- This topic has 93 replies, 14 voices, and was last updated 19 years, 4 months ago by
.
- Posts
Corm, Carlisle
I’ve had a little think. I’ve grouped together how I would personally split the variables as shown below. Also given how i’d start by defining each variable, some need classification, i’ve added qu’s at the end.<br>King George seems a good place to start, not much time though. Please communicate if you disagree with anything i’ve written, we need to be on the same level here. With data in the formats i’ve provided, I think I could produce some interesting statistical findings. <br>Ok so:
Previous Race Variables(4):<br>a) Class of race previously won = Numeric/discrete/bounded (by class boundaries)<br>b) LTO form figure = Numeric/discrete/unbounded<br>i) Days since last ran = Numeric/discrete/bounded (max by age)<br>o) Type of race run in LTO = Numeric/discrete/bounded (by class boundaries)
Horse Profile Variables(7):<br>c) No. Chase Runs = Numeric/discrete/unbounded<br>d) Distance previously won at= Numeric/discrete/bounded (by min/max race dist)<br>e) Age = Numeric/discrete*/unbounded<br>f) Strike Rate = Numeric/continuous/bounded (0-100%)<br>k) Previous form in H’Cap** =<br>n) No. of runs that season = Numeric/discrete/bounded (by days into season)<br>q) Course form? = Numeric/discrete/bounded (win/place:1 ratio)
Raceday variables(4):<br>g) Official rating = Numeric/discrete/unbounded<br>h) Weight = Numeric/discrete/bounded (min/max weight values)<br>j) Running with a penalty = Binary/discrete/bounded (0,1)<br>p) Raceform rating = Numeric/discrete/unbounded
Outside Influence variables(2):<br>l) Influence of trial races*** =<br>m) Trainer form with seasonal debutantes**** =
* Is the foal date important? If precise age is important then this is continuous.<br>** How was this factored? Was this a subjective view or based on ratings?<br>*** How is it proposed this variable is tackled?<br>**** How broad is this? All stable debutants or restricted to 2yo/3yo/4yo+ dependant on the type of race?
Regards,<br>Ken
Hi gang
I have got chatting with someone on another Forum.<br>"forgive me father……"
He knows very little about Horse racing, much like clivex really. But, he can provide all the data that we need.
Lets pick one horse running in the Tingle Creek Chase, and produce a rating for it. I feel we need to thrash out the method manually first.
byefrom<br>carlisle
Hi
I have only got the Weekender at the moment I’ll sent out some scans.
byefrom<br>carlisle
ps Ken PM me your email, if you can……….
He knows very little about Horse racing, much like clivex really
That could be countered by an opinion about those that place too much faith in systems.
Hi clivex
my faith is firmly placed in a multifarious approach.
byefrom<br>carlisle
"your father is your mother, your sister is your brother and your all doing each other…the Norwich FAMILY!"
Just one thought…
Age of horse
I think on the jumps (maybe hurdles too) this should be banded :o Not that type of course
Cos
If say the Tingle creek had been won last five seasons by an eight year old, what would that tell us?
It would tell us that
The Tingle creek has been won the last five seasons by an eight year old
My thing about trends is that there has to be a reasoning. There is no obvious reason in most jumps races why an * year old will have more chance than a 7 or 9. But maybe a younger horse could be disadvantged againts an older one (lthough im still unsure about that)
Will a statistical blip like the above, the whole trend figure could become horribly distorted by a factor which isnt really all that illuminating
Just a thought :)
Hi
with regard to age.  Maybe we should look at the impact value the age of the winner has each year.
If one year the race is only contested be 8 year olds, this fact, (age of the winner), would hold little significance.
We will have to toy around with quite a few ideas….
byefrom<br>carlisle
"Ooohhh misses……….."
With regard to not only age but all other factors what needs to be considered is the profiles of the losers as well as the winners to guage the significance of each factor, which might, of course, vary from race to race.
Don’t worry – I’m on top of that area!
Just to let you know what’s happening – I’m compiling some King George data which I’m hoping Ken1 will then play around with.
Meantime – someone with expertise on the data extraction side would be handy.
Carlisle – what are you working on? Can you get in touch via PM about the guy ‘with all the data we would need’. That would be handy. Ask him what he/she has on teh KG and get him to sent it to you and then me/Ken. Might save me some graft.<br>
Hi gang
I am working on producing a rough, first stab, rating for Kauto Star.  Also thinking of producing Hennessey rating, retrospectively.
I will have a word with my contact, unfortunately he is on holiday at the moment.
Craig Thake wrote this about the King George (26/12/04)
Key Trends<br>*********<br>1. At least eight runs over fences          9/10<br>2. Won over at least 3m                9/10<br>3. Won last time out                   9/10<br>4. Won 47% of chase starts              8/10<br>5. Aged 7 to 9                       8/10<br>6. Within 9lb of RPR top                 8/10<br>7. Adjusted RPR of 173+                 7/10<br>8. Won a Graded chase over 2m4f to 2m6f    7/10<br>9. Previously won a Grade 1 chase           6/10
other factors<br>===========<br>No winner in the last 20 years has won on seasonal reappearence. Only one one had run in novice company the previous campaign.  Four of the nine other winners had contested the race before.
Verdict<br>=====<br>Past winners had nearly always shown a high level of form – unexposed – second season chasers haven’t a good record Repeat winners have been common, but reigning champs have struggled of late, so it is hard to fancy Edredon Bleu. Azertyuiop is a class act, but nearly all winners had been proven over at least 3m, and the horse with best claims is last years runner-up Tiutchev.  The lack of a recent run would be a worry in most cases, but not when the trainer is Martin Pipe.
Does anyone know what he wrote last year for the Sandown renewal?
byefrom<br>carlisle
ps the King George?….you buggers!
Here’s what CT had to say about the 2005 renewal
Stan James King George VI Chase 2.15 Kempton 1995 race run at Sandown in January 1996
Key trends <br>Won last time out, ten winners in ten runnings <br>Eight runs-plus over fences, 9/10 <br>Aged six to nine, 9/10<br>Won over at least three miles, 8/10<br>Won at least 40 per cent of their chase starts, 8/10<br>Within 9lb of Racing Post Ratings top-rated, 8/10 Adjusted RPR of 168+, 8/10<br>Won a Graded chase over 2m4f to 2m6f, 7/10<br>Won a Grade 1 chase, 7/10
Other factors<br>No winner in the last 20 years was making its reappearance.<br>Only two of the last ten winners had run in novice company in the previous campaign.<br>Four of the eight other winners had contested the race before.<br>The record of defending champions is 615P544.
Verdict<br>Crucially, the last ten winners had won last time out. Repeat winners have been common, but the reigning champ has struggled of late, so it is hard to enthuse over Kicking King. Kingscliff, Royal Auclair and Impek are the only last-time-out winners and all meet the required criteria on age and RPR rating. Preference is for Impek, who never runs a bad race at Sandown (two wins, four seconds and a third from seven runs) and looks much improved.
Hope this helps
Crucially, the last ten winners had won last time out
Its comments like this though that get my goat…
Why is it necessary? Say a contender finished 1/2 length to war of attrition last time out, would you dismiss him?
I think with trends its what you leave out which is more important taht what you leave in
Hi davidbrady
thanks for your last posting.  Do you keep old copies of racing papers like I do?  Mine are mostly weekenders….
Anyone with old RF Update? (they could prove to be useful)
Everything is in the melting pot….."as far as I am concerned".
Cheers from<br>carlisle
ps clivex I agree with you CT’s use of the word "crucially" seems a bit OTT.
(Edited by carlisle at 5:50 pm on Dec. 1, 2006)
- You must be logged in to reply to this topic.