Wednesday, 12 March 2014

State The Average

Because of the continuous threat of police action, I am hesitant to say where I read the following statement:

I've spent time putting together some ideas on the futility of some peoples interpretation of game state for use as a betting strategy. I've nearly finished the first spreadsheet which calculates the % of results distributions from HT scores for any league and will be demonstrating why it's useless in some peoples hands along with some suggestions on how it could be made useful.
However, I do agree with the sentiments. All that a statistic such as "18% of away teams losing 0-1 at Half-Time come back to draw" (taken from this season's La Liga) tells you is what has previously happened, on average, over a certain period (in this case the current season). It's a fun fact, and no one in their right mind should consider putting money on fun facts. The idea that backing the draw is a value bet if it is priced better than 5.6 in the above scenario, is very dangerous.

Averages don't work very well in betting, and whether the draw is value at 2.0, 5.0, 10.0 or whatever is all down to what the goal expectancies are at that time, and these values can't be calculated based on irrelevant averages. Adjusted for time, they will most likely be close to the pre-game values, with adjustments made for events such as injuries,tactics, bookings and of course red-cards and goals.

Another question for adherents to the idea that 'game-state' gives some kind of secret edge is that the data is not secret, so why would you consider that you have an edge?  Anyone who cares to look, can see that "18% of away teams losing 0-1 at Half-Time come back to draw in La Liga" and they could go back further, or drill down on specific teams at a certain game-state, but the further you go back, the less relevant the team itself becomes since teams change their personnel and with a coaching / managerial change, also their playing style.

1 comment:

Danny Murphy said...

Another nugget Padrino.

Any chance of a post about shots sometime? Last night QPR had 21 shots away to Brighton of which 8 were on target yet lost the game 2-0. Brighton had 6 shots with three being on target.

It strikes me 21 shots away from home is good going so perhaps the score did not reflect the match on this occasion.

How important is the shots data in formulating goal expectation?