Sunday, 1 March 2015

Advanced Poisson Guide For Sale

Courtesy of Twitter and Mark, it was brought to my attention last week that Hejik is selling an “Advanced Poisson Guide”. I was asked my opinion of this, but contrary to another of Jonny Grossmark’s unfounded beliefs, (he has several), Hejik and I are in not actually in partnership to take over the world, (evil laugh), and the mundane truth is that I don’t know him (Hejik) from Adam, although Hejik does actually exist, which is likely a plus. I therefore have no knowledge of such a guide.

I was a little surprised given Hejik's thoughts on Poisson less than twelve months ago (5th April 2014) where he clearly has major doubts about Possion being a good fit for football, and further admits that the perfect model does not exist:

There seems to have been much debate about Poisson on various forums of late and the danger would appear to be the perfectly natural beginners obsession with building the perfect model.
Such a thing does not exist as you'll well be aware and 'best guess' is a reasonably fair assessment of what professional bettors are doing.
As long as your 'guess' is extremely well educated you have half a chance at the books who are also, apparently surprisingly to some, employing 'best guess' pricing models, not psychic witches with crystal balls.
For what it's worth, I think it's important that we question how efficiently Poisson fits to football and observation would suggest it's far from perfect.
However, even without that information it's easy enough to arrive at the same conclusion when we consider that some of the biggest bookmakers in the world are actively promoting it as a method of beating them. I'm not sure they'd be so inclined to do that were it even near as effective as they suggest.
Note the comment that Poisson is a far from perfect fit for football and the suggestion that bookies promote Posisson because it is ineffective, so we have something of a voltafaccia, as we say in Italy.
As many readers will know, I use a modified form of Poisson myself, (unmodified Poisson is not a good fit for football) so this is an area of interest, and I am only too well aware that the quality of the output (i.e. probabilities and thus prices) generated is only as good as the input, which is the goal expectancies for both teams.

What is not clear at all from Hejik's sales pitch is where these goal expectancies come from. If you want a tool that accepts the two goal expectancies and generates match odds, you can find one for free at web site. I’m not sure if this calculator adjusts for the infamous zero-inflation problem, but turning input into output is the easy part.

Where any value would lie is in the calculation of the input parameters, i.e. accurate goal expectancy figures, and Hejik is very much aware of this fact, so my concern is around the maintainability of meaningful numbers. 

Hejik’s guide supports the top leagues in England, France, Germany, Italy, Spain, Scotland and The Netherlands, as well four lower English leagues down to Conference National level, and claims that they “require minimal input from the user to maintain them – 20 minutes per week is generous”.

From my own experience in maintaining numbers for the top five leagues in Europe, it takes a lot more than 20 minutes per league, so either the data Hejik is maintaining is relatively unsophisticated or I am extremely slow.

I like to think it’s the former, in which case the output will be similarly unsophisticated. I’m speculating here, but 20 minutes to enter match data for 11 leagues would barely be enough time to enter in the match results, and as has been much discussed previously here, match results are often misleading and need to be placed into context.

For a start, as is the case with brains, goals are not created equal, which means that even for this basic starting point, a certain amount of effort is required to make them useful data. One would hope that few would argue that the first goal in a match is more relevant than say the fifth goal of a 5:0 thrashing, although having read some of the nonsense that is spouted about goal times or historical head-to-head results having predictive value, it may in fact be more than a few!

The other concern I have regarding this guide is the lack of any evidence that it actually works, and somewhat related is that if indeed it does work, why would Hejik be selling it? Selling eggs from your tried and tested Golden Goose is one thing, but replicating and selling the geese themselves for just 47 each would seem to be an odd business move, aka shooting yourself in the foot. 

If anyone has some first-hand knowledge about the quality of this product, I would be interested to hear from them (anonymity guaranteed of course). £47 is not exactly a fortune, and while the guide may well be worth that amount for entertainment value alone, or for provoking ideas for your own model, I have my doubts that such a ‘one size fits all’ tool is profitable.

No comments: