Determining Long Term Profitablity

[lm1994]

Hi all,

I have an auomated strategy which is working pretty well, the strategy is in profit and has around a 55% win rate.

I only have a small sample size with around 360 data points.

My question is, at what point is it best to start scaling up considerably with the peace of mind that its long term profitable.

Ive ran monte carlo simulations on my results so far which show 99% plus of simulations end up in profit over 1000 markets.

And I've compared my % return in relation to cumulative turnover which after the initial variability has stabiled positive at 0.74%

any help or advice is very much appreciated.

cheers.

[jamesedwards]

Hi all,

I have an auomated strategy which is working pretty well, the strategy is in profit and has around a 55% win rate.

I only have a small sample size with around 360 data points.

My question is, at what point is it best to start scaling up considerably with the peace of mind that its long term profitable.

Ive ran monte carlo simulations on my results so far which show 99% plus of simulations end up in profit over 1000 markets.

And I've compared my % return in relation to cumulative turnover which after the initial variability has stabiled positive at 0.74%

any help or advice is very much appreciated.

cheers.

You biggest problem is likely to be a high gearing ratio between ↑stakes and ↓ROI%.

I recommend you increase stakes very slowly and keep a close eye on profitability (value and %).

[Euler]

And I've compared my % return in relation to cumulative turnover which after the initial variability has stabiled positive at 0.74%

That's a positive sign. If you gently increase stakes and continue to monitor that, it should give you an indicator as to whether it's influencing outcome.

Be aware of things like seasonality and how they could affect the strategy, or else you could raise stakes just as that factor kicks in.

[eightbo]

Genuine question – how come people use return / turnover ?
Is it not more accurate to factor in odds/liability as if you trade at odds of 20-50 on Strategy A and 1.20-1.50 on Strategy B the results will be completely incomparable ?

Re: scaling – you can assign the bot a starting bankroll and add on the market result, and set stake to x% of bankroll.
Export StakeSize & Result and when you grow too big your edge will lessen and the P&L will gradually flatten off when you've exceeded the max the mkt can take. Just revert your stake to fixed £x liability when you're confident with your rough evaluation (will never be perfect), and check in on it every noe and then

[Crazyskier]

And I've compared my % return in relation to cumulative turnover which after the initial variability has stabiled positive at 0.74%

That's a positive sign. If you gently increase stakes and continue to monitor that, it should give you an indicator as to whether it's influencing outcome.

Be aware of things like seasonality and how they could affect the strategy, or else you could raise stakes just as that factor kicks in.

I was going to say this, but was beaten to it. Even the longest-running strategies have peaks and troughs and periods of loss. It really is all about identifying those repeated, consistent troughs and any pertinent factors that can be adjusted.

CS

[Anbell]

Genuine question – how come people use return / turnover ?
Is it not more accurate to factor in odds/liability as if you trade at odds of 20-50 on Strategy A and 1.20-1.50 on Strategy B the results will be completely incomparable ?

If you're trading, ROI by bucket seems a useful indicator?

[eightbo]

Genuine question – how come people use return / turnover ?
Is it not more accurate to factor in odds/liability as if you trade at odds of 20-50 on Strategy A and 1.20-1.50 on Strategy B the results will be completely incomparable ?

If you're trading, ROI by bucket seems a useful indicator?

Hi Anbell, thanks for your comment
Could you elaborate on Bucket, please?

I have found ∑profit/(∑profit+abs(∑losses)) to be most useful in evaluating the long-term performance of a strategy e.g. 50% for breakeven. Having a rough EV also helps to forecast potential drawdown and work backwards to derive appropriate stake size

[Anbell]

Genuine question – how come people use return / turnover ?
Is it not more accurate to factor in odds/liability as if you trade at odds of 20-50 on Strategy A and 1.20-1.50 on Strategy B the results will be completely incomparable ?

If you're trading, ROI by bucket seems a useful indicator?

Hi Anbell, thanks for your comment
Could you elaborate on Bucket, please?

I have found ∑profit/(∑profit+abs(∑losses)) to be most useful in evaluating the long-term performance of a strategy e.g. 50% for breakeven. Having a rough EV also helps to forecast potential drawdown and work backwards to derive appropriate stake size

A price bucket. So you could group from say 1 to 2, and 2 to 6 and 10 to 100 etc

[eightbo]

So strange you mention that, was about to implement that in some automation files for the first time on my value betting as same logic seems better/worse at certain prices. Never thought about doing it with trading but makes sense, thanks for the tip