Im wanting to move away from the get rich quick approach to betting and treat it more like a bank/investment fund aiming for small steady % growth each month.
So I have been working on putting together a massive database of horse racing selections amounting to about 10,000 bets. In total there is a nice profit if I placed lay bets on ever selection. Im not talking 1000 points or anything like that it is more of a long term approach which slowly accumulates say month by month but it is better to look at it quarter by quarter.
My aim is to build up a small portfolio of systems (2!!!) where i make say 5% a month of my bank staking 1% each bet. I'll be happy with the greyhound one i posted if i make 5% a month (early tests indicate a lot more but i wont get carried away) upto say a £200 stake (1% of bank) as i dont think those markets can take much more than that. If I add this horse racing one to my portfolio i'll be very very happy and I can stake a lot lot more say upto £1000 a bet if the bank allowed me ..... not that i will ever get that far!
So on to my question .....How on earth do i go about analysing my data without backfitting it!!
I was thinking of the following as a starting point ....
removing anything over 20.0 .. there is a profit on some but not others so removing it as the prices here are not value when laying
removing odds on as there is a small profit but the return on the number of bets is less than 3%
maybe looking at the courses .... there is obviously a losses at certain courses .... maybe remove anything that isnt in the top80% of profit and remove all loss making course .... however there is a risk that in the future the course turns the other way??
I dont think looking at race type, runners and all of that stuff is needed as that is way too detailed and where a bit of backfitting goes on.
So in summary if i take the top 80% of profit making courses after taking out odds on and above 20.0 what do i do then ..... is there some statistical analysis i can do?
thanks to anyone who takes the time to read this and points me in the right direction.
So I have been working on putting together a massive database of horse racing selections amounting to about 10,000 bets. In total there is a nice profit if I placed lay bets on ever selection. Im not talking 1000 points or anything like that it is more of a long term approach which slowly accumulates say month by month but it is better to look at it quarter by quarter.
My aim is to build up a small portfolio of systems (2!!!) where i make say 5% a month of my bank staking 1% each bet. I'll be happy with the greyhound one i posted if i make 5% a month (early tests indicate a lot more but i wont get carried away) upto say a £200 stake (1% of bank) as i dont think those markets can take much more than that. If I add this horse racing one to my portfolio i'll be very very happy and I can stake a lot lot more say upto £1000 a bet if the bank allowed me ..... not that i will ever get that far!
So on to my question .....How on earth do i go about analysing my data without backfitting it!!
I was thinking of the following as a starting point ....
removing anything over 20.0 .. there is a profit on some but not others so removing it as the prices here are not value when laying
removing odds on as there is a small profit but the return on the number of bets is less than 3%
maybe looking at the courses .... there is obviously a losses at certain courses .... maybe remove anything that isnt in the top80% of profit and remove all loss making course .... however there is a risk that in the future the course turns the other way??
I dont think looking at race type, runners and all of that stuff is needed as that is way too detailed and where a bit of backfitting goes on.
So in summary if i take the top 80% of profit making courses after taking out odds on and above 20.0 what do i do then ..... is there some statistical analysis i can do?
thanks to anyone who takes the time to read this and points me in the right direction.