Trading Greyhound racing : bot behaves completely differently in practice mode to live

We've gone to the dogs.
Post Reply
alun63
Posts: 13
Joined: Tue May 09, 2017 8:44 pm

Sun Oct 15, 2017 8:26 pm

Im sure im not the only one to experience this but maybe the community has an answer, i have a grehound bot that works very well in practice mode with over a 90% strike rate as soon as i go live its a 90% failure rate i realise that due to the low levels of liquidity that stake size can influence greyhound markets greatly but uts the same results with £2 stakes, im at a loss to work it out. any words of wisdom would be appreciated

User avatar
ruthlessimon
Posts: 1095
Joined: Wed Mar 23, 2016 3:54 pm

Sun Oct 15, 2017 8:48 pm

I saw a video recently about moving from sim to live. Although this was more aimed at manual traders. But the rule of thumb was the following:

"Divide your sim performance by 5"

Out of interest, had you done this calc - would you still be "nicely" positive?

xitian
Posts: 441
Joined: Fri Jul 08, 2011 2:08 pm

Sun Oct 15, 2017 9:19 pm

One other potential issue is latency of live bets vs simulated. I don't know if practice mode simulates latency or not. Are you mostly taking prices or offering prices?

If you're taking, are you finding that you have a lot more unmatched bets than usual? If you want to improve latency, you can try running your automation on a remote server (assuming you're running from a home PC currently). That can shave a few ms off which might make a difference.

Or if you're offering, are you noticing that you're arriving behind a queue whereas you'd normally be at the front? Or it could be that in live mode, people can obviously see your bets and might find them to be good value. In simulation, your naff bets might not be taken as readily as no one can see them.

How many bets are you matching in simulation vs live?

alun63
Posts: 13
Joined: Tue May 09, 2017 8:44 pm

Sun Oct 15, 2017 9:35 pm

thanks for the responses, hadnt heard about dividing the sim by 5 ill check that out, im offering prices on a remote server, i just find that in live mode they are invariably drifting away from the trade and im not getting matched nearly as often as sim mode it does seem that its probabluy only worth developing the bot in live mode but that gets expensive! i dont have the same issue quite so dramatically on the horse racing markets

User avatar
Euler
Posts: 15638
Joined: Wed Nov 10, 2010 1:39 pm
Location: Bet Angel HQ
Contact:

Sun Oct 15, 2017 9:45 pm

Dividing by 5 is pretty harsh. It's very dependant on what you are doing, the spread, the stake etc.

It's always better to use live mode anyhow, even at small stakes as it is much more representative. Practice mode is there to debug and familiarise.

alun63
Posts: 13
Joined: Tue May 09, 2017 8:44 pm

Sun Oct 15, 2017 11:30 pm

yes i came to that conclusion i was just surprised at the dramatic difference still ill just have to spend the money i guess!

Post Reply

Return to “Trading Greyhound racing”

  • Information
  • Who is online

    Users browsing this forum: No registered users and 1 guest