This has been about one and a half years in the making.
Last year, I pitched in to help with the Hybrid roster set, and despite all of the amazing work that came before by the OSFM crew (Knight, seanjeezy's pitch edits, and countless others), Willard, and teeds for making a brilliant roster set, the sim stats were spotty in some areas. And this was after Willard and teeds had already done things like balancing overpowered pitching ratings by adjusting clutch and /9 ratings downward.
So we decided to test certain things before releasing the roster last year by running a ton of sims, crunching through lots of data, and it really just amounted to patching some things. The best of the best hitters weren't quite getting their numbers, walks across the league were low as were stolen bases. We decided to take the upper echelon hitters and apply boosts "by feel" as well as applying a +5 universal boost to plate discipline among other things. We were mostly looking at league-wide stats and team stats to guide us because I felt that there was too much variance in stats of individual players.
Fast forward to this year and after having based defensive attributes on metrics (for the first time?) and adjusting speeds (both of which will feature in the Hybrid roster this year if not necessarily in the stock OSFM roster as well), I decided to take another look at the venerable ratings scales that Knight developed a while back. I started just by changing the stat columns from raw numbers (i.e. 25 HRs) to rate stats (i.e. 4.32 HR%) to make it fair for all guys regardless of playing time. After that I constructed some graphs from which linear fits of rate stats vs ratings were plotted and players were re-rated according to the resulting line equation. I started simming and adjusting the scales according to what I saw (based on trying to get the sim stats as close to 2015 ZiPS projections as I could). After about 3 iterations of refining the stats-to-ratings formulas (yes, I said formulas!), I am at a point where I honestly don't think it can get any better than this. One of the benefits of using "formulas" is that you can use a data entry/spreadsheet program like Excel to import any stats for any players and after entering the formulas in another column, you can have Excel do the dirty work. No more needing to go between several windows/files/screens and fit data into little boxes and "ballpark" that into ratings.
I am not going to publish my formulas quite yet although I likely will do that in this thread eventually. The reason isn't that I'm being secretive, but more that the formulas can get a bit complex and will confuse some people. Some formulas are layered and have IF,THEN arguments in them based on how nuanced I felt it needed to be. Anyway, I am posting a link to my most current google spreadsheet here so feel free to take a look at the ratings and comment. Please note that this has been the result of many, many sims so even if the numbers don't necessarily look familiar for some players (Gallo with 95 power will seem high at first glance, for example, and I didn't really like that at first either) doesn't mean it's necessarily wrong. Remember that no one, single rating determines everything and how a player plays is a combination of several things just as in real life.
Here are the ratings:
https://docs.google.com/spreadsheets...it?usp=sharing
I will also post some comparisons going forward so you can get a feel for how close to the 2015 ZiPS projections that these ratings will get your players. I will compare a couple of players that everyone is probably interested in right now...
Clayton Kershaw 2015 ZiPS projections...
6.75 H/9
0.55 HR/9
10.10 K/9
1.91 BB/9
Clayton Kershaw sim stats using my ratings (avg of 5 sims)....
6.66 +/- 0.82 H/9
0.58 +/- 0.17 HR/9
9.30 +/- 0.25 K/9
1.78 +/- 0.31 BB/9
Mike Trout 2015 ZiPS projections...
.301/.401/.577
.978 OPS
4.96 HR%
20.71 K%
13.48 BB%
85.2 SB%
Mike Trout sim stats using my ratings (avg of 5 sims)....
.293 +/- .025 batting average
.385 +/- .025 OBP
.585 +/- .039 SLG
.969 +/- .064 OPS
5.41 +/- 0.76 HR%
20.34 +/- 1.13 K%
12.27 +/- 0.50 BB%
85.3 +/- 7.0 SB%
The sim engine struggles with Craig Kimbrel and Aroldis Chapman-level K-rates as well as Joey Votto-level BB-rates, but other than a 99 rating in those categories only getting you so far, I couldn't be happier with how accurate these turned out.
I would encourage anyone, whether you are rating your players for OSFM or trying to rate a random player/fictional player based on being similar to one of the players in this spreadsheet, to feel free to use these ratings.