- This topic has 93 replies, 14 voices, and was last updated 19 years, 4 months ago by
.
- Posts
Dave – A week is hardly a long time surely. Anyhow – we’ve made good progress and we’re over the first hurdle which iwas getting of our behinds and getting on with it! We’ll keep you updated.
Carlisle – great stuff. Ken has some interesting ideas on how to work with it to begin the process of obtaining a computer-generated-ish basis for a trend rating. I’ve been doing some stuff on the data also just to give ken something to work with. To manually extract data for about ten horses takes about half an hour so, on average, it’ll take five to seven hours-ish per race to get ten years worth of trends. At one race per week with a few of us doing it we could get the data up and running in no time. After that it’s minimal effort to update each year. <br>But, before we do that, we gotta see whether it is actually worth diggin gthe data out. Be fantastic if your data source can do it quicker/easier.<br>
- Total Posts 17716
<br>To manually extract data for about ten horses takes about half an hour so, on average, it’ll take five to seven hours-ish per race to get ten years worth of trends.
Jesus, Corm! <br>How much are you planning to put on?:biggrin:
Yes RH but between three/four of us that’s only just over an hour. And once it’s done it only takes half an hour per race each season to update.
Of course if someone could come up with an auto-extract script…
Hi cormack
that sounds like a valuable, ongoing, resource.  I know, given a chance, we will come up with some good ideas.
Cormack how do you envisage the raw data to look………
Just checking.
byefrom<br>carlisle
Hi cormack
I sent it to you on Sunday.  I have just resent it.
cheers from<br>carlisle<br>
Thanks Carlisle – must have been in amongst the millions of dodgy emails I get and deleted by accident.
We’re getting there! (albeit at Quixall Crosset pace!)
Hi cormack
we are having a bit of a lull at the moment. I am sure we have est. some good foundations.
It’s health to approach this subject from 2 different angles. This will give a sense of balance and enable us to have some meaningful discussions.
I am also determined to come up with a worthwhile Race Trend Rating.
byefornowfrom<br>carlisle
Not really a lull Carlisle – not at this end anyway as I’m beavering away at the King George data. I’ll pm you later with a suggestion for some data mining you could do to speed the process up.
- Total Posts 17716
Forgive me for being a little cynical, but I fail to see where all this is going get you?<br> As you are no doubt aware, each horserace is a unique event, and can differ from year to year depending on any number of factors.<br> It is all well and good knowing, say,that a race is normally won by low weighted horses, or that trainer X invariably does well at certain courses, but, in no way, can it account for all the circumstances of the race being studied. Chip away as much as you like at past trends, at the end of the day the decision to bet surely has to be based on the unique circumstances of this day’s race?<br>I accept that past results can give some sort of guide, what I fail to see is how they can possibly embrace what may happen in this race or the future? You yourselves must have seen where top weights don’t win the Hennessy or Michael Stoute doesn’t win the Lincoln, but where does this approach allow for the Trabolgans or the Stream Of Golds?<br> Maybe my reticence is a result of my not being a ‘stats’ punter, but surely this exercise has been done a number of times previously, by such notables as Mathematician for instance, but I have yet to be convinced that it is worth the effort you guys seem prepared to expend?<br> Just my view, maybe some others see it differently?<br>
I share some of Reet Hard’s concerns. It’s a commendable exercise in that it is happening in the public domain so to speak and it may throw up all manner of interesting ideas, just as say Nasa projects produce breakthroughs in other areas.
My take on trends has always been that they can be a short cut to spotting something that is happening in the real physical world. However, if, after investigation, there appears to be no real reason for a trend, then that trend is nothing more than a statistical anomaly.
My other concern is that trends tend to lead to arbitrary cut-offs. You might establish that horses carrying above 11-0 have a poor record in a certain race and you might be satisfied that there is a reason for this, perhaps that it is often run on heavy ground and is an extreme test of stamina. But does that mean you should throw out any horse carrying 11-0 and include any carrying 10-13?
In other words, I think trends can be used to help point towards the characteristics that can help a horse win a certain race, but that is the start, not the end of the analysis. Trends that are worthless are those that state a certain trainer never wins a certain race or that horses above a certain price can never win it.
Good points RH and Aranalde.
RH – yes, it is a crucial aspect of trends and stats to consider that the upcoming race doesn’t form part of the sample on which the analysis is undertaken. What we will produce will be an analysis which highlights the statistical significance of certain bits of data and statistically predict the likelihood of aparticular trend continuing. There is the viewpoint that all we’re doing is ‘backfitting’ but there are stats tools that can help avoid that by assigning statistical significances to each trend (i.e. a measure of how influential they are and a measure of how influential that trend is likely to be in the future. <br>Re- Stream of Gold, Trabolgan, etc. Clearly trends will change and clearly individual horses may buck the trends. It happens often. What we’re lokkign for however is not the definitive answer to every race, we’re, looking for something that will give us a slight edge. Something that the market has either overlooked or underplayed the significance of.
I know one ore two are following this thread (with a mixture of disbelief, bewilderment and the type of emotion usually associated with watching sane men jump off a cliff) so i thought I’d let you know that we’ve agreed a set of trends data for the King George and said data has now been laboriously assembled and is, as I write, being fed into the guts of a processor by the stats expert on the TRF Trends Volunteer Force (!!).
More anon.
- You must be logged in to reply to this topic.