Research Question

[Mad Gambler]

I'm currently work out with the aid of a database a system that I hope will produce long term profits. My question is that so far I have some 500 race results. How many would be needed. I think about 5000, but I may be wrong.

I hope to be able to post some of the fruits of my research at a later date.

mad gambler.

[wesmip1]

Mad gambler.

You need at least 5000 races .. I would think of it more in years. I use 3 years at the moment.

1 year for coming up with systems.
2 years for testing

Good Luck

[TWOBETS]

I have noticed that with all my wacko systems I have concocted over the years that the length of time required to test the system is inversely proportional to the average odds of your runners.
Thus if you have a system following odds on runners I believe a shorter sample will be quite indicative of future results. Conversely the longer priced investments will require far longer trial periods.
Just my observation.

[Privateer]

I used 2 years worth of results, devised my current method then trialled it for 4 months on paper before investing. Still working after a number of years.

[wesmip1]

Just thought I would mention it is important to have a long length of time vs a lot of selections.

I am currently testing a system that has had over 896 bets this week (thats not a typo and its not a laying system). It is showing a profit of 34% for the win and 3% for the place. It is also showing a strike rate of 26% for the win and 51% for the place.

But there is no way I would trial this with real money despite the large number of bets. It needs to be tested over at least 6 months to a year before you could be confident in the method.

I have another system which I have tested over a much longer time frame and it has only had 104 bets this year. It is up 60% for the win and 32% for the place with a strike rate of 29% for the win and 61% for the place. I have tested this over years of data and it works consistently over most time frames. The testing data for this was just under 600 races but it encompased 2 years of data.

I am much more confident in the second system then I am the first system even though it has had a lot less bets in the test period (896 vs <600).

Good Luck.

[partypooper]

I have noticed that with all my wacko systems I have concocted over the years that the length of time required to test the system is inversely proportional to the average odds of your runners.
Thus if you have a system following odds on runners I believe a shorter sample will be quite indicative of future results. Conversely the longer priced investments will require far longer trial periods.
Just my observation.

I think that observation is much deeper than you think TWO BETS, in fact that's where I picked up my nom de plume.

i.e. take ALL 1-1 moneys shots S/R = about 45% (not exact figures)
take ALL 2-1 shots S/R = about 30% ''
take all 3-1 shots S/R = about 22% ''

see what I'm getting at? just consolidates what you are saying. if your system targets 10-1 shots you will have a s/r (over time) of something like 7-8%, regardless.
of course the idea of your system is to lift that s/r 11-12%, then youre cooking with gas for life ey?

ps, i did the exercise using pre-post prices and the result was almost identical s/r AND LOT

[Bhagwan]

I have found 4 years of data to be ideal.

What I found was the SR was consistant at 25% accross all the systems batched together, but the average price started to drop away.

So I then started to bet the horse only once instead of multiple times because it was selected by multiple systems & the POT improved.

If one is using multiple systems .
Try not to double up on the bet because it is selected by multiple systems e.g. If the same horse shows up 5 times, it does not mean it has 5 times the chance of winning , so its best to only bet it once,, for better long term results.

Cheers.