Outside the Box

TOUT WARS: Battle of the Experts

I couldn't find a good page for Mat Olkin, but here is one with some stuff he's done.

The reason I incude this is that we had an e-mail exchange. Mat suggested that we do a similarity score where we compare player's season by season and then come up with an overall score, so for instance, two players could have the same career values, but arrive at them the exact opposite way. One peaks early and one late. With current sim scores, these guys would have the same value.

I thought of a sneaky way to do it differently. Part of the problem with adding the year-by-year scores and dividing by the number of years is that players with matches of 1000, 700, 1000, 700, 1000, 700 are the same as 850, 850, 850, 850, 850, 850.

They are similar, but I would say the second match is a better one. Just my opinion. A way to differentiate between these is to divide each score by 1000, so it is between 0 and 1. Then multiply them. The first group would be .343 and the second would be .377. Now if players are the same for 20 years or 8 we would like the same value, so we then take the Nth root of the result and multiply by 1000. .343^(1/6) * 1000 = 837 and the other group comes out to 850 (how about that). The only question comes in how to handle guys where their careers don't align at the beginning, say Vlad Guerrero and Bobby Abreu. Abreu debuted at age 22 and Vlad age 20.

My thinking is that 500 will be the minimum score between two seasons, so we'll just give those seasons a score of 500, or we could just use the seasons that match up I guess. I think I like the former better.

I might try and code this up and see what happens for some well known players. It may be that the scores are so far off that nothing of interest happens. We'll see.