7th September, 2017.
A number of people have asked me to go through my database and establish whether there was a bigger edge for betting higher margin edge bets that my model prices compared to the market.
In the daily spreadsheets, I do include analysis on the implied percentage edge between model and market price, so using the below example, if a player is priced at 2.50 but my model prices them at 2.00, the implied edge would be calculated as follows:-
(100/2.00)-(100/2.50) = 50-40 = 10%
I went through my database from 1st September 2014 to 31st August, 2017 – three years of data – to look to see if betting on players with a bigger edge would yield a higher return on investment, and I generated the following results based on staking to win a hypothetical £100 return based on the Pinnacle prices available at the time I send the daily spreadsheets:-
|
Edge % |
Matches |
Stake |
P/L |
ROI % |
|
15.00% + |
95 |
7576.42 |
962.96 |
12.71 |
|
12.00-14.99% |
119 |
10020.44 |
856.05 |
8.54 |
|
8.00-11.99% |
336 |
35088.15 |
-1197.48 |
-3.41 |
|
4.00-7.99% |
776 |
93487.77 |
4462.13 |
4.77 |
From smaller samples, there certainly were bigger returns as the implied edge percentage was highest, and combining the 12%-14.99% and 15.00%+ brackets generated 10.34% ROI across 214 bets. If a bettor was extremely risk averse, and content to pick their spots, taking this approach historically has yielded very strong results. It also indicated that generally, these high value spots don’t tend to be ‘too good to be true’ and that ‘market knows’ may be an over-rated phrase.
However, the results between 8.00% and 11.99% edge were slightly negative, but the 4.00% – 7.99% ROI was almost exactly where I’d expect it to be, given the huge number of 4.00%-5.00% bets recorded.
For those who are interested, the full database can be downloaded below.
Part two will look at the same concept, but in the WTA, before I look at some various staking strategies (including Kelly and derivatives of) to look at the most efficient staking strategy to adopt.
