THE BOOK--Playing The Percentages In Baseball

Great comparisons. 

Also consider: Roy Oswalt 125, Marcus Giles 112.

WS has some value when comparing position players, or (maybe) pitchers, but never one to the other.  Has James said anything lately about his timeline for WS 2.0?

Did James just not realize this when he invented Win Shares?

Dan, click on my name for my views on Win Shares.

This is the final proof that winshares is a sissy girl.

Chicks dig the long ball.
Therefore chicks must hate pitchers.
Winshares hates pitchers.
Therefore winshares must be a chick.

Dan: Bill has accepted that he needs Loss Shares.  However, I’m not sure he’s figured it out on the pitcher-side yet.

Tango:  is there a reason to expect inclusion of loss shares to increase value of pitchers overall?

I see 3 reasons for the undervaluing, roughly in this order:
1) 35% share of value for pitchers is too low (should be at least 40%, probably 45-50%);
2) using same zero baseline for starters and relievers puts starters at disadvantage (ignores reliever ERA advantage)
3) marginal runs by great starters are worth a bit more than marginal runs for hitters, because they are concentrated in a few games.  If a starter is 40 runs above average (-1.6 RA/G), that’s worth about 14% more in wins than +40 runs from a hitter (+.25 R/G).

James also makes an adjustment for saves and wins, but I’m not sure what it’s aggregate impact on starters is, if any. It definitely increases value of closers, which is probably appropriate.

Guy, I was hoping that just as the WS is too low for starters, that LS would also be too low, so that with this lower baseline, the WS would look better. 

***

I have pitchers at 42% in WAR and “deserved” dollars (including service year discounts), which is right around where their actual salaries is.  I agree that James has an enormous failing in his model with the allocation of shares to pitchers, as well as the split between starters and relievers.

In my model, as well as actual salaries paid, relievers account for 10% of WAR and “deserved” salaries.  I think I looked once at James model, and he has also 10% for relievers as their share of WinShares.  That means that starters are getting something like 25% or 26% in Win Shares, whereas they should get around 32% or 33%.

Indeed, if you see Pedro with 25 Win Shares, that should really be 32 Win Shares.  And that certainly puts him more in line with what we think it should be (and of course, what WAR says he should be).

To add to Guy’s list, another problem is that the breakdown between fielding value and pitching value is done on the team level rather than the individual pitcher level.  This prevents good pitchers from getting the extra credit they deserve for their defense-independent contributions.

I believe that once when WS first came out I tried pretending that Randy Johnson was his own team and doing the pitching/fielding split on that level.  He came up with about 2 extra WS as a result.

Beckett, for example, if we add 30% to his 70 Win Shares, comes in at 90 win shares.  That’s now 10 ahead of Nick Johnson, instead of 10 behind.  Still only 3 wins, but better than what it currently implies…

Doesn’t Studes’ Win Shares Above Baseline correct for this? Four starting pitchers were in the top ten of 2008 WSAB for all players.

No question that WSAB corrects it, if studes uses the correct bench baseline.  I don’t remember if he does.  I seem to remember that his starter bench baseline was still too low.

As long as studes reports WSAB of around 57% for hitters, 33% for starters, and 10% for relievers, then he’s doing alright.