Oddities in Historical DB

[Leafs67]

I'm sure there are lots of these, as obviously there are tons of players to keep track of, but I'm just pointing out some odd ratings in the historical DB that I noticed.

Started a game in 2001-02 with the Toronto Maple Leafs, and Alexander Mogilny, Gary Roberts, and Mikael Renberg are all rated as 1* players. I know none of them were still in their prime years, but Mogilny was coming off of a 43 goal, 83 point season and had 57 points in 66 games IRL in 01-02. Roberts was coming off 29 goals and 53 points, and had 21 goals and 48 points IRL. Renberg had 52 points in 71 games IRL. Seems like all of them should be more in the 2.5* range or higher, no?

[jumpSeat]

Originally Posted by Leafs67

Seems like all of them should be more in the 2.5* range or higher, no?

I would agree. From what I've seen, the historical database does a pretty good job of assigning peak values in and around career years, but then it tends to drop off too soon.

[JeffR]

Basically the same problem for all 3: career arcs where their best seasons were in the first 30-40% of their career, followed by longer periods where they played at a lower, but consistent, level (except for Mogilny, who goes up and down.) The original historical rating system didn't handle that type of situation well, and I seem to have compounded it in Mogilny and Renberg's case by rating them a little low. We've made some improvements in the system that lets me model that type of situation more accurately, but I hadn't re-rated any of them to take advantage of it yet. I just did that, so they should be better next year.

[jumpSeat]

Originally Posted by JeffR

I just did that, so they should be better next year.

Thanks Jeff, greatly appreciated!

[Leafs67]

Originally Posted by JeffR

Basically the same problem for all 3: career arcs where their best seasons were in the first 30-40% of their career, followed by longer periods where they played at a lower, but consistent, level (except for Mogilny, who goes up and down.) The original historical rating system didn't handle that type of situation well, and I seem to have compounded it in Mogilny and Renberg's case by rating them a little low. We've made some improvements in the system that lets me model that type of situation more accurately, but I hadn't re-rated any of them to take advantage of it yet. I just did that, so they should be better next year.

Awesome, that is so great! Really appreciate the level of detail in the historical games on the whole by the way, keeps me coming back every year