It's driving...

[wizardofmann]

...me mad how Betfair figures can 'lie', or to be more precise mislead

All too often in a close run/close finish, the figures my bot produces on the log sheet do not make sense, well to me that is. The last instance of the latter came last Tuesday 24 Feb at Catterick 14.30, 2 mile race. The last log sheet entry read (The last second of the race before it went suspended) Venue £12,659.44p matched units -Sir Safir £5,803.76p matched units. Sir Safir won by a neck. Has anyone else come across this and has an insight as to how a runner being shown in 2nd place in my log sheet, with £6,803.76 less units matched has won the race Ok it won by a neck but the difference in units matched are just not logical (As Dr Spock would often say) I have of course noticed that when the winner has won by a 'mountain mile' and it's odds have gone to 1.01, the units can be accruing for some time before the race is suspended. Where as, a close finish goes into suspension before one can blink and eye-lid

I am afraid my old and fragile brain can make head nor tail of it and goes into melt-down every time it happens.

[Spike]

There are two possibilities:

1: The data is correct - it's your assumption that it should be different that is wrong.

2: You're measuring it wrong - check through your methodology, is it absolutely robust?

You can check if it's number 1 by downloading Betfair's historical data and checking retrospectively whether your figures tally with theirs.

If it's number 2 check though all the assumptions and logic and make sure they're completely correct. Look for any other clues as to what might be wrong (eg data going in the wrong direction). If you're calling the API directly (not through Gruss) then make sure you're handling inactive runners and other unusual circumstances correctly.

Data from the API can be a little erratic sometimes - if you have the facility try sending 100 requests in a second into a busy market, time-stamp them at your end then look at the results, you will see a degree of "noise." However if you're only refreshing once a second you won't be caught out by that.

[wizardofmann]

Thanks for the information Spike. Unfortunately its a little above my head, so please is there any chance of a 'laymans terms' explanation

[Spike]

Ok so I'll assume that your bot works through BA and excel or VBA.

You're "going crazy" because the data you're getting isn't what you expect - well, there are two possibilities; either you are expecting the wrong thing or you are getting the wrong data.

You can check whether your data is correct by comparing it to Betfair's historical data, they publish it all (google it for the download). The files are big and a bit unwieldy but if you just want to know something simple like how much traded on a particular horse in a particular race in running you can get there by loading up the relevant file into excel, performing a bunch of "sorts" and deleting everything else or using a pivot table. It'll be a pain but you'll get there. If you know a little programming it's dead easy.

If you do that and you find that your data recording is wrong then you have to work out why.
It's almost certainly something to do with your methodology - exactly how you are deriving your data. In that case you have to go through every little detail of your bot and ask yourself if it's correct in every circumstance. It's hard to be more specific than that without knowing the exact details of what you do (which unfortunately I don't have time to get into).

[wizardofmann]

Thanks Spike I will get onto it and see what I come up with. My bot works really well apart from close/blanket finishes when punters are throwing wads in all diections, hoping to bag the winner. My bot is very simple in that is looking at amounts matched at the end of each race.

[milfor]

If I understand your problem correctly you think that the winner of a race should always have the biggest amount matched.

Suppose horse A is in front for 80% of a race and the other horses are equal. Then horse A will have the biggest amount matched by far. During the last 20% of the race horse B comes nearer and nearer and improves its amount matched but firstly there still will be a lot of bets on A and secondly the last 20% are not enough time to catch up with the big amount matched during the 80%. So if B manages to win it's not surprising that it will have a much smaller amount matched than A.

I don't think that betting on the horse with the biggest amount matched is profitable long-term.

[Captain Sensible]

Betfairs records show Venue traded 107377.94 in play and Sir Safir 103661.38 in play, also Venue hit 1.01 in running where the big money is traded and Sir Safir never went below 1.11. From that data you assume it was a close race and the winner won on the line especially with it being a RUK track.

Maybe you should look at which tracks these anomalies occur as they do play differently due to the picture delays.

[wizardofmann]

Thank you very much for your replies.

'Maybe you should look at which tracks these anomalies occur as they do play differently due to the picture delays.'

That is a very good suggestion Captain Sensible and I will look into it.

Yes milfor I do but I should have mentioned that my system only monitors the 'crack end' of a race.

With all the suggestions and great input from the person who coded my bot, I now accept Betfair is not doing anything underhand. But I am afraid with my miniscule brain, things tend to get a bit misty now and again. And that's before I've had a dram or three