Re: NAIHL Ratings - Currently Based on 2010-2011 NHL Season
Posted: Wed Aug 03, 2011 12:24 pm
I just completed the "extended" version of the NAIHL ratings for both V1 and V2. The difference in the extended version is that it uses a wider range and does not pack as many players in the 60-80 range. For leagues that have used these ratings before, GM's will have an adjustment period because many of the players will have lower ratings than they are used to seeing. The hope (especially in V2) is that this "extended" set gives a more realistic end result in the scoring leaders than what we saw last year.
Everything in how things were done is the same. It is still 100% based on NHL results/stats from the 2010/2011 season. The only difference is the tables and formulas that convert NHL stats to SIM attributes were adjusted to give a wider margin between top and bottom players. My original intent was to go all the way down to 10 or 20, but the end result looked really funny. What I ended up doing for this file was to take a blend between a full blown extension to 10 or 20 with the original/traditional file. This should give an improvement over what we saw last year with minimizing the risk that the adjustment was too much.
I am done with ratings for the year and will not be making any tweaks to the formula's or files. However, I am interested in finding out how the difference between the traditional and extended version of the NAIHL ratings play out both in V1 and V2. If there are any commissioners who have extra time on their hands, this would be helpful to me: Run a full mock season using the traditional file and the extended file. Let me know what version you used, your league setting, and also the individual leaders page for the "traditional" and "extended" set of ratings. This data will help me figure out what adjustments/tweaks should be made in the tables for the ratings that will be done next year.
Everything in how things were done is the same. It is still 100% based on NHL results/stats from the 2010/2011 season. The only difference is the tables and formulas that convert NHL stats to SIM attributes were adjusted to give a wider margin between top and bottom players. My original intent was to go all the way down to 10 or 20, but the end result looked really funny. What I ended up doing for this file was to take a blend between a full blown extension to 10 or 20 with the original/traditional file. This should give an improvement over what we saw last year with minimizing the risk that the adjustment was too much.
I am done with ratings for the year and will not be making any tweaks to the formula's or files. However, I am interested in finding out how the difference between the traditional and extended version of the NAIHL ratings play out both in V1 and V2. If there are any commissioners who have extra time on their hands, this would be helpful to me: Run a full mock season using the traditional file and the extended file. Let me know what version you used, your league setting, and also the individual leaders page for the "traditional" and "extended" set of ratings. This data will help me figure out what adjustments/tweaks should be made in the tables for the ratings that will be done next year.