betting systems

[reeve26]

does anyone know how to test what the probability is that a betting system has made a profit because it is a good system rather than just by chance? I have heard there is some spreadsheet calculation, can the calculation also be done on a scientific calculator ?

Any help would be appreciated

[mathare]

I have seen the calculations for this but I can't recall them off the top of my head. I think it involves calculating average profit/loss per bet and then working out the standard deviation or something along those lines. Which means it would be very tricky to do on a calculator

[Win2Win]

...Or you could just run it live paper trading for around 100 bets to give an idea of where it is going. Depends on SR how many proofs you need though. 20% SR should be around 500 bets.

[mathare]

OK, one approach would be to calculate the chi-squared for your results.

For this you need to know what sort of strike rate you would expect. Now you can gauge this from experience or you can guess or you can use the average odds of your selections and convert that to a percentage chance of each winning. Of course this worls best if all yours odds are roughly the same and you've not got odds on shots and 40/1 outsiders in the same set of data. And the average you choose is up to you. Do you ignore the statisitical outliers and work out the mean? Do you work out the median? Go for the modal maybe? Work out all three and take the mean of those? Up to you.

Once you have a strike rate use it to calculate the expected number of winners in your sample, and the expected number of losers. Then count up the actual number of winners and losers. Then:

chi-squared = ((Wobs-Wexp)^2/Wexp) + ((Lobs-Lexp)^2/Lexp)

where the W = winners, L = losers and obs and exp stand for observed and expected.

You're looking for a value around 5 or greate, with 5.02... corresponding to 2.5% chance that your observed results were just luck.

[Win2Win]

:ooo

[mathare]

I realised on the way to work this morning that I had made a mistake in the formula given above so I have now corrected it. You need to square Wobs-Wexp before dividing by Wexp, and similarly for the losers too.

[sparkyminer]

Nice signature Mat.:)

[MattR]

I realised on the way to work this morning that I had made a mistake in the formula given above so I have now corrected it. You need to square Wobs-Wexp before dividing by Wexp, and similarly for the losers too.

We were all wondering when you'd correct it Mat, it was so glaringly obvious :ooo :D

[Win2Win]

Just imagine what wonderful fun it would be if we could read Mat's mind :geek

[MarcusMel]

......... Of course this worls best if all yours odds are roughly the same and you've not got odds .......

:D

[mathare]

OK, I made multiple mistakes. I am human it seems :(

[Win2Win]

OK, I made multiple mistakes. I am human it seems :(
Your not happy to see Marcus back? :D

[mathare]

Your not happy to see Marcus back? :D:laugh

Not at all happy, it means he has escaped from my basement where I'd had him locked up for the past few months :yikes:

[presto]

:laugh

Not at all happy, it means he has escaped from my basement where I'd had him locked up for the past few months :yikes:

i would have expected that from vegy................. but not you mat :yikes: