Following on from the previous article I decided to dig a little deeper into the tipster data. The first thing that bothered me was whether I was being a bit harsh on Mr Flanagan. Its hard to imagine how he hit those numbers and he has not contested them on Twitter. Could it have happened by chance I ask myself.
I set up a Monte Carlo simulation on Betfair SPs at Mr Flanagans average price which is a fraction over 8.0. It turns out that there is a chance that those poor numbers could have occurred by chance but that chance is 0.52%
Other questions started to occur to me prompted by Matthew’s comment on the previous blog entry. I am not too sure whether pairing tipsters will produce predictive results, it may be possible if you subscribe to the Machine Learning theory of pooling weak learners together to make a stronger learner. I may look at this later but in the meantime the thought did occur to me that originality may well be predictive. In other words do tipsters who have a higher rate of being the sole supplier of a nap tend to produce better results. Let me put that more simply, does a tipster who say 50% of the time is the only one napping a horse tend to do better than the guy who is stand alone only 20% of the time ?.
I had a quick look at this and it will come as no surprise that original thinkers so to speak did better. Just looking at nap bets those in the top half when it comes to providing original naps
16203 bets and a loss ROI to BFSP of -2.03%
Those in the bottom half of originality produced
12854 bets loss ROI of -3.4%
It would seem that if you are looking to follow a newspaper tipster how original they are is a starting point of choice.
Who is the most original ?
Morning Star Farringdon
Rory Delargy who ironically has blocked me on Twitter comes in second
An alternative to ranking on number of original naps would be to find the average number of naps each tipster shared with other tipsters. In other words The Scout is mid table on this ranking with 3.67 which means on average he/she when selecting a nap is sharing it with 3.67 other napping tipsters. This alternative order improved the dissection
Top half ie least sharing
ROI% = -1.49%
Bottom half most sharing
ROI% = -3.94%
Interestingly the most original on this metric is Mr Flanagan which means he is either original in his naps because they are so bad no one lese would dream of napping them or he is about to defy that 0.52% and come good ?.
Bottom of the originality list on this average metric is Lee Sobot Yorkshire evening post