Saturday, January 14, 2006

Slow on the Draw

As a statistics junkie, there is one key thing that a careful analysis of Charlton's results brings up, namely the fact that we have drawn only one game this season.

I have written previously about the irrational 'value' placed on a draw by managers, and it is pleasing to see that despite a very inconsistent season, we sit on 28 points after 20 games, on course (amazingly perhaps) for 53.2 points, our highest ever Premiership total. The frustration the fans feel this season comes from the ten defeats (50% of the games) but perhaps we all need reminding that our goal should be maximisation of points, not minimisation of defeats since after all, 38 draws would often see you relegated.

In the unlikely event that I am ever Chairman of a football club, any manager of mine should know that they would face instant dismissal if they reveal in a press conference, "..we set out to get a point." Obviously with ten minutes left in a match, with the score at 1-1 and the team under the cosh, it makes sense to protect what you have. But at the start of a game, any attacking tactic deployed needs only to increase the probability of a win by just over half the amount it increases the probability of a defeat to make sense, thanks to three points for a win.

The Chelsea game next week is a prime example and it's fair to assume Curbs would 'take a point.' However, Chelsea will no doubt begin the game on Betfair at odds of approx 1/4 (which expressed as a probability is 80%) whilst the draw will be approx 5/1 (16.7%) and Charlton will be approx 30/1 (3.3%) implying punters expect on average Charlton to take 0.26 points from the game.

If Curbs approached the game with two forwards (as he no doubt will) but also two pacy wingers, and perhaps Ambrose in midfield (instead of Holland), I (and other punters in my view) would be willing to take a punt on Charlton winning at 12/1 (7.7%). This 4.3% increase in our probability of winning would need to be offset by an increase in Chelsea's probability of winning to 88.6% (80%+ (2 x 4.3%)) , which expressed as odds would be 10/77 or roughly 1/8. I do not believe that any sensible punter, upon seeing the line-ups, would back Chelsea at slightly better than 1/8, implying in my view this would be a perfectly rational tactic.

Paradoxically perhaps, this approach is likely to make more sense in games against Chelsea than against say, Birmingham because the probabilities are so mismatched to begin with. Managers however tend to see it the other way around, and they are wrong to do so. I'm not suggesting Curbs will suddenly discover the wonder of statistics, but as New Yorkers often say, "do the math."

Question: Which is the third team out of the 92, aside from Chelsea and Charlton ironically, to have only drawn one game this season? Answer: Wigan (whose fans one can assume are relatively happy with their lot, despite nine defeats).


At 7:14 PM, Blogger Pedro45 said...

..but what would happen in your scenario, if Charlton played with two forwards, two wingers, Ambrose in the middle, and then lost 9-0? Would the laws of probability allow for any equalisation later in the season if we followed the same principle away at Arsenal, Liverpool and Man Utd? I think not...

Good idea though!

At 9:54 PM, Blogger New York Addick said...

The accuracy of any probabilities can only be judged with hindsight, but I'm only suggesting that a cavalier approach to the Chelsea game might give us a chance of winning just 7 of 8 times out of 100. When we had our 4-2 win at Arsenal, or 4-2 home win over Chelsea, we achieved it by not giving the opposition much respect. In my view, if we go to Stamford Bridge and put ten men behind the ball as most teams do in the hope of getting a 0-0, they will have too much quality (plus we can't defend for toffee).


Post a Comment

<< Home