Forums

General Betting

Welcome to Live View – Take the tour to learn more
Start Tour
There is currently 1 person viewing this thread.
Artisan
05 Mar 10 19:39
Joined:
Date Joined: 15 Jun 06
| Topic/replies: 208 | Blogger: Artisan's blog
This strategy, which says that you back before the off every horse above odds 12.5, and place a lay to be taken in running at the square root of the back price, looks fantastic if you analyse the Betfair data. Why doesn't it work in practice?
Pause Switch to Standard View Back above 12.5 and lay at the square...
Show More
Loading...
Report Artisan March 5, 2010 7:40 PM GMT
PS - Must be to level stakes
Report cunningplan March 5, 2010 8:04 PM GMT
never heard of it enlighten me more
Report Artisan March 5, 2010 8:14 PM GMT
You back every horse in a race above odds of 12.5 and below say 950 for £2 before the off. You place a lay bet at the square root of the back bet. So e.g. backed at 25.0, you lay at 5.0 with a stake that is 2 * back price / layprice.

Play this through historical data the results are stunning. Working in practice another thing.

My question is why? Liquidity? Changing market shape? etc. etc.?
Report Wrong Hole Poker March 5, 2010 8:16 PM GMT
lol
Report Artisan March 5, 2010 8:38 PM GMT
WHP, "Wrong Hole Poker", although I hope I will never fall foul of that predicament, thank you for the sentiment.
Report Chosen® March 6, 2010 6:30 AM GMT
A more interesting question is why this would produce 'stunning results' from historical data.

Btw, how do you actually run such a system through historical data?
Report Daveyboymoyes March 6, 2010 8:16 AM GMT
True, no way of finding such results is there?
Report brendanuk1 March 6, 2010 8:28 AM GMT
its problem with backfitting "random" criteria that turned up a sequencce of results but do not point to value or any basis on why it would be a true "edge"
Report Artisan March 6, 2010 2:15 PM GMT
You test it on historical data by downloading the betfair data and derive the price at the off, and the low in-running for each horse.

The theory behind this is that it is a volatility play. Tested on 10,000 races, the results look very far from random.

Has anyone else looked at this?
Report Gallivanter March 6, 2010 5:44 PM GMT
You'll often see layers offering 6.0 about a horse that went off at 36.0 but that doesn't mean they'll get matched.

When you analyse historical races, what prices are you looking to lay at?

If you use the price available to back then there's no guarantee that you'll get matched because that price was put there by another layer. If you use the price available to lay then you will get matched but only up to the amount the backer offers. If you use the last price matched then you may or may not have got matched depending on your position in the queue and the amount you wanted matched.
Report Bayes. March 6, 2010 6:01 PM GMT
If this was a winner you almost certainly leave the back part out because it is very unlikely to be profitable in it's own right.
Report Gallivanter March 6, 2010 6:18 PM GMT
Reading this again, I see the injunction to lay "with a stake that is 2 * back price / layprice". This could be simplified because the figure will always be twice the lay price, as dividing any number by its square root returns the square root, in this case the lay price.

Prices aren't magical and there's no reason to lay at an amount that is twice the price.
Post Your Reply
<CTRL+Enter> to submit
Please login to post a reply.

Wonder

Instance ID: 13539
www.betfair.com