Adjusting bet size to current win rate

[freeindeed]

I am in the value betting game. I mainly lay odds against the underdog winning. With odds between 3.0 - 4.5 the liabilities of a loss are very high, as I take the full hit.
I have a well established strike rate over a large sample size, so when a strategy goes on an exceptional run of wins or losses, it stands to reason that the chance of the strategy soon returning back to its proven strike rate grows with each bet.
Therefore, If the chances of a win are higher than usual - we should bet more. If the chances of a loss coming are higher - we should bet less.

I would like to formalise this into my staking system, but without any mathematical grounding, I'm not sure how to go about it. Does anybody out there have any ideas about it? Once I veer from my strikerate should I bet a certain percentage more or less until it realigns?

[Anbell]

I am in the value betting game. I mainly lay odds against the underdog winning. With odds between 3.0 - 4.5 the liabilities of a loss are very high, as I take the full hit.
I have a well established strike rate over a large sample size, so when a strategy goes on an exceptional run of wins or losses, it stands to reason that the chance of the strategy soon returning back to its proven strike rate grows with each bet.
Therefore, If the chances of a win are higher than usual - we should bet more. If the chances of a loss coming are higher - we should bet less.

I would like to formalise this into my staking system, but without any mathematical grounding, I'm not sure how to go about it. Does anybody out there have any ideas about it? Once I veer from my strikerate should I bet a certain percentage more or less until it realigns?

Games are independent and have no memory of each other. You should not change your stake based on previous results.

You say that you dont have any mathematical grounding, so this will probably seem counter-intuitive but it is mathematically true in the same way that 7 consecutive reds at the roulette wheel don't have any impact on whether the next roll will be red or black.

I hope that you disagree and do the opposite to what I'm suggesting.

[sionascaig]

Agree with Annbel...

Probably a better approach is to focus on the "value" side rather than strike rate...

If you can work out your expected value then there are approaches you can take to maximise that for a given bankroll, e.g. Kelly Criterion https://en.wikipedia.org/wiki/Kelly_criterion... But I would not apply this approach to betfair markets as: you don't really "know" your edge on any specific selection / market... and it will change by selection odds / market...

I'd suggest you work out your edge / margin and how quickly you churn the stake over a period of say a month and use that to set liability and stake sizes appropriate to the level of risk you want to run per market.

So you might end up with something like

- the amount of capital I want to risk on this strategy is X
- my max liability per market is Y (which could be X/20 or X/40 whatever)
- my max liability stake size is Z (which could be Y/10, Y/20 whatever depending on strategy)

I usually set Y & Z dynamically within a bot as X goes up & down depending on results...

[freeindeed]

Thankyou very much for the responses. I was not aware of this and it is counter-intuitive indeed. Apparently this very debate led to a major acadamic schism. It's called "The Gamblers Fallacy" or the "principle of independent events" and it took a while to fully get my head around it!