Mysteries of practice mode

[guy333]

Testing systems is so frustrating when two sets of rules are almost exactly the same but produce different results.

I am trying out two systems, the only difference is that the stop loss is slightly more cautious on the version one.

Lo and behold the first race of today the back bet triggers on system two at a different time than the back bet on system one, so the stop loss triggers on the less cautious system but does not trigger on the more cautious system even though the back bet conditions are exactly the same for both.

It seems the only reliable way to test is in live mode, which could be quite expensive.

[jamesedwards]

Testing systems is so frustrating when two sets of rules are almost exactly the same but produce different results.

I am trying out two systems, the only difference is that the stop loss is slightly more cautious on the version one.

Lo and behold the first race of today the back bet triggers on system two at a different time than the back bet on system one, so the stop loss triggers on the less cautious system but does not trigger on the more cautious system even though the back bet conditions are exactly the same for both.

It seems the only reliable way to test is in live mode, which could be quite expensive.

Practice mode is excellent for testing rules are triggering as and when you are expecting.

But it's less useful for testing profitability for several reasons;
> You're not part of the market in practice mode. When you become part of the market you add negative pressure to the price you achieve
> Bets in practice mode will match more easily and more completely. eg if there's 10p available at top price then your bet won't use it up because your bet doesn't really exist, so you could match £100 to that 10p.
> Bet placement time is slightly reduced in practice mode because the dummy transaction happens on the host machine rather than travelling to Betfair servers and queuing through their matching process.

The best way to test profitability is in live mode with the smallest possible stakes. Expect ROI% to drop as stakes increase, so increase stakes slow and steady.

[ShaunWhite]

It seems the only reliable way to test is in live mode, which could be quite expensive.

As James said, smallest possible stakes, and even if you have no edge at all then you shouldn't lose too much as prices are generally reasonably fair.

But testing things does cost money, the best mental approach to that is think of testing as buying a lottery or raffle ticket, ie you don't expect to win. But if you do then unlike the lottery, your strategy will keep winning every week. You'll have to speculate to accumulate.

Key though is knowing how long to run it for, almost nothing is going to win every day or even every week. But then again you can't just keep plugging away throwing money at something hoping it's going to suddenly have a reversal of fortune.