do players outperform in their free-agent year? phil birnbaum
TRANSCRIPT
Do Players Outperform In Their Free-Agent Year?
Phil Birnbaum www.philbirnbaum.com
Free Agent Performance Do players outperform in the year
before free agency ("contract year")?
Conventional Wisdom says "yes" By "trying harder" that year,
players immediately turn their effort into higher salaries
John Burkett John Burkett Component ERA
1999 5.44
2000 5.28
2001 2.86 (contract year)
2002 4.95
2003 4.41
But, Jeff Fassero Jeff Fassero Component ERA
1997 3.60
1998 4.10
1999 8.02 (contract year)
2000 5.25
2001 2.97
Jack Clark Jack Clark RC27 (avg-HR-RBI)
1985 7.15 (.281-22- 87)
1986 5.58 (.237- 9- 23)
1987 11.08 (.286-35-106)
1988 5.98 (.242-27- 93)
1989 7.11 (.242-26- 94)
But, Terry Pendleton Terry Pendleton RC27 (avg-HR-RBI)
1988 3.43 (.253- 6- 53)
1989 4.07 (.264-13- 74)
1990 2.81 (.230- 6- 58)
1991 6.73 (.319-22- 86)
1992 5.79 (.311-21-105)
Effect not obvious For every example of a sudden
contract-year star, there’s a counterexample of a contract-year collapse
Need a systematic study
How to figure it? What is evidence for a player
having a better contract year? Can’t go by the raw numbers
because of aging effects
Aging Free agents tend to be older
players Older players are on the decline A 35-year-old in his contract year
would be "outperforming" just by keeping his numbers the same was when he was 34
Methodology Used the "luck" algorithm Calculates expectation based on two
previous seasons, two following seasons
35-year-old compared to his numbers at 33, 34, 36, and 37
Takes care of regression to mean Predicts fairly accurately for all ages
The Study If players deliberately find ways to
outperform in their contract year, they should appear to be "lucky" by this algorithm
Calculated for all contract years to 2001
Thanks to Retrosheet for free-agent transaction info
Results: Hitters All contract year hitters, 1977-2001
Season outperformance: -0.1 runs Only hitters with 300+ batting outs
Season outperformance: +1.9 runs Same, normalized to 400 batting
outs Season outperformance: +2.2 runs
Results: Pitchers All contract year pitchers, 1977-
2001 Season outperformance: -0.2 runs
Only pitchers with 100+ innings Season outperformance: +0.6 runs
Same, normalized to 200 innings Season outperformance: -1.1 runs
No evidence of any effect Results indistinguishable from zero
Statistical significance not met For instance, standard error of pitching
estimate –1.1 runs is 0.8 runs Algorithm is not 100% precise
… but it’s pretty good: within 1-2 runs per season for regular players
No evidence (cont’d) Possible bias in data
Players who retire after contract year (because they lost effectiveness) are not counted, biasing the sample higher
Players who re-sign before the end of the season are not included in the sample
Including only regulars biases data in positive direction – players who are struggling won’t make 100 IP or 300 batting outs
More Results Batters, min. 300 batting outs, normalized to
400 batting outs Contract year: +2.2 runs Everyone else: +1.1 runs
Pitchers, min. 100 IP, normalized to 200 IP Contract year: -1.1 runs Everyone else: +2.6 runs
Other Studies "Baseball Between the Numbers," Chapter 5.3,
"Do Players Perform Better in Contract Years?" by Dayn Perry
Found "genuine phenomenon" of about half a win per season (5 runs!)
But – used "prominent free agents" – not a full or random sample
"Prominent" after the fact may have biased the results upward
Other Studies "The Influence of Free-Agent Filing on MLB Player
Performance," Atlantic Economic Journal, Dec. 2005, Evan C. Holden and Paul M. Sommers
Used 2003 only, but examined every player filing for free agency
Found no significant contract year effect, but found that performance decreased significantly in the year after
Effectively, the authors don’t discuss the "contract year" issue so much as the decline following
"… youngest players exhibit the smallest decline, largely because they (unlike their older counterparts) will have the opportunity to sign another contract before they retire."
Could the effect be simply due to player aging?
Other Studies "Shirking or Stochastic Productivity in Major League
Baseball?", Southern Economic Journal, April 1990, Anthony Krautmann
Checked all free agents, 1976-1983, who signed 5+ year free-agent contracts
Counted the number of players with significantly outlying performances in contract years, and following years
Found only the expected number of such players Conclusion: no evidence for the contract-year effect
"A Test of Additional Effort Expenditure in the "Walk Year" for Major League Baseball Players," Benjamin D. Grad
Regressed performance on a bunch of variables including contract year
No effect found for contract year
Pitchers with best/worst free-agent
years +44 – John Burkett, 2001 +39 – Darryl Kile, 1997 +35 – Danny Darwin, 1996
-57 – Jeff Fassero, 1999 -39 – David Cone, 2000 -33 – Kevin Brown, 1994
Hitters with best/worst free-agent
years +40 – Bret Boone, 2001 +38 – Albert Belle, 1998 +37 – Mark McGwire, 1992
-38 – Delino Deshields, 1996 -34 – Johnny Damon, 2001 -32 – Roberto Alomar, 1998