On July 20, 2010 at FanGraphs David Golebiewski authored “2010 BABIP-xBABIP Splits So Far” (http://www.fangraphs.com/fantasy/index. ... ts-so-far/
BABIP (batted balls in play) had been a sabermetric standard for some time until the recently improved xBABIP, which is hailed an even more reliable predictor of future performance.
Dave is an astute, and valued, contributor. He included this qualifier, “To get a better feel for how a hitter will perform in the future, it’s vital to take a good hard look at multiple seasons worth of performance.”
Given the parameters of the study, it seems to me a clear parallel to my Risky Hitters (now into its third year of tracking). (viewtopic.php?f=33&t=13641
Where I used drop in OPS as the indicator of which to be wary, Dave uses the completely independent xBABIP. The results are strikingly similar.
The thrust of Dave’s article was to provide two lists of players:
1) “And here’s how these guys rate in terms of the difference between their actual BABIP and their projected rest-of-season ZiPS BABIP. This is a more pure measure of who has been ‘unlucky,’ as ZiPS incorporates multiple years of data and regresses it:” Ah, “a more pure measure,” “multiple years of data,” and “regresses it.” The elixir that virtually guarantees a lock on predictability.
2) “And finally, here’s how they rank in terms of the difference between actual BABIP and projected rest-of-season ZiPS BABIP. This is a good measure of who has truly been “lucky”: Yesss!!! Not just been lucky, but “truly” been lucky.
To recap: It’s July 20, the top, and bottom, 25 players each have been unlucky and lucky. Armed with the irrefutable weight of the metrics, the unluckies are ORDAINED to get LUCKIER and the luckies, well, they are the new PARIAHs.
Dave doesn’t explicitly SPELL this out. He doesn’t have to. The elixir makes it clear.
I just mention the “elixir” and the “truly” to point out that we stat heads try to make the stats seem the ALL answer. In truth, they are just a piece of the puzzle.
Reader/Commenter, “Best post of the year, thanks so much for posting this. I have Quentin and Lind and felt their batting averages would definitely go up, but this gives me even more hope. Smoak could be a good sleeper for next year.”
Let’s measure the actual results.
Time didn’t allow for a full account of the impact of each of the five roto stats for our players. The goal is to see if, indeed, players turned the corner, as per the expectations. It was easy enough to go to the Game Log of each player to capture the OBP/SLG/AVG on July 20. Then again at end of season. How did our players fair?
The UNLUCKIES (we expect, no COUNT ON, these guys to progress):
July 20: .305/.386/.228
So, on average, each guy improved a whopping .003 OBP, .005 SLG, and .003 AVG.
(12 players progressed in at least 2 of the 3 cats; 13 regressed; all-in-all, NO GAIN)
The LUCKIES (soon to be pariahs):
July 20: .367/.482/.303
On average, regressed -.016 OBP, -.025 SLG, and -.017 AVG. These ARE significant losses. You would have lost ground in R/HR/RBI/SB/AVG. Still, the fall back end numbers aren’t that bad?
How did our Reader fair?
July 20: .344/.517/.244
Oops, what happened to that holy grail?
July 20: .273/.375/.218
Yesss, he did turn it around. Not that they are numbers that won the league for our commenter, but a turnaround nevertheless.
What are we to make of it all? We had STRONG, almost INFALLIBLE metrics. The players who had BABIPs that were too-good-to-be-true did, indeed, regress significantly.
However, the group that had been unlucky, simply remained unlucky. Huh…?
Well, that’s pretty much what I’ve been concluding from my Risky Hitters analysis and results. My studies show that if a player drops off more than 100 points in OPS in a given year, expect, on-average, a 14% gain back of the loss the following year. From the FanGraphs’ post, expect near-ZERO gain back from reported bottom of the barrel mid-season performance to end of year.
That doesn’t mean that players will NEVER return to form, just not for the immediate future.
BTW: I had begun review of players who showed gains of over 100 pts in OPS (average +139). Eerily, the results are similar to the FanGraphs’ post and review. Yes, as a group the players drop off, but several hold, and even BUILD on the gain. Indeed, roughly one-fourth continued to add, on average, another 29 points; examples, Votto another +43 (.874>.981>1.024), Cano +43 (.715>.872>.915), Tulowitzki +20 (.733>.929>.949).
Since both FanGraphs and my Risky Hitters project some drop-off for top improvers, x-metrics (expected) might help you project for 2011:
So, if you think you are going to get a guy down on his luck for cheap, a fabulous steal, and phenomenal return, who do you think is fooling whom? The only way these “losers” are a “win” is if you can get them so cheaply that 14% gain back on the dropoff provides “value.”
The “winners” are a little trickier. Since they have recently, and dramatically, improved their performance, their cost will go up as well. If the cost is going up, and it is likely they will fall off to a degree, it can be a tough call gauging the return. For my money, I would much rather have the prospect of a solid return (even if it costs me a little more) than to screw myself believing that I was going to get bounce-back numbers that have a very low likelihood of bouncing. Value is relative.
Haven’t quite figured out yet how to identify the players who will emerge from the dark to post that first 100-pt OPS gain, but you KNOW I am in search…