Four Corners

The Racing Horse is indifferent to the plethora of supposed new betting advice systems and strategies that come on to the market. A few deserve scrutiny but the vast majority are either bogus or not fit for purpose. Some authors/sources deliberately procure false hope, others albeit unintentionally, enlist less important or flawed features and give wrong credence to certain aspects. More recently we have seen those that create systems and strategies that seek to complicate in order to raise their own profiles and then of course, we have a few that make a profit.

The Race Advisor sifts this pool of information so we don’t have to!

Essential prerequisites carrying argument weighting are what we require before wagering. This common sense rationale has stood the test of time and make up the ‘Four Corners’ of our betting slip. This raison d’être was vital in 1950, essential now and will be relevant in 2050. With the help of Race Advisor we would like to herald and trumpet this lost message and reaffirm those quadri-points in order of importance!

Trainer Form
Requisite Class
Optimum Ground Conditions
Jockey Competence

The single most important element to any winning bet is ‘Trainer Form.’ The sum of the other parts is not as important. If there is a cornerstone to win betting this is it!

Acknowledging trainer statistics and the mathematics that accompany them before making a win single bet is vital. It is a decisively indicative test that removes randomness, opinion and bias, it reveals how well the stable’s horses are currently and delivers an up to the minute litmus test. If backing the horses of trainers on the cold list is madness then it must be sensible to back trainers who have recorded winners?

Most of us know that there any amount of random factors that can affect a result and it takes small margins to win most races. These issues are more likely to be compounded if the horse is not fully wound up! Furthermore, if one accepts that handicap racing forms more than half of all races on the cards of UK race meetings a few pounds of improvement or decay can mean the difference between winning and finishing down the field.

Each horses has a potential ability level, whether they achieve that is another matter. The better the horse is trained and looked after the more it is likely to achieve but it is accepted that genetics and skeletal soundness are factors outside the trainer’s remit. Finding the bandwidth to which each horse can aspire needs the trainer to be totally in tune with his charge and, whilst some achieve most fail in this regard. Based on this incongruity we only trust recent form.

We were fortunate to spend the day with Monty Roberts (The Horse Whisperer) and he gave us lots of information on horses and we adhere our betting to his golden principles. He told us that every horse that has ever been born or ever will be born would have optimum ground conditions. The onus rests solely on the bettor to find out before making an investment what that is!

Some racing people including trainers, irrationally spew that good horses go on any ground – they do not, some just handle it better or have something in hand of their rivals. A problem for systems is that they are unlikely to recognise/weigh/measure optimum or deteriorating conditions against the class the horses are racing in! It is here where most bets are won and lost.

So our intended wager has a tick in every box and a winning chance looks obvious, why would anyone compromise the bet once the horse arrives at the course after weeks/months of schooling and training? We know that trainers are not duty bound to give the ride to the best jockey available! They might use an apprentice, someone who is contracted, rides work or a favoured son. The owner might have some input or the decision might be political. But the bettor has no such restrictions. It is hard to measure or be scientific about the importance of jockeyship, but they are the trainer’s connection to the horse and it is often their decisions that dictate the result and our profit and loss.

The quadri-points listed probably represent 90%+ in importance towards a winning bet. Of course stable shenanigans, course configuration, draw and track bias, pace in the race, distance, weight, breeding and more can influence the result but the big four represents a core and serious starting point to any bet. This is where we part with most systems and strategies!

We always imagine placing a bet without one of those elements satisfied. In the cold light of day and without emotional fueling how confident would you feel about your bet then?

Paul Moon
The Racing Horse

Race Advisor

The Race Advisor is one of the leading blogs on betting in the UK. Our goal is to help you generate more profit from your betting than you’re currently doing. With thousands of blog posts and some of the leading horse racing analysis software out there, if you’ve got a question about betting then we’re here to make sure you get the right answer.


  1. I am retired with limited resources but following the and watching horse racing on Channel 4 allows me to continue with one of life’s pleasures. Paul Moon speaks more knowledgeably than most of the (highly paid) so called experts on Channel 4’s Morning Line. My biggest bet so far is £5 and after starting with a £10 pot in September and following the Pacafi system my pot now stands at £100. Now I’ve tested the system I am ready to increase my bet. Watch this space!

    1. Ella

      Thank you for your kind words. Do not have that much knowledge but hopefully recognise winning opportunities.

      Pleased that you have followed us and hope we continue to win you money. We will watch this space! Any clues as to what we are likely to see?

      Thanks again…

      Paul Moon
      The Racing Horse

  2. Hi

    An interesting article in which a major emphasis is on the trainer. When Adrian Massey produced his statistics for every race every day I produced a system that took into account over 90 factors for every horse from its first run.

    I analysed these factors to produce statistics from which I rated every horse in every race. These statistics produced some unusual results.

    You cite class as one of your four corners. There is an apochryphal saying in racing that ‘A drop in class indicates a winner’. Statistics proved otherwise. When a horse rises in class it has a greater chance of winning than one that drops in class.

    The other corner you highlight is ground conditions. Very difficult to ascertain when predicting winners the day before. No need to explain why!

    I rate races that take three of your corners omitting ground conditions as it would be impossible to give a range for selections for every ground condition. It is not impossible that the punter takes the three selections and then rates them nearer the time of the race on ground conditions from its past form.


    1. Alan

      You mention 90 factors for every horse from its first run? Crikey, I remember counting critical factors before starting the Pacafi and I think I got to 20! Would love to know some of the more obscure features. Please tell?

      Alan, wanted to talk about the class differential in a bit more detail but had to work within a certain amount of words. It is interesting and surprising that you have statistics that show ‘when a horse rises in class it has a greater chance of winning than one that drops in class.’ From a naked point of view the maths of that seem strange?

      In any case when we bet we are not the slightest bit interested in literal class raisers or class droppers per se unless they carry the rest of our strategy with it. We know the folly too well of indulging in a pic ‘n’ mix mentality from a strategy. Of course if the horse comes into the race down from Listed class to a class 2 displaying excellent trainer form, with optimum ground conditions and William Buick replacing previous rider Ann Stokell our eyes light up! Of course every horse selected must come with perceived value attached irrespective of claims. Conversely if we saw a horse down from Listed to class 2 but the trainer showing poor current form with his horses, on going the horse has never encountered and at odds with breeding we totally ignore it!

      Furthermore, we have about five years of historical data and our worse mistakes have been going in big with one of the big four components missing or weak. On 3rd October 2014 we took a chance with Shane Kelly when riding Rembrandt Van Rijn at Wolverhampton. We had taken 4/5 and 8/11 and the horse was backed into 2/5. It scooted clear going into the straight and would have won 10l eased up. Only the jockey knows why he decided to hit it with the whip, the horse jinked and he fell off and the colt remains a maiden. Jockey competence! Our profit for that month read +5.62pts it should have read +7.34.

      Of course you are right to highlight how difficult it is to predict winners the day before because of changing ground but that is our task. We were mindful of our bet today for those very reasons and hope the weather stays fine for Cheltenham. Go Balthazar King!

      Alan, thanks for kicking off my stint with Race Advisor. Please tell us a little more about those 90 factors…

      1. Hi Paul

        Sorry to be late in responding to your query. Currently am in a morass of program alterations that are holing up my systems. Will hopefully be able to reply tomorrow


      2. Hi

        There is a gap in racing and have managed to get up-to-date. Yesterday, in order to get my selections on line last night I ran my system in the wrong sequence and as there two meetings repeated today that ran yesterday, meant I had to do some work manually and have lost some other data.

        A days recording of results is completed after running 27 programs. The selections for all my systems is completed when I have run 25 programs. You will understand if I get prickly when people imply I am a charlatan which I take you are NOT implying.

        To understand where I am coming from I am more interested in system development and programming than horse racing. Though the results of my developments give me lots a of satisfaction. Lingfield today 6 winners out of 8 races. Wetherby 6 out of 7.Uttoxeter 5 out of 7 and Cheltenham 6 out of seven.

        To answer your query re 92 factors.

        If you take the form of a horse, I didn’t just take the form over its whole career, but its form over the current season, the form over each type of race (flat, all-weather and jumps), plus the type of race (handicap, seller, nursery,etc) form over distance, form over going, form with trainer, form with jockey, form with trainer/jockey combinations and so on. I then took the jockey apart and then the trainer. Performance when changing class, performance when changing class up and down plus number of changes, etc. Its performance after days since last run,etc. The checks were endless.

        All these checks were based on statistics I kept myself resulting from races which obviously took a while to build up. One conclusion I did come to in all this process was that the statistics were meaning full after a month and changed very little in the later months.

        To give you the statistics in which you find yourself ‘naked.’

        Class changes:
        Drop of 4 classes produced 7.5% winners
        Drop of 3 classes produced 6.68% winners
        Drop of 2 classes produced 6.8% winners
        Drop of 1 class produced 6.82% winners
        Increasing class by 1 produced 7.45% winners
        increasing class by 2 produced 9.57% winners
        increasing class by 3 produced 12.59% winners
        increasing class by 4 produced 12.5% winners
        Increasing class by 5 produced 8.6% winners

        The above statistics indicate to me that trainers run unfit horses to lower their rating and when they have them at peak fitness put them in better class races. Not illegal just tactics.

        As Wolverhampton is due to start and I have unpublished selections at the meeting I will draw to a close. I already have a 20/1, 6/1, 15/2 and 9/2 winners on this system giving me a nice profit.

        If you have further queries I will happily answer them.

        1. Excellent post Alan! Your numbers are an antidote to the usual ‘hand-waving’ that passes for analysis in some handicapping circles.
          It is OK to have an opinion but unless it can be quantified and then tested the opinion is often worthless.
          Trainers are supposed to present their horses in a suitable condition to win the race in question but quite often this rule of racing is ignored. I remember only once seeing the stewards in my jurisdiction fine a trainer for violating this rule. The horse in question was grossly overweight and unfit yet this prominent trainer was given a very small fine. Cheers, Peter.

        2. Alan

          What a fabulous email and thanks for the time and detail! You have given me plenty to think about. Can say in response to your “The above statistics indicate to me that trainers run unfit horses to lower their rating and when they have them at peak fitness put them in better class races. Not illegal just tactics” that it makes absolute common sense. Your email has now been etched into my thought processes!

          Alan, apologies if I implied anything. I can say that I am grateful you took time out to explain. Fascinated by the result list of class changes, will digest the contents and contact you again if I may?

          Thank you

          Paul Moon
          The Racing Horse

  3. I use these elements for my study always,but make allowances for class (improving horse)the only bit we have to
    take for granted is how the horse is on the day.None of us feels A1 everyday

    1. Melyvn

      Totally agree with ‘improving horse’ comment. All of our 2yo, 3yo and 4yo bets would come into this category! Generally speaking we do not look for horses to spring back after a poor run or a poor couple of runs. Also agree with selecting the horse for today. We totally disregard last season’s form apart from the ground conditions and whether the stable backed it or let it drift. We might make a mental note of early speed or fast finisher and how it carried itself.

      You remind me of what Monty Roberts said regarding the racing of equines when you said ‘none of us feels A1 everyday.’ They are not machines and we must never get upset with them. When we have a losing bet we blame ourselves.

      Thanks for your comment…

      Paul Moon
      The Racing Horse

  4. I think the first paragraph goes a long way to summing things up-it is so true. Yes this article is the real thing and for those who are a bit unsure and to be sure, that must be most of us, then it is a must read, sooner rather than later.

    1. John

      Thank you very much for your comments. You are obviously a gentleman! I can tell how you write…

      Best Wishes

      Paul Moon
      The Racing Horse

  5. Excellent article and write up Paul most informative totally agree about trainer form trainers can goes for months without a winner yet people will still follow them and the great late big punter Alex bird would agree on going conditions his father was a bookmaker and would rub his hands when the heavens opened and the rain came I for one am very wary of backing any horse on heavy going

    1. Stephen

      If I had 43 lives would never understand why a punter would back a horse without checking trainer form first! Working for Ladbrokes I had seen punters put four figures (in the 1970s) on horses when the trainer had a bug in the stable! On a couple of occasions I warned punters and once was threatened physically by a son of a father claiming I was interfering. I could not put that episode in our black book behind the counter because I would have got into trouble with Ladbrokes! That horse was pulled up and it was a Flat race!

      I love William Hill Radio and listen most days but the pundits like Mick Naughton keep telling punters that the trainer must come off the cold list one day! Yes exactly – one day! I want every box ticked on my bet and leave hope and prayer to those of a religious persuasion! When it comes to pounds, shillings and pence I like maths, it never lies to you, no matter what!

      Totally agree with heavy going comments. The more testing the going the more it dilutes our strike-rate and we only back with it in our favour. Ideally we would not give selections on heavy ground but have to provide a service to our members. Managing this remains our hardest task! It is no coincidence that our biggest profits historically are in June and July.

      Stephen, thanks again for your comments.

      Best wishes

      Paul Moon
      The Racing Horse

  6. Thanks to Paul and Alan for a very interesting discussion. It tallied with many findings from my 50 years of research.

    They obviously don’t need my posting to achieve their success but I thought that
    I’d share my experience as an example of what an ordinary bloke can do by following a long-term patient approach to applying certain principles to understanding what lies behind data.

    As a professional statistician I don’t look at data, I look through data to speculate what it means.

    The list of the main factors in my prediction equation does not included any of the four that Paul lists. But I do use the trainer and going factors in my models.

    In contrast, my number one focus is the horse as an individual. I try and quantify as many relevant features of each horse as I can, in order to develop an overall profile that I can then match with the race conditions as they arise.

    I focus on 2yos but the quantitative models also work well for 3yos and older horses as far as I can tell. I analyse and predict every horse for each 2yo race in the UK and for their 3yo races until the new turf season starts when the roundabout begins again. I enjoy news of how my babies make out as mature horses as my family love horses and our own in particular. But time is limited.

    The problem for me is the time required for analysing and making the predictions. The process is computer supported but not fully automated (called a Decision Support System) as there are judgements to be made in some cases. Although the overall model only comprises 7 factors, some of them comprise a large number of possible variables (over 50 in one case).

    The research has been conducted in a very disciplined way, meeting professional statistical standards. Thus some variables took years to quantify whilst the sample sizes reached adequate levels to conclude a non-random effect on horse performances. Only then would the new variable be added to the model. I think it was Mark Twain who said something like ‘the problem is not what you don’t know, it is what you really really know that ain’t so’.

    So I think its better to have a smaller/simpler model that one can be sure about rather than a complex one that includes falsehoods.

    Whilst I am on that point, it is obvious that UK racing market has not (yet?) detected some of the patterns that I have identified. This has been useful to me in providing personal confidence in my own figures when my odds line diverges from the market odds by a significant amount. Years ago I would have passed up a race if my economic value bet was over 10/1 in the market. Now I am not affected by this (although I still admit to double checking the judgements and calculations in such cases), for example, since September 1st (start of my current spreadsheet of race predictions – excel seems to grind to a halt when I have accumulated around 20,000 rows of data and equations) there have been 16 winners of 10/1 or greater.

    Also in this regard, my research and prediction performances significantly increased after I cancelled subscriptions to the racing channels and I stopped reading the racing journalists. I realised that to find value, I had to develop my own solid knowledge and avoid being seduced by ‘what they are saying’.

    So, the current (partly) computerised process has been in operation since 2003. Designed as a learning process, the first parameters included a minimum 30% margin before an economic bet was flagged. At the end of December 2003 the ROI was about 34%. So that looked as if I had a stable process. But the key thing was that when I checked each win probability category, the percentage of winners was about right. That is, the horses with 0.3 probability of winning, won about 30% of the time. The ‘end’ categories were not as good. More horses won in the 0-0.1 category than the model expected (debs etc), and fewer horses won in the >0.8 category than the model expected. In stats speak, the model was well calibrated and the ROI has steadily increased with each tried and tested new addition to the model.

    There is still a long way to go. There are several likely developments being being tested and I fill a 200 page A4 book with ideas, post mortems, diagnoses etc. each flat season from 2003 to the present.

    The principles that I have used include using only racecourse performance data (let the data speak), concepts from relevant known sciences (like physics), personal experience of horse behaviour and
    statistical thinking ideas. So that, a new variable is not introduced to the model equations unless it is based on actual reliable track racing, is backed up by scientific knowledge, makes horsey sense, and based upon adequate sample sizes to deal with uncertainty in data and measurements.

    I reached a stage when friends and colleagues suggested that I was making things too complex. But my confidence in the research principles got me through that and when a new finding would pull bits of the jigsaw of knowledge together and produced a simpler but more powerful prediction model, I was re-energised to continue.

    In the book ‘Outliers’, Malcolm Gladwell reckons that it takes about 10,000 hours of practice to achieve mastery in a field. I am not claiming ‘mastery’ but I can suggest that a lengthy apprenticeship can pay dividends provided that one is continually questioning, adapting and improving ones craft.

    1. Hi Denis

      Thanks for your comments. What an interesting response from you. Like yourself I prefer to play with statistics and finalise my resultant factor using probabilities. Hence the three selections in the method used on my site.

      Incidentally, the system I developed with over 90 factors would need several factors to change significantly in order to produce a change in the final ‘score’. Currently I use seven factors and am happy with the results. However, using such a low number means a small change in one of the factors can have a significant effect on the final ‘scores’

      Re your comment of Mark Twains – some times I think I have too much detail to hand. In the over twenty years I have been amusing myself I have written in excess of two hundred programs and have data on all aspects of UK racing. My main source of data is a file on the performance of favourites on which I can extract hundreds of statistics. In spite of this I still cannot predict a possible favourite winner. In spite of this, things that are not apocryphal but based on fact and experience stick in the mind and guide me away from pitfalls.

      Best regards to you and your methods. I spend most of my time computerising all the manual operations I have to go through in order to achieve my aims. Writing an A4 book is anathema to me.

  7. i like to look at improving horses if i can find an improving horse where it has improved in the last 2-3 races and looks a decent e/w bet 5/1+

    i also feel filed size is a factor i don’t bet in fileds bigger than 14 or smaller than 8
    i also won’t bet in soft or soft/heavy heavy ground unless it opposing a short priced fav e/w

    1. Michael

      Finding improving horses is always the key but the problem with that is everyone else is doing the same.

      Definitely reluctant to bet in heavy ground. All our statistics tell us that it dilutes our strike-rate and profit. We are much more selective in testing going if we bet at all.

      Of course, when backing in races with few runners it can mean the race is more tactical but that can be a positive or a negative. Lots more runners mean a better chance of an even/stronger pace but then traffic can become a problem. In sport visualisation is important. Former jockey Jimmy Lindley’s school of thought says each horse will have early speed, middle speed or finishing speed and it is how the jockey best uses that to determine how much prize money the horse will win!

      Michael, thanks for your thoughts!

      Best Regards

      Paul Moon
      The Racing Horse

  8. i also start by eliminating horses which i feel after analysis have no chance,,,
    then look at the horses that are left for value.

    1. Michael

      Definitely the best place to start! 100% agree. It is funny you should say that because we are currently writing a piece called ‘Removing The Clutter!’ If good enough was going to submit it to Race Advisor.


      Paul Moon
      The Racing Horse

  9. Hi Paul

    I’d just like to add some comments to one of your many insights in the Four Corner article.

    “It is hard to measure or be scientific about the importance of jockeyship, but they are the trainer’s connection to the horse and it is often their decisions that dictate the result and our profit and loss”

    One statistical quant approach to measuring jockey competence is to use the like-for-like idea by comparing jockeys who rode the same horse. That assumes everything else being equal, which it isn’t. As we know, a horses’s performance is affected by many other factors. But if you do this comparison for the same two jockeys who both ride many different horses , if there is a difference it will show up after accumulating enough data. If one eliminates cases where a horse had a bad run for other reasons, then the data (comparisons) should be a less variable and the difference in abilities should be clearer.

    If the comparison is made in terms of the ratings (lbs) achieved by the horse, the systematic difference between the two jockeys will be in lbs.

    So, if a horse appeared to run ‘true to form’ in two races and it ran an 85 for jockey A and an 80 for jockey B, just on that evidence A would be 5 lbs superior to B.

    If A and B rode another horse which had ‘clean’ runs and it ran 72 for A and 75 for B, then A-B would be -3, where as it was +5 in the first case.

    When this careful analysis is made and the differences are accumulated until the sample size is adequate, one produces interesting statistics. Jockeys that have low numbers of rides will be difficult or impossible to assess with reasonable accuracy.

    One has to adopt a scale for expressing the relative ratings for jockeys but this is arbitrary as one would only be interested in the lbs differences between the jockeys.

    Where this has been done to good statistical standards (sample sizes etc) its has been found that if the scale was set at 10 lbs for good professional jockeys, the ‘top’ jockeys (on this basis) will rate around 18 to 20. The remaining ‘very experienced’ jockeys could rate around 5lbs.

    As crazy as this may appear, this method has identified top performing jockeys before they were noticed by the ‘crowd’ and who later became champion jockeys.

    The analysis can be taken further. Having quantified A, one can use that to compare A and C, and D etc. It is possible that in exploring the differences, one finds a jockey that gets on so well with the horse, that the performances exceeds that expected from a ‘superior’ jockey such that the trainer sticks with that jockey. One can imagine other special cases like this. But the overall analyses can still provide the ‘general’ picture.

    There are stats techniques/packages that will quantify different people or teams that compete with each other. Used for American football, basket ball etc.

    Just a thought

  10. Oops

    Correction: Top jockeys on this scale would rate around 13lbs compared to the average ‘good’ jockey of 10lbs. (not the 18 to 20lbs difference given in my previous post – read the wrong column in a table!)

    My quantification of optimum going differences has only just started. The interest in that was triggered by noticing some of the trainer explanations to stewards inquiries into the reason for horses poor runs were given as ‘unsuited to the going’. Drops in rating for a limited number of these cases were around 9lbs, which is much greater than the 3lbs average difference between a top jockey and a good one. The sample of these ‘going’ cases needs to be greatly extended before useful estimates can be made.


  11. An interesting response Denis which highlights the stewards and the trainer. Your conclusion may well be correct as regards the jockey and the weight difference but as it is the trainer who chooses the jockey I think in the cases you illustrate that it finally is down to the trainer. (One of the Four Corners).

    The principle of the Four Corners is ideal. As I have used in excess of 90 factors previously I did increase them to in excess of 100 at one point but reverted back when I noticed that many differences in the extra factors had minimal effect on the resultant score. Keep the number of factors to the minimum and when they fail monitor the results and you will learn the reasons for failure and these results will aid in the improvement of the method you employ.

    Michael has a clarion cry that you should always monitor your results and this is the only way you will improve.
    I monitor my results daily and publish on my site the percentage strike rate I have achieved at each course and the value in points of the winners. This morning Market Rasen had a strike rate of 45% this evening it is 47% the reason being I had 6 winners from 7 races. Additionally I had a 25/1 winner. The maximum winner at this course previously was 13.96 points gross BSP. Why did this winner stand out? Its previous form was very good compared to the rest plus its best form was at Market Rasen.

    Two other horses had better form but finished third and fourth out of 4 finishers. A significant factor was the difference in weight between the two and the winner plus when Master Of The Hall last ran its weight was 16 pounds higher than todays run. I didn’t rate the horse highly but I did rate it sufficiently to make it a selection.
    The other two selections finished last and unseated the rider.

    I am pleased that I made it a selection but was unaware of the above. Had I been aware of the 16 lb drop and the soft ground conditions I would have made it a lone selection.

Back to top button