Tweaking the Hall of Stats

23 Dec 2017 19:10

Adam Darowski's Hall of Stats is by far my favorite of the various Hall rating systems. Its main competitor, JAWS, suffers from two major flaws in my view: (a) an inherently arbitrary measure of peak value by focusing on a player's seven best years, and (b) an excessive focus on the HOF median, which leads to an ever-increasingly high bar for induction.

The Hall of Stats, in contrast, is an elegantly simple setup. Adjust every hitting season to 162 games, but give players half credit for unplayed games instead of full credit. Give catchers and relievers a boost and pre-1893 pitchers a penalty. Add bWAR and WAA*1.8 (WAA essentially measures peak value, the 1.8 gets WAR and WAA onto the same scale). Empty out the real Hall and fill it up again with the same number of players based on that formula.

This is so very close to how I would handle it. I thought I would try building my own Hall rating formula and see where the process took me. A few initial tweaks:

Blend fWAR and bWAR instead of using only bWAR. Tom Tango is an advocate for blending the two.
Instead of WAA, use WA2 (WAR above 2). Another idea taken from Tango. This better handles pitcher hitting.
Throw out the Union Association and Federal League as minor leagues. Honestly no idea if this is correct, just a preference.
Apply the schedule adjustment to everyone except pre-1893 pitchers.
Instead of 1.8*WA2 to measure peak, let's try WA2+WA3. WAR+WA2+WA3 will be called WAR+.

That's enough to start. No adjustments for catchers or relievers yet (I already know I won't adjust for relievers), and no additional penalty for pre-1893 pitchers. Right now the real HOF has 222 players inducted for their MLB performance. Let's start by seeing who among eligible players is in the top 222 by WAR but not WA2, and vice versa. The players shaded yellow are those who are in the top 222 by WAR+.

Folks tend to put a lot of emphasis on peak value but it's not at all clear to me that the WA2 guys are more deserving than the WAR guys. Koufax is the thorn in the side of any system that doesn't include a boost for peak value but other than him there are really no guys in the WA2 column who would be egregious omissions. Regardless, there are too many pre-1893 pitchers by WA2 and not enough catchers by either measure. So let's do those adjustments. Instead of going by WA2 and WA3, for catchers we'll do WA1 and WA2, and for pre-1893 pitchers we'll do WA3 and WA4. With these adjustments added, we'll call WA2 aWA2 for adjusted WAR over 2 (meaning WAR over 2 or occasionally 1 or 3). Here are the in&out guys by WA2 and aWA2.

That's a lot better, I won't really miss any of those guys in the first column. Here are the guys for WAR+ vs aWAR+.

Again, no big losses in the first column and I like all those catchers being included. Seems like the adjustments are the way to go. Now let's compare WAR to aWA2, highlighted guys are in by aWAR+.

Again, it's not clear to me that the WAR guys are less Hall-worthy than the peak guys. But, Tango advocates for WAA/WA2 as a better HOF starting point than WAR, and he's way smarter than me. So let's go with aWA2 or aWAR+. Which is better? Well, is the yellow group from the WAR column better than the unshaded group from the aWA2 column (plus Chet Lemon, who oddly is in by both WAR and aWA2, but not aWAR+)? No real must-haves from the aWAR+ group, whereas again I like the extra catchers from the peak group. Let's go with straight aWA2 over aWAR+. Here is the comparison of the in&out guys in my Hall of aWA2 vs the Hall of Stats.

Again, not immediately clear that one group is better than the other. I do like Barnes, Ward, and Caruthers being in (I know Adam would too), and I don't mind losing any of the HoS guys.

Ward, Caruthers, Whitney, Uhle, Spalding, and to a lesser extent Kaat, Gooden, Cooper, and Grimes all benefit from being pitchers who could hit (the Hall of Stats doesn't count pitcher offense in WAA).

Barnes, Caruthers, Stovey, McGraw, Kiner, Hahn, Spalding, and Jennings are all peak guys who benefit from aWA2's reduced emphasis on longevity.

Kaat and Lolich benefit from excellent fWAR totals compared to bWAR.

Everyone other than Whitney, Barnes, Kaat, Lolich, and Jennings has a Hall Rating above 90. Lots of guys at 98 and 99.

The ones that stand out to me personally as names I don't hear a lot about in HOF debates are Whitney and Uhle, two good hitting pitchers with good peripherals. Whitney actually had the two best single-season K:BB ratios ever until 1994; he still ranks 4th and 5th.

Here are the top 222 HOF-eligible players by aWA2, by position.

Pitchers

Catchers

First basemen

Second basemen

Third basemen

Shortstops

Leftfielders

Centerfielders

Rightfielders

Designated hitters

And finally, the Hall-worthy currently ineligible players by aWA2.

So, did we learn anything? I learned that I can get a Hall that's the same basic shape as the Hall of Stats with I would argue fewer arbitrary adjustments. I learned you can take WAR out of the formula entirely without losing any must-have players. I learned that introducing fWAR seems to have a surprisingly small effect compared to other tweaks like swapping WAA for WA2. Anything else jump out at you about the process or the results?