THE BOOK--Playing The Percentages In Baseball

Yes! Yes!

And yet, when I looked at their stats called “OPS Wins” and “Clutchiness”, I clicked on the “glossary” to see the definitions...and they aren’t there.

More work to be done…

It doesn’t matter at all.  He’s got such an enormous resource up, and in no time at all, that I’d consider the entire site Beta.  Any issues should be directed on David’s Forums site, as that’s the best way he’ll make it better.

Not to get snide, but you could also find most of those things in your last paragraph on the THT site.

Is it fair to think of WPA+ and WPA- as won-lost records? Frankie Rodriguez would be 12-6, Papelbon 11-6, Santana 17-13, Pujols 16-8, Ryan Howard 17-10, Jeter 15-9. It makes sense but the winning percentages seem a bit low (eg Santana). Is this because only half a win and half a loss are awarded each game? Would it make sense to double the player’s distance from .500 like so --

Ryan Howard
WPA+ 17.44
WPA- 10.00
Decisions 27.44
.500 record= 13.72-13.72
Games over .500 17.44-13.72= 3.72
Doubled= 7.44
Added to .500 record=21.16-6.28
Rounded= 21-6 record

This corresponds more closely to his offensive winning percentage (10.95 rc/g in 4.89 r/g league = .834).

For a team, if you add up all the win advancements (WA or +WPA) and loss advancements (LA or -WPA), you will get a total of about 230 WA and 230 LA.  Essentially, you get almost 3 times WA and LA as you get wins and losses.

However (W-L)/2 = WA-LA, so a team that is 109-53, their WA and LA would come in at 244 and 216 (WPA of +28, which is 28 wins above the .500 of 81-81 or 109 wins).  If you insist on calculating it as a percentage, that 244 to 216 would be .530, compared to the team win% of .673.

A player like Howard, at +17 WA and 10 LA, is truly great.  There are about 460 game advancement (GA or WA+LA).  Howard has 27 of them, meaning he’s 1/17th of the team.  If they all performed like him, the team would be 17*17 WA and 17*10 LA, or 289 WA and 170 LA, or +119 wins above average.  Of course, this is impossible, since the maximum possible is +81.  What happens is that a team of Howard/Santana-like ballplayers wouldn’t be able to get so many WA, simply because they would not be in so many close games.  While the average team LI is 1.00, a juggernaut like this would probably have an LI at 0.50 (just an illustration).  Now, all of a sudden, a team of juggernauts would not have 289 WA and 170 LA, but maybe 150 WA and 85 LA.  As you can see, the GA is 235, which is much closer to the 162 GP.  (I don’t know what the minimum number of GA you can have, but it’s probably pretty close to that.) This is a +65 win team, or 146 wins, or .901 team win% (say .750 off win%).

All that to say that I think you can get the offensive win% to lineup as you want, but you won’t get the W-L record to add up to the team record.  Or, if you want the individual W-L records to add up to the team record, the individual win% will be off (and you’ll certainly end up with negative losses in extreme cases).