One thing about playing fantasy baseball: it forces you to ask–and answer–some questions you’d otherwise never think
twice about. One of the ones a friend and I debated this past off-season was: Is Phillies outfielder Pat Burrell worth keeping for the 2004 season?

For what it’s worth, Baseball Prospectus 2004 had this to say about Burrell: “Burrell’s batting average dropped 73 points from 2002 to 2003, the 42nd-largest drop in history for a player with 500 at-bats in both seasons. That’s bad enough, but there’s only one other player above him who, like him, didn’t hit .300 before plummeting. What happened?”

Thing is, batting average is one of the most volatile stats in baseball; when it spikes, we call it a “career year,” and when it falls off a cliff we call it…well, a lot of things, not many of them printable. But take another look: While Burrell’s batting average declined in 2003, his hitting peripherals–patience and power–remained very close to levels which he’d previously achieved:

2000  23 PHI  474   260 359 463   129 203
2001  24 PHI  618   258 346 469   102 211
2002  25 PHI  684   282 376 544   117 262
2003  26 PHI  599   209 309 404   117 195

(PAT is a stat I use to measure hitter patience–it’s simply (unintentional walks)/(plate appearances). ISO is of course SLG-AVG.)

The big question facing the Phillies and fantasy owners around the world was: Would Burrell bounce back in 2004? Or, put another way, to what extent was Burrell’s decline simply a factor of bad luck, tough defenses, and other factors beyond his control?

I wanted to take a look at players who experienced declines similar to Burrell’s, following two years of relative productivity, and see how they bounced back the following year. So I looked at:

  1. All hitters from 1948-2003 (translated stats);
  2. Who had at least 350 plate appearances in three years running (Y2, Y1 and “Down”);
  3. Where their Down AVG fell at least 40 points below their lower AVG between Y2 and Y1;
  4. Where their Down PAT fell no more than 10 points from their lower PAT between Y2 and Y1, and
  5. Where their Down ISO fell no more than 20 points from their lower ISO between Y2 and Y1.

Then we see how they performed in their “Next” year after their “Down” year. Burrell meets these qualifications, which control for the possibility that a player suffered a drastic injury which prevented him from playing at all, or that his decline was just the player returning to normal after a career year.

This resulted in a pool of 107 players, at many different stages of their careers: Superstars young and old, solid contributors, injured players and healthy players, sluggers and slap hitters, prolific walkers and free swingers, young players and old veterans at the very end of their careers, players at all positions on the field, from almost every year
in the range considered. The most common ages represented in the set were 26 (13 players), 29 (16 players), and 34 (12
players). Harmon Killebrew appears twice in the set. Travis Fryman retired following his Down year–the only player in the set to do so.

Putting them all together, here’s what their average season was for the four years in question:

  107  Yr-2  567   311 375 502    82 191
       Yr-1  573   308 377 495    87 187
       Down  533   254 332 434    93 180
       Next  455   285 356 473    87 188

As a group, these players lost about 54 points in AVG in their Down year. However, their PAT actually improved by several points, while their ISO dropped about 10 points. Then, in the year following their Down year, their AVG regained 60% of what had been lost, while their PAT and ISO again remained about the same as the years before their Down year.

“Ah!” you say, “but doesn’t this set include a lot of players who were in their 30s and at the end of their careers? We wouldn’t expect them to bounce back as far, so the younger guys probably did even better!”

Fair enough. Here’s what we get when we divide the data into two sets, by age:


   57  Yr-2  566   314 373 514    75 200
       Yr-1  579   309 377 503    84 194
       Down  548   255 330 438    88 183
       Next  495   288 358 487    85 199


   50  Yr-2  569   307 377 489    89 182
       Yr-1  565   307 377 486    89 179
       Down  515   253 335 429    98 176
       Next  409   280 354 453    90 173

In fact there was very little difference between the two sets. The older players had a little less power, but a little more patience. Younger players regained a little more AVG (33 points vs. 27), but the gap is not large.

The other point to notice is that players lost playing time in their Down year, and continued to lose it in their Next year–despite their rebound. This is where older players are most distinguished from younger players, perhaps due to the “end of career” factor for a few of the oldest players. In any event, a huge drop in batting average one year perhaps shakes the confidence of a player’s manager in that player, and it takes more than one year for him to regain that confidence.

Back to Burrell: What did PECOTA think? Its-weighted mean projection for 2004 was that Burrell would lose a little playing time in 2004 and would regain about 60% of his losses from 2002-to-2003, with a .263/.356/.506 line. If we follow the guidelines of the numbers above, we might expect Burrell to add about 36 points of AVG, and his ISO to improve some, returning him to a .245/.345/.480 level. So PECOTA’s somewhat more optimistic, but the figures are in the same general vicinity. Of course, PECOTA takes into account many factors beyond what we’re studying here, and uses a more complex algorithm for making its projections.

There are 13 26-year-olds in the set above, and none of them are among Burrell’s top 20 PECOTA comparables. This is interesting because both approaches imply a substantial rebound for Burrell, which suggests that a profound drop in batting average–while other skills remain intact–might be due to factors beyond the control of the player. The similarity in rebound between the old and the young players also suggests this.

Of the 12 other 26-year-olds in the set, only Mickey Mantle, Donnie Hill and Ellis Burks didn’t see a significant rebound the next year. Mantle and Burks dealt with well-known injuries during their careers which obstructed their development, while Hill was a part-time player who is probably an outlier in this study.

Finally, Burrell is one of six hitters who qualify for the set based on their 2003 seasons. Of the other five, Craig Counsell–who suffered a dislocated thumb last season and had to deal with serious neck injuries for a couple years–has rebounded nicely; Bobby Higginson‘s average has bounced back but his power is gone; Jason Giambi, Kevin Millar and Bernie Williams have not yet rebounded.

To the relief of Phillies fans and Burrell owners everywhere, Burrell has followed Counsell’s example and not that of the others–presently he’s batting .276/.391/.483. That might change by the end of the season, but right now 2003 looks like a fluke, and Burrell’s career appears back on track.

(Darn the luck, I should’ve traded for him before the season began!)

Michael Rawdon is a programmer living in Silicon Valley. A long-suffering Red Sox fan (aren’t they all?), in the off-season he drowns his sorrows by reading science fiction and comic books. You can reach Michael Rawdon by e-mail at