• Hi Guest Just in case you were not aware I wanted to highlight that you can now get a free 7 day trial of Horseracebase here.
    We have a lot of members who are existing users of Horseracebase so help is always available if needed, as well as dedicated section of the fourm here.
    Best Wishes
    AR

Where Do I Begin?

With any rating speed or collateral i have found the best use is to ask "what relevance to today's race" ? I also think that S student shares a good thought that sometimes one can be used to validate or degrade the other.
 
student said:
Hi Giusepee
I’m a newbie to this forum and as Mick will confirm I have difficulty in writing short postings. I’ll try and make my points simple and clear, but apologise in advance.
I’ve researched and used my own speed ratings since the early 60s. Not many punters used speed then and one could make profits in 2yo races and I got restricted and laid off by bookies on Merseyside.
With speed being considered by more people now, I use them to inform my collateral ratings and vice versa as just one way of identifying ‘false’ run races/ratings. For example, a low speed rating (compared to its collateral rating – on same scale) can result from a slow early pace or a ‘too fast’ early pace where a few horses battled for the early lead and quickly reached their exhaustion point & slowing to allow an ‘average’ hold up horse to continue past them at one pace in ff. The OH will have to uprate that winner probably by about 6lbs because its beaten better horses (on paper)– false rating?
If one calculates a going allowance based on all races on the card, a low speed rating could result from runners in the straight running into a head wind, whilst runners down the back straight get wind assisted for a few furlongs. Then one can calculate separate allowances for each wind strength and direction, etc etc.
My impression over the years is that most punters who base bets solely on speed ratings may make relatively small profits mainly by getting the occasional longshot winner (including specialists Phil Bull and Alex Bird).
Mike and Dave have given you good advice, so I’ll finish my bit on speed.
I was interested in was your ability to get good strike rate but not profits with HRB ratings, which I’ve no experience with. That’s a positive achievement in developing your methods. I’m guessing that the lack of profit could be the result of using ideas (good recent form etc) that the rest of the market uses. Any successful investor in anything has to be sufficiently ‘contrarian’ from the rest of the investors, to make a profit. For example, the market usually likes a winner last time out, they don’t like a bad recent run. But good trainers will take that horse home, work out any problem and put it right as quickly as possible. I think that’s why you hear a lot of good punters say they can’t get a certain trainer ‘right’; trainers like Clive Brittain, Mark Johnston.

Investigating a ‘bad’ run lto may reveal that it was excusable or much better than it seems on paper. That run could put the market away and offer value odds to you if they haven’t looked at the right details. All you have to do is find a few details that matter that don’t get the headlines. There are lots. One way is to listen to what the best jocks talk about; Johnny Murtagh and Dettori. Don’t judge them by their tips, its what they say about why horses run better or worse. They know about horses.
In summary, I suggest that you don’t give up on the use of collateral ratings. They are more consistent than speed ratings for reasons given above. You may like to consider balancing your research time between the two types of rating and linking your findings together that make horse sense.
Very best wishes
Click to expand...

Hi S student

Thank you for your reply. I have been trying to think of various angles that could help me with my own horse race handicapping system. Speed Ratings are just one of the many things that can be considered. I am currently creating a SQL Server data warehouse with horse race information from historical races. The primary aim is to build a data cube with these races and results and then carry out some Data Mining. I have built the data warehouse, I just need to build the cube and data mining template. I will then base my selections from the data mining results.

Regards
Giuseppe
 
mick said:
Hi S student keep on keeping them long. :)
Click to expand...
I'm sorry this is so long. You want to see the chunks that I've edited out; no you don't. But here goes.......

Hi Mick, how’s things. Long time no news?

I like to find contrarian ideas. Not just in horseracing but in life in general. That way I can practice and develop that skill every waking moment (that’s 4.30am till 11.30pm) by asking questions of myself and others.

I reckon that there are different kinds of intelligence (soccer players see the spaces and the runs of their team-mates, or not as the case may be) and thinking. Each needs practice to develop the skills, and for me, developing contrarian thinking in horseracing took years.
But that was odd, because I’d been a successful business consultant and consultancy is about delivering better results than the client’s current system can. In that case, I was expected to be contrarian? And yet I hadn’t been achieving the same success in horseracing predictions.
Adopting the consulting analysis and thinking skills to horseracing (main one is not to be trapped by the data as presented by others – as in the form book) the level of ROI significantly increased.

So that suggests to me that we might be able to improve our skills and knowledge in one area by importing our skills from other areas. By looking at things from different points of view (handicap and speed ratings) I have used ideas from consulting analyses in racing, and vice versa.
In my last posting I tried to illustrate that it’s possible for a winner to beat better (on paper) horses without improving its performance. It won by running even pace whilst the ‘better’ horses ran too fast early and ‘collapsed’. The idea that the winner quickened (pace) at the end, is an optical illusion. But for other reasons the official handicapper could (and often does in these cases) raise the OR for the winner in future races. Perhaps punters should include the OH in their prayers at night?

But to focus on contrarian thinking in racing and escaping from traps in ‘official’ data, I’d like to take a different case from the ‘pace’ one above. I take a number of factors into account to measure a horse’s capability. Many of them are interrelated but I need to simplify the next example (running the ‘VDW’ risk) to keep the wordage down.

The race decs will show the distance of the race to be run. It’s there in black and white. But not all races at that distance have the same challenges on the horse. Think of the energy demands for example. My aim is to produce the most accurate predicted rating for a horse in its next race when I know the distance and likely going. I do not focus on the exact distance and going, but think in terms of ‘about’ those conditions. With movement of running rails affecting race distance, and iffy forecast going, ‘about’ isn’t too farfetched? That avoids me trying to seek exact matches for race conditions (between its past performances and ‘today’s’ race) which I suspect much of the market does – I want to be contrarian to the market if I possibly can be, without making bad decisions just for the sake of it.

I start with the pedigree and get an idea of the sort of distance required for the horse to achieve its maximum rating. Taking all the other conditions into account, I then check, race by race, what the race distances tells me about the horse and trainer’s likely plans and changing previous assessments if need be. I must point out that I still specialise in younger horses; 3yo hcps now.

In a simple historical case to hand, my ‘pedigree’ distance suggested about 8f.
Its deb distance was about a ‘net’ (taking other factors into account) 1f shorter; OK, quite common for the deb race to be ‘under distance’.
Race 2, distance as for deb race, and its running strongly suggested that further would be needed for a ‘true’ PB (see the first example on pace above). Although it is noted that the horse has not been moved up yet.
Race 3, distance is still about a ‘net’ 1f shorter, but running commets confirmed pedigree distance for its potential PB.
Race 4, same as for race 3, but horse achieved a true PB todate. Hoping now that the market thinks that it’s a 7f horse?
Race 5, entered in race at ‘net’ 2f short of its pedigree distance; ALERT for intent nto. And when results and running comments come through and say “ridden in rear from the off” giving it no chance; ridden to lose? Confirms the alert.
Race 6, entered at pedigree distance (and confirmed by some runnings) for the first time after being UNEXPOSED in first 5 races. Extra headgear for first time suggesting +ve intent even though it probably will not help. (My time around horses tells me that unreported changes to bits and other tack are far more important than those changes reported). Longshot winner.

Notice that I am positive about the expected value in race 6, despite the horse running over a new distance, longer than anything before. For years I used to heed the gospel “only bet a horse that has proved itself on the race conditions”. I can understand big stake punters not wanting to take too many chances but I find value in horses meeting new conditions. In this case, several pointers were ‘telling’ me that further would be better in terms of future rating (based on proven historical data analysis); also that the market would probably think it was a 7f horse and its ‘bad’ run lto would have further deterred market support, hence an expected long price (again based on historical data). Expected value (using the mathematical terms) is a function of win probability multiplied by expected odds. So in both cases I am contrary to the market but based on sound evidence.

I am hyper active with many interests and have to specialise in each due to limited time. But it is amazing to me how often an idea in one interest is triggered by another.

Just a few thoughts. Best wishes
 
One thing to take from some of the posts on this thread is that some members put much effort into creating there own speed or collateral ratings.I would suggest that although there are plenty of frustrations along the way this is always time well spent.During the process your knowledge of many aspects of form will increase and when you are finally happy with what you have created and your confidence in same grows then you will also be fully aware of all that is involved and exactly why it is working, something which cannot always be claimed by those who use the commercially available ratings.

The other important aspect which standalone justifies your efforts is that sometimes your ratings being unique will disagree with those used by the majority and this can buy you a valuable market edge.
 
Last edited:
I agree with your points Mick re developing own ratings. I do that but also use the knowledge gained to know when the race result/rating is worth retaining or is false and should be discarding in updating your assessment of the horse. In some cases discarding 'false' races from a horse's lifetime PPs can reveal a hidden inprovement ratings trend that my simple eyeballing the raw data didn't detect.
 
S student most backers tend to use ratings as a numerical means of deciding which horse (s) are well treated in a race yet to be run and this can be important,but i find they also play a most useful part when attempting to profile the former. :)
 
Good morning all

I guess I've drifted a bit from the original motive for me to post to this forum. Speed re Gioseppe requests about where he should start.

My own understanding now is that 'speed' is used for measuring end to end or 'final time' speed. In track & field we think of that as endurance; to maintain 'secs/furlong' for most of the way. We know that in the majority of races most horses are slowing down towards the end. I think of 'secs/flng' over a short or part of a race as 'pace', using speed and pace in that way when I post.

I don't use sectional times. In fact I queried the call for these extra data when UK punters started to talk about them. My main question was 'what are you going to do with them when they come'. The other point was 'what will they tell you that the existing comments in running don't'. When the 'new' philips VRs arrived I experimented with taking mid race times etc. The first shock was that even in a 2yo 5f race, the short term pace did vary in some races. They didn't just dash out of the gate and maintain a high pace all the way. Of course, for some green 2yos in debs, the penny would drop mid race with sig change of pace. But that was usually obvious from eyeballing and confirmed by Raceform race readers.

This change of pace would not coincide with furlong markers, and landmarks were not always where you wanted them to be. As I improved my predicted ratings with my owm measures that were reliable I have not returned to the sectionals data uses. I have read Simon Rowlands timeform articles and can see how some punters could make consistent rating adjudtments calcs from sectionals; even though I sometimes recognise a limitation of statistical science concepts. I think some folk would benefit from having at least an intro idea of these; such as the RPost/Raceform rating staff who often try to fit them round the previous rating of a specific horse rather than find a best fit with a group of finishers (having omitted the false ones?,Lol).

I find that both speed and pace provide useful interpretations of results data. I sometimes watch racing on TV and noticed a really good winner that went away in ff. Couldn't understand its good price. Not a race that I work. Looking back, it had led and lost in ff of a strongly run race (high speed rating?) lto. But in the race I watched, it was a slow early pace and that seemed to make the difference for this horse.

So, I'm interested in Giuseppe's approach of building a data warehouse and cube to produce his ratings; will it's initial use be for research purposes or has he already got some models that he can run in his new set up. I bow to our professional computing friends; they would laugh, or cry, at my ms access/excel set up.

But that triggers a thought. I am writing up my consulting problem solving cases from time to time, and recently tried to put them in order of extra value prduced for the client; high to low. This was to see which tools and approaches figured in the best ones. Apparently, I made benefits worth £1m or more in 6 cases. The shock I got was that computers were not required in any of these. The top 6 were solved by thinking.

best wishes to all
 
Good morning S student Re your above post i really liked the final para.I think your findings have great significance for many current day horse race backers.Computers and automation can be a wonderful aid but i have found no quick fix final solutions as part of turning a profit.I do see increasing numbers of backers seeking this and perhaps this alone is a good enough reason to take the opposite view.?

I have subscribed to RI for many years and like it ,but has it made me a more profitable punter NO, what it does do is make the process more enjoyable.Racing data bases and search engines now enable a fast look at the most complex combinations and aspects of the past but i wonder if this easy find has now negated any edge and the future one will be the Pen and A4.?

I like the idea of long term some things going full cycle you can see this in racing and i have found it within my own MO.Who knows maybe we will even revert to having Bookmakers who will take a bet.!
 
Last edited:
I really enjoy reading your responses to this discussion. I have attempted my first go at using Data Mining from the horse races dating back to the 25th July 2018.

The first issue i have come across whilst trying to work out the probability of a horse winning a race is the volume of variables i have tried to use. With more race variables for each horse in a race, the more complex it is to identify any potential trends.

I am going to streamline things and maybe start off with 4-5 variables, such as Official Rating, Racing Post Rating, Going Wins etc and see how this does.

If anyone has any thoughts on the key variables i could focus on, I am open and grateful for any suggestions.

I agree with Mick that, for those who have followed and done their own horse race handicapping for a long time, that the traditional methods of pen, paper and knowledge may well be more profitable than using other methods such as data mining.

Thanks for your input everyone!!
 
giuseppe_esq giuseppe_esq as you have discovered there are many variables involved but to answer your question i would prioritize six Class Weight Fitness - Readiness and Distance Course Going.The other advice i would offer is to specialize this could be race type ,age or whatever suits. My own choice is All - Aged Hcaps of 5 - 10fur .With so much racing these days i feel my time is best spent attempting to learn more about less.
 
mick said:
giuseppe_esq giuseppe_esq as you have discovered there are many variables involved but to answer your question i would prioritize six Class Weight Fitness - Readiness and Distance Course Going.The other advice i would offer is to specialize this could be race type ,age or whatever suits. My own choice is All - Aged Hcaps of 5 - 10fur .With so much racing these days i feel my time is best spent attempting to learn more about less.
Click to expand...

Thanks. I am not going to invest too much time working out my own speed figures initially. Instead i will use the Racing Post Ratings.

In response to the 6 variables you suggested:

* For class, distance, course, going - do you mean focusing on horses that have previously won a race at the same class, distance, course and going as the race being entered? This could be deemed as a silly question, but there are other attributes available, such as percentage won in the class, distance, going etc.

* Weight - I havent really focused too much on weight, as some authors have argued whether weight actually matters. However, i have created some systems before that only focus on the top 4 in the weights in any given race.

* Fitness - Readiness - How do you assess/calculate this? How many days since a horse last ran?

I have only tried to create my own horse race handicapping system for a few months, so i am still new to this and appreciate i have a lot to learn.
 
giuseppe_esq giuseppe_esq Within the six variables i suggested you will find the reasons why the majority of horses win or lose.If your talking only ratings components then my own which are collateral based would have just two main ingredients Class and Weight and of course the OR is part of both.My own are framed to mirror those of the BHA Official Handicappers with the hope that i can sometimes disagree with them.If you look at the BHA site there is plenty of info Re how they think and work which should give you much food for thought.
 
Last edited:
For some reason i missed post #23 on this thread and would suggest that all backers could learn and improve by reading and thinking about the contents.As the author states contrarian thinking will buy you a market edge and this is an important aspect for anyone serious about achieving long term profit but to feel able to disagree with the majority for the right reasons is no easy ask ,and in this respect i liked S student final sentence " But it is amazing to me how often an idea in one interest is triggered by another"

His thinking is to apply knowledge and experience of non racing activities to how you currently view the racing task,but i also find it works within its self because its never enough to develop a profitable strategy you also need a good understanding of why it is working and what part each component plays,and sometimes when reviewing this an aspect of form seen or used in a slightly different context can result in just a small change bringing a significant improvement.

Being contrarian with yourself. ? :confused:
 
Last edited:
Hi Mick

re contrarian with yourself. reviewing this an aspect of form seen or used in a slightly different context can result in just a small change bringing a significant improvement.

Big point. When this has happened to me, I have then been able to 'see' other examples of the new revelation that I had missed in previous working form data. A need to stay alert and maintain curiosity at all times? Last phrase was written down last night in some notes I was making about judging evidence of UFO's!
 
S student I agree Alertness and Curiosity should both be a constant part of. :) Although on balance i am comfortable with the way i work ,i still constantly challenge my own thinking and also remain receptive to that of others.
 
Last edited:
giuseppe_esq said:
I am going to streamline things and maybe start off with 4-5 variables, such as Official Rating, Racing Post Rating, Going Wins etc and see how this does.

If anyone has any thoughts on the key variables i could focus on, I am open and grateful for any suggestions.
Click to expand...

Hi Guiseppe

I think your change of tack is sound; fewer variables to start with.

But a story may provide you with extra ideas for doing that. A very successful consultant was hired by an american petro company to identify the best places to build some extra petrol stations. Scanning the report of a previous marketing consultancy he took the opposite (contrarian) approach. They has asked thousands of drivers to list the most important factors in choosing a station to use. It gave over 20 factors and the computerised stats package gave a long 'best fit' equation for data from their existing statitons. It was a bad predictor of profit per station and gave no insights about how these decisions are made (we usually don't try to solve 20 factor equations whilst we drive along?).

The method of the new consultant was to always start with trying to find the one most important factor, and see how much understanding you can get from that. Then try to explain any anomolies in the fit of data to this factor, and thereby find the 2nd most important factor, etc.

So his survey asked drivers to give the most important factor that led to the decision to stop at a station. This clearly showed that as 'the expected added time to the journey in manouvering in, being served, and rejoining the traffic flow'. When he compared this with the marketeers list, most of their 20 factors were just detailed factors that affected 'added time to the journey'. In other words they had missed the wood for trees? Essentially he had got into the head of the driver needing petrol soon. (in usa at the time, stations had no price differences, only sold petrol, attendents serviced you, cleaned the windscreen etc).

Because the moment this one factor became clear, the company got a massive financial advantage over their competitors. The gospel at that time was that you couldn't make any money from building a second station at a cross roads (max traffic flows) if there was an existing station already there on one of the corners (easiest manouver routes are corners - you by-pass the traffic lights). But in that case, there are 3 other corners and the traffic flow through a diagonal opposite corner to one with a station already, will not loose much custom to that or any other one.

The consultant didn't have to write a report. He just presented a drawing of a cross-road and drew the 16 possible routes through it (including u-turns - desparation?) with an existing station at one corner, and wrote the key driver's decision aim below it; to minimise added time to the journey, and he stood back.

The board of directors stared at the flip chart; few minutes silence then some pointing and chattering at the opposite corner on the drawing. When the logistics and financial directors pointed out that they could buy sites on all the corners opposite an existing services one cheaply because no other petrol company would think of trying to buy them (industry gospel) and push the price up, there was uproar at the idea of $ signs and big bonuses.

Now I know this is a different context to horseracing. But I can vouch for the benefits of using the principles illustrated above; the contrarion approach to starting by throwing masses of data and computing at a problem. In the race distance case above I was trying to find a way of 'getting into the head of the trainer/connection' about entering a horse to have a chance of winning at a good price.

And as used in that case, I am not saying use exactly the principle, but something like/about that sort of idea.

Hope this helps at some stage in your investigations.
 
Last edited:

Thanks for your insight. Really appreciate it.

One of the issues I am having is the wide range of data values for some of the attributes.

I will try creating some buckets with ranges to try to simplify things. For example Official Rating between 61-70 etc

I have built a data mining structure, and thought making the predictions would be easy. Unfortunately it isn't and will need fine tuning along the way.
 
You must log in or register to reply here.
Back
Top