Well, altogether 262 horses ran on Sunday, so that's an initial cost of £2.62, and, with over £4.00 returned you didn't even payout - but there maybe a better way.
Lay all horses at 1.01 again for 1p. But, that might not guarantee matching on the winner.
Ok, what is the reason for wanting a list of winners?
Can Bet Angel give a list of winners?
Winners are listed in a nice CSV file here, but get updated a bit erratically: -
https://promo.betfair.com/betfairsp/prices/
https://promo.betfair.com/betfairsp/prices/
Hi Euler,Euler wrote:Winners are listed in a nice CSV file here, but get updated a bit erratically: -
https://promo.betfair.com/betfairsp/prices/
That's great - thanks. I think what I'll do is get data from a variety of sources (your suggestion, the RSS feed and have screen scraping from the sporting life website as a fall-back for any that I don't match.
I really don't have to be 100% - I'm looking at probabilities over a large data selection, so if I miss a few races out of the 100+ that occur each day, then that's still ok.
Thanks again
Ian
You're right, I didn't take into account the winners. However, I think you're talking about 'Rounding' - finding a set of odds across the field and a set of varying amounts to bet (or lay) that mean you win more than you lose over a period of time... That's very rare with Bookie odds (as in, I've seen it once in over 4,000 races) and almost as rare with Laying on Betfair.pdupre1961 wrote: Well, you're probably not quite right there as Wincanton would have given a nice profit...
Personally, I think hybrid-rounding can work - don't look for event/odds combinations where you can bet on 100% of the field - bet on a smaller percentage (shortest odds) and rely on probability to sort things out over the long term.
If it were as simple as betting or laying amounts on all runners, then lots of people on here would be doing very nicely indeed
-
pdupre1961
- Posts: 410
- Joined: Fri Feb 18, 2011 8:01 pm
EyePeaSea you didn't answer my question...pdupre1961 wrote: Ok, what is the reason for wanting a list of winners?
pdupre1961 wrote:EyePeaSea you didn't answer my question...pdupre1961 wrote: Ok, what is the reason for wanting a list of winners?
How long have you got?
My 'edge' is to look at historic data and apply an algorithm with different parameters, looking for a permutation of bets in a race (number of horses to back/lay, how much to bet/risk) that provides, over time, more profit than loss.
So, I have my historic data, then run a specific permutation and look at what the outcome would have been.
To expand slightly...
I have two systems running at the moment, one using BetFair (just the Lay side) and the other betting with an online bookie.
For the Bet system, I ran an entire analysis the other week; testing my algorithm against 4,309 races. I have 6,171 different permutations of parameters to run each time and each permutation requires around 100 calculations. So, 2,659,083,900 individual calculations (give or take).
On the Financial Investment adverts they say "past performance isn't an indication of future performance.". However, with statistics and probability, the exact opposite is true. Size (of the data set) matters... at least when it comes to statistical analysis
The Lay system is similar in principal - it's just a lot of number crunching and running through scenarios. But without the results - none of that would be possible
Ian
-
pdupre1961
- Posts: 410
- Joined: Fri Feb 18, 2011 8:01 pm
How far back do you need?EyePeaSea wrote: How long have you got?
Going back isn't an *option (see below if you're really bored) so despite what I said about data set size, it's just going forward that I'm interested in. I just ran a scan for winners from the 26th Oct to yesterday, and I got all but 22 winner names from 787 races, and so I'm only having to ignore about 3% of the data. Once I add in the extra source of winners (thanks to Euler) I'll get that as close to 100% as I need to.
Ian
* I collect data and odds at multiple points in time - so even if I got the winners all the way back to the beginning of the year, I wouldn't have the matching odds data - just SP, which doesn't help.
** And yes, I'm probably more interested in the challenge, than the result. I think of a new mathematical approach (although no approach is ever really 'new'), spend weeks coding it, months collecting and analysing the stats and then (usually) see that it doesn't work out. So I move on to the next idea. It keeps me out of the pub
Ian
* I collect data and odds at multiple points in time - so even if I got the winners all the way back to the beginning of the year, I wouldn't have the matching odds data - just SP, which doesn't help.
** And yes, I'm probably more interested in the challenge, than the result. I think of a new mathematical approach (although no approach is ever really 'new'), spend weeks coding it, months collecting and analysing the stats and then (usually) see that it doesn't work out. So I move on to the next idea. It keeps me out of the pub
Would this site help?EyePeaSea wrote: * I collect data and odds at multiple points in time - so even if I got the winners all the way back to the beginning of the year, I wouldn't have the matching odds data - just SP, which doesn't help.
http://www.fracsoft.com/
I've not used it myself, but I've read the site mentioned on this forum in the past. You should be able to get that second-by-second historical data which you're after.