I appreciate your comment and completely agree. This single hypothesis is far from enough to model the much more complex reality, probably constituted by many different principles of this kind.pythonic wrote: Your model is maybe a bit too simple because in reality you typically won't have a set of players knowing the true odds, but rather a population of (more or less) sharp players with maybe somewhat different opinions about the true odds.
Great statistics Ando. I did not know of the existence of a source of betbrain’s closing odds. Is it publicly available?Ando wrote: At >=1% above Pinny no-vig odds we are looking at a yield of 6.75% for 12962 bets
2% and above: 7.60% yield for 10426 bets
3% and above: 8.43%yield for 8330 bets
I generally find the yields higher than expected, at least if I assume that column 1 and column 2 are equivalent to common definitions of EV and ROI. I would expect the average yield to be somewhat aligned with the interval of edges. Any edges at 3% or above could be very large and cause higher yields, the distribution is unknown. It is interesting to look closer at the intervals with very low edges/EV, to justify betting on these via pinnacle no-vig odds as reference.
So to remove the high edges from the statistics, I split up the population and just looked at the 1-2% interval: (P1+/Y1+ = P2+/Y2+ + P21/Y21). So P21 is the population with 1-2% edges, 2536 bets. The yield/ROI for these bets (Y21) is 4.62 %, which is still high, as I would expect something closer to 1-2%.
Maybe I am getting something wrong or we are talking different definitions or concepts. The yields in general, are very comparable to my own yields for an equal number of bets, but that is where my reference is a surebet percentage for a soft with pinnacle/sharp, and not EV percentage.
Regardless, it is always great to receive actual numbers! Looking forward to see this sample enriched with data from other sports as well.