- This topic has 93 replies, 14 voices, and was last updated 19 years, 4 months ago by
.
- Posts
Hi everyone
There are 3 racing columns, that I know of, that specialize in producing trend based selections.
KEYTRENDS     by Ali Whitehouse-Jones  (RP Weekender)<br>Trendsetters    by Scott Moore         (Raceform Update)<br>Big Race Trend  by Craig Thake        (Racing Post)<br>=============================================
They all operate a steady process of elimination, eventually ending up with a Tip. I feel this methodology should be used as a tool and viewed alongside other forms of analysis. They all research their features with painstaking detail.
However they often manage to come up with 3 different Tips. There is obviously a good deal of subjectivity at play.
I have spoken to cormack15, and we would like to try to develop a key trends rating for every runner. In this way a horse can be place within a pecking order of suitability for any given race. Form and value selections with a high key trends rating can be backed with confidence.
The computer MUST take 90% of the strain of any ratings production.  Once the code has been written I can furnish it with it’s necessary data, via a comma delimited file.
Anyone interested in helping to develop the software, please post a reply.
The 3 previously mentioned articles will be a good source of reference, I can make copies…….
It all sounds a bit ambitious I know, but it will be very worthwhile.
Comments welcomed.
byefrom<br>carlisle
<br>(Edited by carlisle at 10:37 am on Nov. 27, 2006)<br>
(Edited by carlisle at 10:36 am on Jan. 7, 2007)
Very interested in the idea, though I only work with basic spreadsheets and functions.<br>I’d like to see the articles and maybe help where possible.
not a great fan of trends that are without a stated logic, but what are the returns on the aboves selections?
I give cursory attention to Craig, but he always seems to be way off…
You knew there would be a post like this :)
First step Carlisle would be to decide which trends were felt to be important.
There are a wide variety of ‘bits’ of data for each race. Which are worth collating?
Anyone?
Hi there
thanks for the replies.  I will scan the 3 articles and email them to you.  All the possibly useful trend angles must be considered, then the relevant ones ticked for any given race. In this way the program will have a generic nature.
I don’t pretend to know all the answers, at this stage. We will firstly have to isolated what the questions are.
Reading the source material, debating and giving some inspiration a chance to grow, will help.
Are there any Forum members that are aware of any books about Race Trends. Seem to recall something about race profiling, and what was "The Pattern" about?
byefrom<br>carlisle
"Intelligence is what you use, when you don’t know the answer"
(Edited by carlisle at 7:31 am on Nov. 27, 2006)
Hi Factorman
I am a self taught programmer, rather amateurish.
An interest in horse racing has driven me to learn about computers and my efforts can be a bit long, but effective.
Any help will be great.
thanks from<br>carlisle
ps clivex "give this piece a chance". Logic and commonsense must be paramount. I don’t know what their returns are. Probably only average. But I am not suggesting that we follow them.
(Edited by carlisle at 7:36 am on Nov. 27, 2006)
Thanks Flagship
we will write it in BASIC to begin with.  A VB end user application would be ace though. Some help with the design flow of the program would be appreciated.
Chip in when you can.
Early days yet.  If you want a copy of the previously mentioned articles PM me with your email address.
Cheers.
byefrom<br>carlisle
(Edited by carlisle at 9:51 am on Nov. 27, 2006)
Are there any Forum members that are aware of any books about Race Trends
Not quite the answer but Trainers statistical analysis by Timeform is good stuff. One trend i do like is the pattern of trainer’s form and behaviour.
Hi clivex
I agree, we should be able to consider how the trainers perform in the race in question.
Trainer/owners must have a favourite races.<br>For example Dandy Nicholls dreamed, as a lad, of winning the Ayr Gold Cup.
byefrom<br>carlisle
So carlisle/corm .. what do you want to end up with a rating (implying a chance) or a winner by process of elimination?
How will you decide what is important/meaningful and what is just a fluke?
If you are just looking at big races, you will need to come up with a way of assessing overseas trainers. You will also have to bear in mind that you will not get many betting opportunities .. so your stake will have to be high to make it worth your while.
Just a few thoughts.
An analysis i like in timeform is the comparison between the actual results the trainers achieves and the average mark of his horses. Its a rough guide to under/overperforming trainers
Applied monthly this could be even more interesting
There are plenty of holes in this but again trends like this are a factor rather than the sole criteria for selection
What I have done in the past is use excel spreadsheet macros to interrogate a database set up in excel (using the advanced data filter).  Even though I’ve used only basic criteria this has given me a good starting point to narrow a field down or highlight certain horses. It’s never been sophisticated enough to compile a tissue on its own but I’ve used it to influence (to varying degrees) the tissue I’ve come up with based on other factors.
<br>Here’s a wish list of items I’d want to know as a minimum about each race and horse who has contested a particular race (and I’d want ten years worth of data) –
Going<br>Number runners <br>Finishing position of the horse<br>SP<br>Racing Post Rating achieved in the race<br>Topspeed achieved in the race<br>Previous best RPR<br>Previous best Topspeed rating<br>Last three RPR’s (in same code) prior to race<br>Last 3 topspeed ratings (same code) prior to race<br>Trainer<br>Age<br>Weight carried<br>Draw (flat)<br>Previous best RPR over Course (returning null if not run over course)?<br>Previous best RPR over Distance (as above)?<br>Previous best RPR on going (as above)?<br>Days since last ran<br>Number previous runs that season<br>Number previous runs in total<br>Race completion percentage (jumps)
Tall order? Maybe, I don’t know what you can do Carlisle.
DJ – it would be interesting (and novel) if we could arrive at a ‘trend rating’ – I think that is an interesting idea of Carlisle’s and one we should strive toward. Once we had it we could work out what to do with it!
Once we had the data then what I would be looking to do would be to run a large number of queries, returning a ‘result against expectation’ figure on which to judge each criteria or set of criteria. If a particular criteria (say 5 year olds) had returned results better than their expectation (based on odds) then it would be highlighted in some way. You could run sophisticated queries with more than one variable (say, for example, examining 5 year olds carrying >11 stone starting at <10/1).
When I’ve done similar types of analysis in the past I look for a very large results/expectation ratio (expressed in Level stakes profit on turnover %) to ensure stats are exteremly favourable thus countering teh high level of unreliability from a statistical viewpoint. That is a bit rough ‘n’ ready and some stats gurus may be able to come up with a better way of putting a figure on the reliability of the findings from a a statistical viewpoint.
More on stats later. (if you can stand it!)<br>
Hi cormack
you kick on, I won’t nod off.  Some of my best friends are numbers.
It’s good that we can approach this from different angles. Personally I think most of our work will have been done by <br>the 3 Trendmeisters.
Craig Thake writes this about the Hennessy.
Verdict<br>======<br>"A race that strongly favours lightly raced second season chasers with a good win-to-runs ratio, who are proven at class 1 or 2 level."
The race Trends have given rise to real feel for the race.
However instead giving out zeros and ones (True or False) we can issue a best match score, either 10 to 0 or 5 to 0. This will stretch out the range of the ratings.
The follow example doesn’t seem that useful.<br>www.betfairpromo.com/betfair_million/analysis1.asp
We can do better.
byefrom<br>carlisle
<br>
(Edited by carlisle at 11:16 pm on Nov. 27, 2006)
Hi cormack
I see what you mean. Running retrospective analysis would be very very interesting. Mind you I am glad you are going to do it.
"keep smiling"
byefrom<br>carlisle
<br>
(Edited by carlisle at 11:14 pm on Nov. 27, 2006)
Thats a big wish list Cormack but it would be hard to know which if any of the factors listed to leave out.
Likelihood ratio’s jump to mind, with a big enough database to ensure stability.<br>Thats one way of producing a probability figure.
What I feel the example on Betfair fails to take account of is the stats on teh losers. It’s all very well saying 8/10 who did such-and-such beforehand went on to win the race. What if 8/10 of the also rans also did such-and-such beforehand? Then the 8/10 stat is meaningless.
We’re looking for trends which are meaningful in a given race rather than generic trends? Right? Or both?<br>
- You must be logged in to reply to this topic.