Tuesday 19 March 2013

Chi tests, statistical analysis and self-improvement.

A very lucid comment was left after yesterday's post by Stephen:

Hi Rowan,

While I do agree with you that it is wise to have set rules for your portfolio I do think there are factors you are not taking into account with your 2 year time limit (disclaimer: any such limit will be arbitrary and the stuff below will get a little technical).

As an example let's take Summer of Football which I have in my portfolio as well.

Last year James produced 424 picks and we know his average odds is around evens. His ROI was 8.4%. To test whether he has an edge based on last year I have done the following:

I run 100 simulations in Excel of 424 variables with even odds and mean 0% (no edge). In betting terms we have 100 tipsters, who each have provided 424 picks and who has no edge. The results show 3 of these (3%) has delivered better results than James and 97 has delivered worse. Thus it is fairly safe to say that James has an edge.

If I do the same calculations with 9/1 as the odds I find that 27 tipsters with no edge will actually have outperformed James. I am sure you already know this but for higher odds services you will either need a larger sample, a higher return or some qualitative assurance to say if they are any good.

Conclusion: I think your 2 year time limit is good, however, it also need to be calibrated against the number of tips supplied. In my opinion, one of the main advantages of the new approach from Football Elite is thus that if Matt still has an edge (which I suspect) it will take a shorter time-period for it to be revealed and his seasonal results will be less luck based.


Well doesn't that just ooze common sense?!?

Of course Stephen is absolutely bang on when he asserts that we must, when considering a sample of bets to ascertain the presence or otherwise of an edge, take account of the number of selections supplied. Even with my limited insight into statistics and their interpretation, I can see that this is the case. The only conclusion that can be drawn, is that for me to impose an arbitrary two year "testing" period is a little silly, but I think Stephen is too polite to tell me in so many words.

This is an area where I know I have a weakness. I don't understand the processes that a half decent ststistical analysis involves, nor do I run simulations, chi tests, and the rest. This sounds like excuse-making, but I simply don't have the requisite time (perhaps there are those of you who do these things who are now screaming at me that it doesn't take any time at all!) to get to grips with this side of running a portfolio. Were I gambling for a living, then I would make it my business to learn how to do these things and I would begin to allow my findings to guide me as to the best way forward for my betting, but this wouldn't come naturally. My brain is more geared towards the arts - writing crap has never been a problem for me. One segment of my law degree was what is known as jurisprudence - the philosophy of law - which allows the student all sorts of scope to write a load of old bollocks providing he can back that bollocks up. Needless to say I excelled, passing that segment of the course with a rare First and no little flair if I do say so myself.

Anyway, I digress. The point I'm making is that my brain is not so efficiently programmed to get to grips with the sciences, maths or statistics, and therefore when I start to read about simulations, probabilities and chi tests, I struggle to keep up. That doesn't mean I'm not willing to learn mind you, and fully intend to do so when the time is right. In the meantime, I have to be guided more by instinct and "feel", and to me, right now, a two year period over which to assess a service as opposed to one year "feels" right.

In the meantime, if anyone can point me in the right direction of some good reading material/internet articles on statistical analysis, chi tests and the like, please do let me know. I'm serious about wanting to learn and genuinely feel that it is the next stage in my development as a gambler, so any help appreciated. Failing that, I'll be relying on old man Google.


Stephen mentions Matt of Football Elite in his comment, and Matt himself left a comment yesterday with a link to an article that he knows is relevant to the subject we're currently discussing. Having read it, I can say without doubt that it is very relevant and provides an interesting read. Click on http://www.rwbaird.com/bolimages//Media/PDF/Whitepapers/Truth-About-Top-Performing-Money-Managers.pdf. When reading it, just replace the term "Money Manager" with "Betting Tipster", as although the article is written from the point of view of "traditional" investment, the principles are equally as applicable to gambling as an investment. It really is an insightful read. For some reason the link appears not to work, but if you copy and paste it into your browser, it should be fine and is well worth doing.


Monday 18th March
I've been banging on about the fine recent form of On The Nose recently, but one racing service that has enjoyed a fine all weather season and start to 2013 is Northern Monkey. Subject to a 20p/£ Rule 4 deduction, finding an 11/1 was nevertheless rather sweet (Clear Praise - Wolverhampton). Elsewhere, no joy for the aforementioned On The Nose (0/2), and that was it on a quiet day.

Northern Monkey: Staked 1pt, +
On The Nose: Staked 1.25pts, -1.25pts.

2 comments:

  1. Does anyone know how equine investments is doing..in terms of results and getting prices?

    ReplyDelete
    Replies
    1. Equine still get every account restricted or closed. Results been good this year but always a struggle getting on.

      Delete