Paper betting - advice sought

[Zoob]

I'm more of a football fan (when it comes to both watching, and betting), but when the season finished I started gathering info on horse racing. I know fairly little about horse racing, (that'll set the alarm bells ringing...) therefore, because I have a background in stats, I started paper betting based on the odds of the horses in each race. My bets are mostly lays (average of around 9/2), but also involve some backing. Depending on the the betfair odds, number of runners and some other factors, I place one of 5 bets (i.e. laying horses at certain odds, or backing at certain odds...) . For each type of bet, I have followed between 100 and 300 races. In total, I've paper bet in over 800 races over the last 40 or so days. I'm currently averaging around +4 points per day, with 2 out of 3 days being winners, and the average winning day being better than the average losing day. 800 races seems a lot to me, but since horse racing goes on all year round with an average of 20 UK races a day, I'm not sure if 800 over 40 days is a big enough sample size. The fact that I'm doing so well from my paper betting makes me think I should start betting for real, but I'd be interested to know how long any of you punters think a system should be tested for before going live? Any thoughts would be much appreciated... Cheers.

[slapdash]

Re: Paper betting - advice sought :welcome to PL, Zoob! It's the size of the sample that matters, not the total number of races. So the information you get from a sample of 800 races will be just as reliable, whether there are 7000 or 7000000 races a year, so long as the sample is chosen randomly. What I'd be most careful about is that if your sample comes from 40 consecutive days, then it may well be biased. For example, your sample covers the start of the flat season, when there are a lot of horses running with no recent form, and it's quite likely that a system based on market information would perform differently during this period.