OZmium Sports Betting and Horse Racing Forums - View Single Post

Quote:

Originally Posted by Lord Greystoke

Hi Paul,

Picking up on your point here and also another in one of your earlier points(re-posted by Dale above).. what would you say is a LARGE enough sample with which to be confident in identifying:

1. what the market values correctly / incorrectly
2. where there is a pattern or theme that is more likely to be repeated with a 'reasonable level' of confidence

I have seen various suggestions on here regards a suitable sample size but given that you are in the business of quality data / database / testing over the longer term, I am interested to see what your view is on this. Apologies if I have missed one of your earlier posts which spells this out!

Cheers LG

Hi LG,

In terms of sample size, obviously the more the better although the benefits than can be gained analysing non-current data (say more than 4 racing seasons old) starts to diminish. Why? Because the market place changes and accordingly commences to revalue various factors. My own analysis of these and others is a big part of how I stay ahead of the game. This applies to point 2. that you raised.

In terms of point 1. , the benchmark returns (average returns) from backing all horses using SP/NSW tote returns around a LOT of 11% (using a proportional stake). If one can identify factors that perform much better than that say LOT of ~6% or less then that is the beginning of the process that I and others such as the major syndicates adopt. Essentially you model needs to be continually updated to cater for this aspect (it's called adaptive market hypothesis).