 seem to be way too short for this runner, particularly considering he is also carrying more weight than he’s used to.

### In Summary

After doing our analysis and due diligence, we’ve found that there’s one selection in this race that looks strong enough to place our bet on, Geological.

There is a concern over the amount of weight being carried, but the odds do seem to take this into account.

At the time of writing, the odds on offer are 8.20, and our PR Odds rating is suggesting that they should be 8.49.

Taking everything into account, personally I will be looking to place an 80/20 bet on this runner, 80% to place and 20% to win. I will also be aiming to get minimum win odds of 8.50.

Whether this selection wins or not, the process outlined in this chapter shows a quick and effective way to analyse a race and create your own horse racing predictions even if you’re new to betting.

## CHAPTER 4: Advanced Methods

If you’re an advanced bettor, or if you like creating systems from data, then there’s more ways that you can use Reynolds Ranking scores.

In this chapter I’m going to look at one of those ways which you can use the scores to create systems and automate your horse racing predictions. I cannot claim creation for this method, it came from Dave Schwarz from whom I first learned of the Reynolds Ranking scoring method.

If we take a range of similar races to the one we’re looking at, and check the Reynolds Ranking scores for the winners, we’d get something like this.

Rating 1 | Rating 2 | Rating 3 | Rating 1&2 | Rating 1&3 | Rating 2&3 | Rating 1&2&3 | |

Race 1 | 3 | 1 | 5 | 3 | 15 | 5 | 15 |

Race 2 | 1 | 3 | 8 | 3 | 8 | 24 | 24 |

Race 3 | 6 | 2 | 1 | 12 | 6 | 2 | 12 |

Race 4 | 2 | 2 | 3 | 4 | 6 | 6 | 12 |

Race 5 | 5 | 1 | 4 | 5 | 20 | 4 | 20 |

Race 6 | 8 | 5 | 1 | 40 | 8 | 5 | 40 |

Race 7 | 3 | 6 | 2 | 18 | 6 | 12 | 36 |

Race 8 | 2 | 4 | 1 | 8 | 2 | 4 | 8 |

Race 9 | 5 | 1 | 6 | 5 | 30 | 6 | 30 |

Race 10 | 4 | 1 | 7 | 4 | 28 | 7 | 28 |

If we then look at the worst scores across this sample of races we get this:

Worse Score | 8 | 6 | 8 | 40 | 30 | 24 | 50 |

If you take the strongest two ratings or combinations, then you’d come to the conclusion you’re looking for horses that:

- Have a Rating 2 score of 6 or less
- Have a Rating 2&3 score of 24 or less

Remember that the lower the score the better.

What does this mean?

It means that we can look at the Reynolds Ranking scores for winners in similar races to build a profile of the type of horses that win these races, and then highlight them very rapidly.

Yes, you’d need a sample of more than ten races.

And yes… you’d also probably need to use more than three ratings.

However, you could use a similar approach to the method outlined in Chapter 3, and perform this for a set of ratings in each of the major categories Form, Speed, Class and Connections.

Doing this would mean that you’d not only be using the method I’ve outlined, but you’d also have baseline scores for the winners of similar races to optimise the elimination steps further, making the process more systematic. The more systematic you make the process, the greater the possibility of automating your horse racing prediction process. If you can automate the process, then you can analyse more races.

## Where To Go From Here

Throughout this case study, I’ve introduced you to Reynolds Ranking scores, how to calculate them, how to use them to find selections and how you can take this further if you’re interested in using them to create horse profiles and systematise the selection process as much as possible.

I’d love to know if you already use Reynolds Ranking scores in your betting, or if you’re now going to. Please let me know by leaving a comment below.

For me, I’ve added the idea of adding Reynolds Ranking profiling to the RA Pro Members Club development list. The concept being that we’d do the profiling for you on every race you open, so that you know exactly the type of horses which have won similar races in the past. *If that’s something you’d like to see then please let me know in the comments below.*

Interesting and worth testing. As for you carrying out the profiling on the races I’m interested in, that’s fine with me !

Thanks Graham.

The concept being that we’d do the profiling for you on every race you open, so that you know exactly the type of horses which have won similar races in the past. If that’s something you’d like to see then please let me know in the comments below. That would be a great timesaver for me. Great article, thanks.

Great, thank you Gary 🙂

Thanks for that Michael. I’m afraid its just beyond me. I need a simple tipster i think who just gives me their selections

Not a problem Ian, if there’s any questions you have then just fire away. In the meantime I hope you’ll be watching the 18:10 race today 🙂

Hi Michael, your proposal is just up my alley as I like to use the ratings and certainly anything purposeful to eliminate selections has to be a no brainer. yes please do get it into the develop mode as I think impressive stuff could result. regards

Thank you Eric.

And after all that it finished seventh 🙂

I did say I couldn’t guarantee a win 🙂 However the selection process is solid, and this selection offered value. It’s possible to see it offered the value we thought because when I published the article the odds on the selection were 8.20, although I wanted to get 8.50 it never got quite that high and I took the 8.20. However, it went off at 6.21 Betfair SP. As we know the Betfair SP markets are very accurate in their prediction of probability of winning, which means this horse had a 16.1% chance of winning according to Betfair SP, and we took odds that indicated it had a 12.2% chance of winning, giving us a massive edge on the bet. Although this one didn’t come in, if we bet this equivalent type of bet lots of times, we would expect to get 16 winners out of 100 bets on average in the long-term, and make a 9% ROI. Watching the race, I think the weight got to him, he was doing well but couldn’t keep it up in the last two furlongs. At a lower weight in similar competition he should perform excellently, could be one to add to your EyeCatcher tracker.

How about finding these horses for a back to lay if the odds at the time of finding them are showing a higher value then they should be, that way we can make money straight away without having to wait for a winner only 16% of the time.

That could also work. The overall strike rate may be higher, the above example was showing how we can see that although that specific selection lost, we found value and that means we’d make a long term profit

Well…………….. In your first example you gave “You’re the man” a score of 3 !! Even the most basic hand held calculator will tell you that 3 x 0 = 0 Therefore, as 0 was the lowest score perhaps that should have been the chosen selection. I gave up soon after that as the calculations became even more difficult to follow. ( I’m 85 next March so perhaps the time HAS come for me to give it all up ) Whatever, it’s not for me. Thanks anyway !!

Thanks for the message Ronald. You’re right about the score, someone else mentioned that as well. I’ve gone back and corrected it now. The score is 3, it shouldn’t have been multiplied by 0 because the horse did not have a ranking for the second rating. A horse can’t be ranked 0 in a race, it can only be ranked from 1 (best) to the maximum number of runners in the field. This means it should not have multiplied by anything, making it the worst horse of the three.

I’d love.to see the profiling development!

Hi Michael interesting article. Could you clarify a couple of things? 1/ In what kind of races is this method best used? Handicaps, large fields? 2/ You mention using factors that are rank, but your cards feature raw factors – am I missing something? BTW particularly enjoyed your novice racing article. Have developed a betting strategy around this which is showing promise – not there yet. Will post something on the forum should it prove worthwhile.

Thanks Ian. I tend to prefer to use this approach in races where we have a solid amount of data for all horses racing, the more data the better. In terms of race conditions, it doesn’t tend to matter, what’s more important is that the ratings chosen are good and don’t overlap each other too much.

In the RA Pro Members software the RR (Reynolds Ranking) will automatically use the Rank versions of the ratings on a race card. If there aren’t Rank versions then it won’t use them. You can see which of the ratings on the race card it is using by clicking on the Reynolds button.

Uncheck any that you don’t want it to use. You can also create your own race cards from the Settings page, as shown in Training video four, and you can choose to only look for the Rank versions of a rating if you want to use that instead of the raw rating. Most of our ratings have a Rank (Rnk) and a Difference From Top (DiffTp) version so that you can use those instead of the raw versions. They can be easier to use when you’re getting started as they provide an easier way of determining difference between horses 🙂

Thanks Michael, that’s answered my questions nicely. There is quite a lot hidden in the software such as knowing that a translation happens when using raw data. However I am learning more ideas from this than elsewhere and that is good.

Hi Michael. My first ever comment so just like to say I’m enjoying playing with the site (although not too profitable at the moment). I’ve been playing around with this and built an Excel macro that calculates these ratings from the spreadsheet download. Looking at all flat turf Handicaps since 16th August I found 500+ qualifiers (I won’t have an exact number until NRs are removed). I’ve pulled up the results for the first 102 selections and it’s returned a loss of -13.54 points to SP – that’s straight win <11.0, EW 11.0 – 21 and 20/80 21-30, discounting anything 31 or above. The strike rate was a little disappointing 12 winners and 21 placed (although not all bet on). I'm wondering if you think the method would work better if the higher class races were concentrated on? Thanks

Hold the front page! I’ve just taken a look at my code and see that I’ve made an error in my code (< instead of <=) which will have drastically reduced the number of qualifiers – I'm really trying to establish hoe successful the method is in producing a short list. I'll report back. Thanks

Sorry to hijack the thread. But I’ve done a quick check of the result for handicaps on 16th August using the correct criteria. there were 5 winners from 16 races (32 selection in all) which is just over 30% SR for the shortlist. Betting without any form study to flat stakes on SP would have produced a profit of 3.5 points, almost certainly better using BSP. Obviously I’ll need to go through a lot more races but I’ve got to start wok now and I didn’t want anybody to discard the idea because of what I might have written. Thanks

Thanks for the comments 🙂 Great to see you’re starting to play around with the idea and have already found the possibilities are there.