Saturday, 29 October 2011

Cardinal Red - And It's All Over

The World Series Game 7 turned into a bit of a personal disaster for me, as I strayed from the strategy of laying the road team if they take the lead early. I actually used that strategy to get Green All Over, but for some reason couldn't just let it rest. Possibly because it was the last game of the year and I wanted to make the most of the opportunity, I then opposed the Cardinals at various prices as they extended their lead incrementally to 4 runs, but unfortunately the Texas Rangers could never get anything going, and in one game, my profits on the season were reduced by 40% to end slightly below those of 2010. What have I written in the past about greed? Unless it's a stellar weekend ahead, the Premium Charge will not be a factor this week. The Texas Rangers will be feeling even worse than me, after losing last year and coming so close to winning this year, trading as low as 1.02, with the Cardinals twice just one strike away from elimination in game 6. Ouch.

I received an e-mail from Scott who wrote:
Hi Cassini
Another good XX weekend...

As a trained mathematician, statistician, and football trader, what is your opinion on the statistical merits of using a database of previous goal times to make bets on the time bracket of (for example) the first or last goal of the game, e.g. first goal of the game after 25 minutes or last goal of the game after 70 minutes?

You can do statistics and back-testing on anything, but would you have low, medium, or high confidence in such bets if they are backed up by statistics or could one just as easily make the argument that while numbers for total goals, total shots, total shots on target, and match outcomes can be reasonably predicted within certain confidence intervals, the actual timing of those goals is largely random?
Hopefully Scott won't mind me publishing the question, but finding topics to blog about isn't always easy, and this is an interesting question. The answer is that this isn't something I have looked into at all. For me, while a team's rating is likely to hold true over several games, on any one day, anything can happen. By this I mean that selecting the top three teams over a Premier League season is a lot easier than picking a winner on any match day. It's why we watch football. So to break down one single game into even smaller segments and try to predict how things will play out in those few minutes seems even more prone to chance. In my opinion, the prices available on bets such as the first goal after 25 minutes, or the last goal after 70 minutes, are unlikely to offer any value. I'm not aware of any consistently profitable strategy using these bets, except by the laying bookmakers of course.
Somewhat related perhaps, and there was an interesting piece on the BBC web site about statistics yesterday, specifically Liverpool's poor rate of conversion of shots to goals. While the ratio of shots to goals varies across leagues and teams, from Barcelona's ridiculous high of almost 1 goal per 5 shots in mid 2010-11, to Serie A's approximately 1 goal in 12 shots, the BBC article mentioned Liverpool's recent game v Norwich City where they scored one goal from 25 shots. The EPL produces approximately one goal for every 8.5 shots, so whereas over a season, that 90 minutes could be expected to generate close to three goals, in one game, or one part of a game, anything can happen.

No comments: