How to apply trend analysis to horse racing?

[Dumb Racing Junkie]

I have come to this forum hoping to learn how to apply financial trading techniques to gambling on horse racing.

Specifically, I wanted to learn how to identify and rate 'hot' trainers and jockeys using their past results/profitability statistics. I wanted to place much more weighting on their current short-term results using something similar to regressinon analysis.

Can anyone point me in the right direction on how I might go about getting started learning how to do this?

I know for a fact that very successful gamblers are using this as the basis of their form analysis but I am not exactly sure how they do it.

 

[bottomdollar]

I am not sure you're in the right place for this but I know that you can get software such as Bet Angel which monitors trends for race horse odds... they call it sports trading rather than sports betting!

 

[Dumb Racing Junkie]

Yeah I am aware of Bet Angel but that is not what I am talking about.

I want to learn about specific techniques to analyze and interpret past data.

Techniques they use in the stock market.

Techniques they use to try and predict what is going to happen in the future based on past performance.

Can anyone help, or know where I may be able to get help with this?

 

[everyonerich]

ermm.. what about put all your horse data on excel then create a chart? then use TA to predict the trend or something.. lulz

 

[Rossini]

I wonder if you could successfully use TA as an aid to bet on horse racing, would it be more successful than TA on financial instruments, because less people are doing it? Or are the odds stacked so much in favour of the house that it wouldn't make any difference?

 

[shadowninja]

If you can create some sort of graph then I would think that looking at very basic EMA crossovers could work for giving a simple trend direction. You couldn't trade off support and resistance... unless everyone reads the same charts you do... *scratches his chin thoughtfully*

 

[megamuel]

Horse Racing Systems | Betting Tips

 

[Calinor]

Interesting thought Although I have no idea whether this would actually have a chance of winning, I guess intuitively there are some similarities. There are two different approaches: "'hot' trainers and jockeys using their past results/profitability statistics" in my opinion doesn't lend itself too easily to statistical techniques. In horse racing, you have trainers, jockeys and the horse, and also any weight if it is a handicap, (maybe even track condition). So if you change any one of these, it isn't the same product/asset any more. So applying statistical techniques to find patterns in things that are fundamentally different has problems. Also this isn't the market. So techniques used on these forums applied to markets will not necessarily work for this.

The other approach would be to consider the market as the actual odds. Any quoted odd could be converted to a number. Odds of 2-1 should convert to 1/(2+1)= 0.333, odds of 4-1=1/(1+4)=0.2 etc. These odds are changing in time, and so you essentially have a number that is moving through time. Since this is a market, and buying these odds has an effect of pushing that converted number up, you could try to use some technical analysis on this. If there is then some inside knowledge or trend, it may show up somehow. But for this you would need quite a few changes in the odds to see a pattern, and even then, I don't think it would work in quite the same way. If I do all my analysis and find that there is a horse with odds of 10-1 and my analysis says that those odds will go up/down, then suppose I take those odds, and it moves to 4-1. Have I made profit? In the markets you would.

There seem to be some people with excellent knowledge of applying statistical and technical anlaysis here. But you have to give them something to work with, i.e. a graph of numbers representing a market. And you would also have to decide which of the two approaches I mentioned above you really want to go for.

 

[case777]

You need to find a horse where the odds offered are longer than the "true" odds of it actually winning.

So if the odds against a horse were 3-1, but you felt these were wrong and it is really is a 2-1 horse (ie, if it ran 3 races, it would win one of them and lose two) then you should bet on that horse. For every £1 you bet, you will lose £1 every 2 times out of three, but gain £3 every once out of three times. (1*£3)-(2*£1) = £1 profit.

To find the "true odds" you can use Beyer speeds or any other data on the horse. Use the data in a system to estimate odds on a horse, and see if it has a positive expectancy over the long term.

 

[Ferru]

Have a look at the databases at Adrian Massey's Horse Racing Ratings. You can do searches to see how much profit or loss you would have made had you bet on, for example, all favourites under 4.0 which won their previous race.

What you will find is that the Betfair odds are an extremely accurate reflection of the horse's true price, and therefore finding an edge over the market is very difficult.

Have you considered trading on the betting exchanges? You might find this trader's blog of interest: Horse Racing Trader [ £0 - £150,000 Challenge ]

I hope this helps.

Jeff

Yeah I am aware of Bet Angel but that is not what I am talking about.

I want to learn about specific techniques to analyze and interpret past data.

Techniques they use in the stock market.

Techniques they use to try and predict what is going to happen in the future based on past performance.

Can anyone help, or know where I may be able to get help with this?

 

[pseudo straddle]

I know for a fact that very successful gamblers are using this as the basis of their form analysis but I am not exactly sure how they do it.

Perhaps, DRJ, you could enlighten us as to the source of these ' facts '.

 

[Bovreuil]

What you will find is that the Betfair odds are an extremely accurate reflection of the horse's true price, and therefore finding an edge over the market is very difficult.

Really? I know everyone says this, but I don't find this in my own data.

Which "betfair odds" do you mean? 4 hours before the race / 1 hour before / 15 mins before / at the official start time / the last price matched before the actual off?
These prices vary so wildly that the "extremely accurate" claim cannot be made for all of them!

It's not hard to find horses with a 20-30% difference between the highest and lowest price traded before the start. With those kind of movements, finding an edge of a percent or two is not difficult.

(It's a bit different for predictable football or tennis matches where the prices don't move more than a couple of percent, and there really is a high correlation between price and result.)

 

[Ferru]

Hi Bovreuil

I see where you are coming from, although maybe the Betfair odds represent the true odds given the available information. For example, if a horse drifts it may be because it has emerged that the horse is carrying an injury or is looking nervous, but the earlier odds accurately reflected the information available at the time.

The stats on this website also suggest that the market is extremely accurate:

Probability Theory

I do wonder, however, whether the odds become inaccurate when people see a steamer or a drifter and jump onto the bandwagon, thinking there must be insider knowledge at play, causing the price to fall or rise to a level that doesn't represent its true price. Betfair's own stats suggest this is so: http://www.betfairpromo.com/tactemails/camb1a.htm#How

Regards

Jeff

Really? I know everyone says this, but I don't find this in my own data.

Which "betfair odds" do you mean? 4 hours before the race / 1 hour before / 15 mins before / at the official start time / the last price matched before the actual off?
These prices vary so wildly that the "extremely accurate" claim cannot be made for all of them!

It's not hard to find horses with a 20-30% difference between the highest and lowest price traded before the start. With those kind of movements, finding an edge of a percent or two is not difficult.

(It's a bit different for predictable football or tennis matches where the prices don't move more than a couple of percent, and there really is a high correlation between price and result.)

 

[Ferru]

Hi Case777

I imagine very few people can get more accurate odds that bookmakers' odds compilers, who have an encyclopedic knowledge of racing and contacts in every major stable in the country.

Regards

Jeff

You need to find a horse where the odds offered are longer than the "true" odds of it actually winning.

So if the odds against a horse were 3-1, but you felt these were wrong and it is really is a 2-1 horse (ie, if it ran 3 races, it would win one of them and lose two) then you should bet on that horse. For every £1 you bet, you will lose £1 every 2 times out of three, but gain £3 every once out of three times. (1*£3)-(2*£1) = £1 profit.

To find the "true odds" you can use Beyer speeds or any other data on the horse. Use the data in a system to estimate odds on a horse, and see if it has a positive expectancy over the long term.

 

[Dumb Racing Junkie]

Perhaps, DRJ, you could enlighten us as to the source of these ' facts '.

Hi again everyone,

Thanks for the interest in my question.
I work for a betting agency. My passion is horse racing. I have access to a database that can run just about every test imaginable over past results.

I am not asking for help with handicapping. I have an intimate knowledge of ratings, market efficency, speed maps, formulating prices, overlays etc...All of this. I have read Beyer. I know Adrian Massey.

 

[Dumb Racing Junkie]

Lets look at some stats similar to something you would find on Adriam Massey's site and then, hopefully, generate some discussion on the specific part of the puzzle I want help with.

Here are some statistics for my favourite trainer:

Classification Sts 1st 2nd 3rd Win% Plc% Windx
====================================================================
01/08/04 to 31/07/05 1081 213 158 134 19.7 46.7 75.9
01/08/05 to 31/07/06 914 199 157 102 21.8 50.1 86.1
01/08/06 to 31/07/07 745 151 124 87 20.3 48.6 92.1
01/08/07 to 31/07/08 477 107 72 59 22.4 49.9 86.7
01/08/08 to 04/03/09 376 108 51 42 28.7 53.5 104.8
Total 3593 778 562 424 21.7 49.1 86.3
Last 100 100 30 15 12 30.0 57.0 114.4

However, strike-rates are not at all meaningful for our purpose - $$$.

More important: Assuming $1 was invested on each runner.

Last 50 runners gives 15 winners @ 30% and a return on investemt of 118.8%
Last 200 runners gives 61 winners @ 30.5% and a return on investment of 102.8%

An even better way to look at this: The A2E Measure. The A2E (or Actual-to-Winning) measure takes into account the ability of the horses trained by the trainer in question. The Expected number of wins is calculated using the market price of their respective horses and compared to the actual number of winners ridden or trained, with 100 being nuteral.

Runnners Winners A2E

Last 12 months: 634 153 109.6
Last 24 months: 1144 262 110.4

What sort of analysis would be suitable to use to predict what is going to happen in the future if I am using something similar to an A2E rating - but with MORE WEIGHTING placed on recent results. This trainer is probably not a good example because they are very consistent but other trainers run very hot and cold (I am not going to speculate why this is so). A trainer will be very hot at one carnival and train many winners then be cold at the next and not get any winners. Their overall stats will not reveal this. The only way is to place a lot of emphasis on their recent results.

I hope this interests somebody. Any advice on what type of analysis would be suitable to combine with my stats to pick up on a trainer when they are starting to get hot or drop them when they are going cold?