22 January 2010

Cost Efficiency of Pitchers and Hitters in the Free Market

Remember to follow Camden Depot on twitter to receive any new updates (@CamdenDepot).

Yesterday I posted an analysis on organizational rankings using John Sickels' grades (AL, NL) and the Wang methodology to determine worth. In that column, I wrote my suspicion that Wang underestimated the value of pitching prospects about half of what they are truly worth. The basis of that thought was how free agent pitchers are typically a rather poor buy in comparison to free agent hitters (which was rather anecdotal) and that ranking systems and trades seem to view positional and pitching prospects somewhat similarly. Why would this matter? Buying free agent pitching may be so inefficient that having a stable of young pitching prospects may be twice as valuable even though they have a low success rate.

Upon that idea, I decided to engage in a quick study analyzing the cost efficiency of free agent pitchers and position players. It is all after the jump.

Players included in this study were all individuals that received more than 5MM in total salary signed in the 2007 off season (pitchers, n=26; positional players, n=34; t-test, alpha=0.05). Players with contracts longer than three years only had salary earnings to this point included meaning that the only seasons addressed in this study are the 2007, 2008, and 2009 seasons. Five pitchers and seven positional players are still under contract with more total money devoted to the positional players. This means the numbers presented here are still dynamic. Metrics used were MM/WAR and WAR/MM (to avoid issues with 0 WAR players. Cost efficiency based on previous WAR production was derived from information at fangraphs.

Results and discussion:
Cost efficiency was significantly higher (p=0.03) in the positional players (0.20 WAR/MM; 0.04 SE) in comparison to pitchers (0.092 WAR/MM; 0.026 SE). Both of these means are less than the rates by which they were paid (0.24 WAR/MM). This is to be expected as players are typically paid with respect to what they have accomplished as opposed to what they will accomplish. Some risk management is utilized, but the team with the most optimistic projection or the team who is positioned competitively to more greatly value wins will likely succeed in signing the player. Another way to look at this information is by MM/WAR, which is the typical way this information is expressed. For instance, pitchers from 2006 have wound up getting paid 10.66MM per WAR. Positional players, on the other hand, have earned 5.46MM per WAR.

After taking into consideration that several contracts have yet to end, it appears unlikely that the findings presented here would change greatly. We would expect though that the cost efficiency for both populations would decrease more. Based on projections of a 0.5 WAR decrease for every remaining season under contract, we see a reduction for pitching cost efficiency from 0.092 WAR/MM to 0.089 WAR/MM and for positional players a decrease from 0.20 WAR/MM to 0.19 WAR/MM.

In no way is this a conclusive piece of research. It is just a simple study used a somewhat small data set. The preliminary indications are that the system has already taken into account the free market cost to acquiring a pitcher versus growing your own. Pitchers are 53% less cost efficient than batters. In Wang's valuing of prospects, he has batter worth 49% more based on a weighted average of the top 100 prospects as ranked by Baseball America. If this is true, it casts doubt on how many try to use the Wang methodology to determine organizational worth. Wang's work appears to be best suited in comparing absolute worth of two different players outside of the pressures of the free market talent available to teams . . . more or less a closed system only considering MLB projection of MiL talent.

UPDATE: How would this change the prospect rankings?
Well, one way to do it would be to multiply the prospect worth as designated by the Wang approach by the factor difference between the the cost efficiency of pitchers and positional players. There are probably better ways, but this is a quick way to do it. Using this approach, pitching prospect values would be multiplied by 1.93 (using the WAR/MM numbers). The following would be organizational worth of the top 20 pitchers in each organization.

1. Rays
2. Rangers
3. Indians
4. Braves
5. Giants
6. Angels
7. Royals
8. Athletics
9. Dodgers
10. Brewers
11. Orioles
12. BoSox
13. Cubs
14. Rockies
15. Mets
16. Tigers
17. Reds
18. Padres
19. Pirates
20. Blue Jays
21. Astros
22. Nationals
23. Yankees
24. Marlins
25. Twins
26. Mariners
27. Cardinals
28. Phillies
29. ChiSox
30. Dbacks

Of the three lists I have shown in the past two days . . . this one actually looks the best to me.


Crawdaddy said...

Daniel Moroz posted a comment on this article from his website: http://camdencrazies.com/2010/01/22/valuing-prospects/

It is a long comment, but really the point to be made here is the contention of whether or not pitching is overvalued on the free agent market (in other words, pitchers are far less likely to live up to their contracts than positional players). Moroz comments that he thinks including relievers into this grouping is not appropriate as he thinks fangraphs does not properly calculate WAR for them (which is arguable because replacement level for a bullpen pitcher is not the same as replacement level for a starting 1B in terms of how it is calculated).

Anyway, I removed the relievers and it is still a significant difference (P CE rises to 0.12WAR/MM from 0.092WAR/MM). As I mentioned this is a small data set, so a more robust system should be evaluated. I don't think though that relievers are what is making this appear to be a misvaluation of pitching prospects in general.

Stotle said...

List definitely passes the smell test.