I Make Predictions as an Experiment

I noticed that Mark Lawrenson has picked us to lose 0-1 on Monday.   He could, of course, be right, but it seems like a stupid prediction.  We are playing well.  Aston Villa is playing poorly.  The only reason to pick Aston Villa to win (setting aside the views of optimistic fans) is to try to get credit for being unexpectedly right.

In my view, the goal of a prediction should be to get the prediction right and, perhaps, to learn something, not to look especially good when right and hope people forget when you are wrong.  As an experiment I have predicted this weekend's results based upon a simple set of rules.   I am posting this not because I believe there will be great interest in my predictions, but because I would like to be on the record so that I can know if I do well and not forget if I do poorly.

My rules are based upon the following assumptions:

First, I assume that eleven games worth of results gives real information that is more reliable than my feelings or “expert” analysis.  Therefore, I will pick the clubs who are doing well to beat the clubs that are doing poorly.

Second, only 33 games in 220 team games have been decided by a margin of two or more goals so I will never pick a team to win by two goals.

Third, clubs have scored 3 or more goals only 31 times in 220 team games so I will never pick a team to score more than two goals.

Fourth, so far 30 percent of the games ended in draws.  I will create a system that picks about 30 percent of the games to end in draws.

Fifth, there is a home field advantage.

Here are my rules:

I look at the two clubs’ current place in the table.  I subtract three from the home team’s rank.  If the modified table places are within three, I pick a draw.  Otherwise, I pick the team with the best modified table place to win.

To decide the score, I looked at goals allowed by my predicted winning team and the goals scored by my predicted losing team.  If they total more than the games played by both sides (22 this week) I pick a score of 2-1.  Otherwise I pick 1-0. 

If I am picking a draw I will just pick 1-1 because I cannot come up with a simple way to decide between 0-0, 1-1, and 2-2.  (So far this season, this would not have worked well because there have been a lot of 0-0 and 2-2 draws and relatively few 1-1 draws.  I do not believe this reflects any real tendency.)

Based upon this system my picks are

Chelsea-West Brom        2-1

Everton-West Ham         1-2

Leicester-Sunderland     1-1

Man City-Swansea          2-1

Newcastle-QPR                2-1

Stoke-Burnley                  1-0

Arsenal-Man U                 2-1

Palace-Liverpool              1-1

Hull-Spurs                         1-1

AV-Southampton            0-1

Let’s see how I do.