Win to place prices on horse racing - Harville

[rubysglory]

With punting on the slow side at present, I have been running some stats over varying scenarios. As we know, it is true that there exists a direct relationship between the win starting price of a horse and its actual winning chance. Curious however if anyone is aware of whether study has been done and if such a relationship exists between placed pricing and actual probability of running a place ?

rg

[PeterLe]

Have you had a look around the Adrian Massey web site??

[rubysglory]

Yes Peter. Thanks. I found the Fate of the Favourites page very interesting which got me thinking about the percentage price versus placing relationship and existence or otherwise of such a link. I know from an Eachway bookmaking viewpoint the prices are distorted where the bookie is required by the BCB to pay a set figure for the place relative to the Win (in Australia it is 1/4 the Win Odds), whereby the shorter the win price theortecially the better it is for the backer, but I was unsure if any data existed on " What is a Fair Price for Place relative to Win ? " based on actual win and place probabilities. Hope that makes sense.

rg

[Iron]

There's a site somewhere which calculates the true place probabilities using the current win odds (see http://www.paceadvantage.com/forum/prin ... hp?t=35264 if you're interested in the maths involved).

I can't remember what the edge was from exploiting the differences, but it was very small, suggesting that the place market is very accurate.

Jeff

[Iron]

PS The service I referred to in my previous email is now free - http://www.calcodds.com/default.php

Jeff

[Euler]

http://www.google.co.uk/search?hl=&q=ha ... ille+place

The problem with these forumlae is you need take into account the draw bias and how the horse actually runs and no formula directly does that. Whichever way you look at it, you still need to know form.

[PeterLe]

Yes Peter. Thanks. I found the Fate of the Favourites page very interesting which got me thinking about the percentage price versus placing relationship and existence or otherwise of such a link. I know from an Eachway bookmaking viewpoint the prices are distorted where the bookie is required by the BCB to pay a set figure for the place relative to the Win (in Australia it is 1/4 the Win Odds), whereby the shorter the win price theortecially the better it is for the backer, but I was unsure if any data existed on " What is a Fair Price for Place relative to Win ? " based on actual win and place probabilities. Hope that makes sense.

rg

To be honest i dont know very much about horse racing believe it or not...but I would have though that it would be very difficult trying to find an edge or profitable system based on stats? Surely there must be tens of thousands of casual punters doing the same (and no doubt very clever Phd candidates too with large computer modeling systems?!)

Mind you having said that I did read a blog sometime ago about a girl who had a "Laying System" (just google "Maria Laying system"), who did extremely well, and from memory I think tht was statistic based? (She never disclosed her system by the way!)

regards
Peter

[rubysglory]

Thanks guys for something to work with. Jeff, your links particularly enlightening especially the 'Discounted Harville' theory. Euler, I understand what your saying but not sure if a ratings type analysis is any more or less necessary than some might already undertake. The Maria Laying system sounds like it might make interesting research as well PeterLe. Found this link which others may like : http://www.premiershiptips.com/betting- ... ing-system.

Expanding further, by being able to calculate the Fair Odds for a place based on market Win odds, what I hope to do is perhaps find a way of exploiting any overlays that may exist. By example, mathematically speaking, the place odds can be particularly bad for a bookmaker. Consider a 1/1 (2.00) chance from an eachway bookmaker's perspective. If this was in fact five chances each of 9/1 (10.00), the eachway bookmaker would be far better off. This is because at 1/1, the BCB enforces a payout of 1/4 (80%) whilst across 5 runners, the place odds would be 5 runners each of 9/4 (30.76%), or a spread totalling 153 %. This is representative of a difference in favour of the layer (the bookmaker) of approx 73%.

Cheers again guys.

rg

[to75ne]

rubysglory i think the maria laying system is not really worth spending any real time on . if i remember correctly, she lays to a set liability over 3 or so odd ranges. so if things go wrong she will lose a set amount in proportion to her bank etc.

the bit that made her successful is the bit she will never reveal (and quite rightly) her edge; which is her selection criteria for which horse to lay. she only revealed her bank management method.

[rubysglory]

Thanks to75ne. I think you are spot - I suspect that 'maria' may have had access to a good source of stable information.

Ultimately, it is the possible existence of any relationship between Win Price (assuming a Perfect Betfair Market of 100%) and the Place Price that I seek to find. What is the 'true' Place Price for a 2.50 Win Chance ? Pie in the sky stuff no doubt, but will no doubt keep me busy for a bit.

rg

[Euler]

It's perfectly possible to do, just very complicated and when you do it, it turns out the market is better. The market seems to discount, running style and the draw bias.

But if you want to work out the odds on horse B in a five runner race you would need to do the following steps.

Algebraic description: -

Horse A wins = A
B beats all others = (c+d+e)/(b+c+d+e)
Horse A win and B 2nd = a wins*b 2nd

C wins B beats all others (2nd) c*(b/(a+b+d+e)
D wins B 2nd d*(b/(a+b+c+e)
E wins B 2nd e*(b/(a+b+c+d)
B comes in second = All the above added together 40%+0.53% etc etc

B wins = Implied probability

B wins or B comes 2nd = b comes second + b wins

It can get quite complicated.

[Iron]

I once heard there are some horses which rarely win, but which often place, meaning you can't derive the true place odds of such horses from the win odds. This is possibly because horses are pack animals, and being in front might make some horses feel exposed.

Also, I believe low-odds favourites are less likely to place than their implied place odds suggest (presumably because they have a greater tendency than other horses to either win or finish nowhere, depending on whether they run to potential).

Jeff

It can get quite complicated.

[rubysglory]

Thanks Euler. Designing a trifecta formulae similar to that which you exampled and then extrapolating the required data might be the way to go. Mind you, Jeff, you may well hold the key with your 'pack animals' theory. Some horses just win or run nowhere.

Phew ! Going to keep me busy I think.

Thanks again guys.

rg

[toptrader]

Interesting to understand if you think this pack theory applies to Greyhounds also, or if you think a calculation of the place odds would be 'fairer' as there are less other factors in the way

(there are different factors to do with the course etc, but at least there's not jockey to interfere)

[maycontainnuts]

Been getting some amusing emails recently on a fool proof way to make money trading on horses. Found this thread on the forum so worth a bump.