- This topic has 34 replies, 13 voices, and was last updated 19 years, 9 months ago by
.
- Posts
Maybe you should warn us next time you’re going to tell a joke, just in case we miss it
:biggrin:
Surely though the bookmakers are using some programme or other to pull togther this information even in a basic form (based on official ratings say?)
I wouldnt have thought that probablities within a band using the deviations was beyond most staticians ability?
Can everybody calm down – I thought it was obvious that clivex was joking – the smilie says that – and it was kind of funny.
I’m pretty lost regarding the statistical deviation etc due to being out of practice with my maths so using the race given in the example, can somebody please post up exactly how the probability that Bert’s Memory will beat Grand Art (for example) if BM runs to a rating of 80 is calculated – or at least to a point where repetition only is required to calculate the probability
I found this website via Google (the first site returned when searching for "calculate betting odds from horse ratings" ) but the odds are pretty unuseable as the odds given range from only 8/1 to 11/1 for the 10 horses given in the example.
http://www.ozracetools.com/modules.php? … s_To_Price
(edited to include the link and get rid of the erroneous smilie caused when you try to put things in brackets.<br>
(Edited by davidbrady at 12:48 pm on Aug. 9, 2006)
Apologies if I overreacted a little – I had an early start today and my sense of humour may be having a lie-in.
Surely though the bookmakers are using some programme or other to pull togther this information even in a basic form (based on official ratings say
This is an interesting point – I have often wondered whether bookies use particular software and whether it might be commercially available.
NP Arandale :biggrin:
Would be intresting if that Coral bloke on here could enlighten us? Maybe he could nab a few copies for TRF members? TDK? :)
Or maybe Barry Dennis could input?
Im genuinely intrested. With so much racing, they must have some template which i would imagine is based upon the formulas we are looking for.
I suspect that David’s find is not the one though ;)
I’ve been trying to convert my own ratings into odds for quite some time, but it’s not easy as several contributors have already mentioned. For a start, common sense dictates that you cannot ignore the market completely, yet if you factor it in, there is a chance that you will miss the value opportunities that can make the difference between profit and loss.
I think the best way forward is to get a ‘feel’ for your own particular ratings and develop some flexible ideas about translating ratings differences into prices.
As an example, using the 3.00 Newcastle(where there is limited scope for using ratings because of the absence of information), my own ratings come out:
Bert’s Memory 90<br>Grand Art 84<br>Ronnie Howe 83<br>Seaton Snooks 82
There are a few unraced horses, so any tissue based on ratings can only come up with prices for rated horses unless you have some method of assessing unraced horses.
My very approximate conversion of ratings to odds is to allow about 4 percentage probability points for each ratings point. For the above race, Bert’s Memory would be about 24 percentage probability points ahead of Grand Art. So, if we put Bert’s Memory in at about 3/1(25%), she would be well clear of any opposition. Move her down to about 7/4(36%) and Grand Art comes in at 12%(8/1). Ronnie Howe would be 8%(12/1) and Seaton Snooks 4%(25/1). I think you have to fiddle around like this and decide whether to bet or not – it’s a lot easier where there are a lot of exposed horses.
Using 4% prob = 1 ratings point is rather arbitrary, but it works reasonably well in practise. If you try to convert it into a formula to encompass the entire field in a tissue, you get reasonable results.
e.g
Horse A rated 106<br>Horse B rated 104<br>Horse C rated 100<br>Horse D rated 98 <br>Base = top rated -10 = 96
A +10, B +8, C +4, D +2
A spread of 24 points to allocate, 1pt =4%
A =40%(6/4)<br>B =32% (2/1)<br>C =16% (6/1)<br>D =8%(12/1)
Doesn’t always work out so coveniently – although I didn’t ‘cook’ it.
The adjustment you make for your ratings (assuming they are accurate in the first place) will depend on the type of race besides other matters.
It would not be unusual to find a Group race for older horses in which there is just 10 lb between the top-rated and bottom-rated and yet those horses will be at 2-1 and 100-1.
If the same ratings applied to a claimer the difference might be more like 10-3 and 8-1.
This is linked to the reliability of the race and not just the standard deviations of the horses themselves.
I think that Coral use a sophisticated computer to figure this all out, namely TDK’s brain…
Human beings are relatively good at solving these problems that require pattern recognition, just so long as the human concerned has been exposed repeatedly to similar patterns in the past (otherwise known as a misspent youth).
It is one of the reasons why the level of complexity involved in horseracing analysis is so appealing to human beings in the first place, as opposed to doing brain surgery in your spare time, or tiddlywinks.
Evening Prufrock, hope you are well. You’ve not been stuck with the late shift at Kempton have you? Like some of us!
Evening DJ. Yarmouth, not Kempton, though therein lies a story I may regale you with over a beer some time…
Speaking of which, I was hoping to get to the first two days of the Ebor meeting. Was hoping to head your way but seems I may not manage it. Will you be on the Knavesmire at all?
By the way, I do know of people who use the sort of involved calculations (bell curves etc) referred to earlier, and apparently it works.
Will be there all 3 days Prufrock, so I look forward to discussing with you over a pint whether the Gimcrack is up to standard!!:biggrin:
EC – Sorry, just in. Here’s how it shaped up –
93 Reload 1.37/1 :o <br>82 Spinning Reel 7.32/1<br>78 Aryaamm 10.25/1<br>78 Changing Wind 10.25/1<br>77 Country Escape 11.03/1<br>75 Tiz Timely 12.83/1<br>72 Mosharref 16.18/1<br>72 Mantolini 16.18/1<br>55 Dik Dik 113.53/1
<br>Bsed on your ratings alone Reload was the clear value (at 20/1) and ran a fair race in 4th. Moshareff was diabolical value at odds on, again according to your ratings and my ratings/price tissue calculator!!
Interesting point about ratings and the market. A short story really, but I’ll keep it brief.
There is (or was) a syndicate that played exclusively on the Hong Kong pari mutuel. The used a prediction model with 20 or more variables and used mathematical techniques of multiple regression analysis. They couldn’t make it pay UNTIL they factored in the opinion of the betting public at large(including, of course, the ‘shrewd money’).
I read this a few years ago in Nick Mordin’s book, Winning Without Thinking. In fact, there was an appendix containing details of the method the syndicate uses(d). I loaned the book to someone who never returned it, so cannot give any more details.
- You must be logged in to reply to this topic.