Debunking the second-half-surge myth
Fantasy owners being misled by belief that a big second half means a big next season
Editor's note: This story was originally published in the 2010 ESPN Fantasy Baseball Draft Kit. It is being republished unaltered here for your convenience.
After the All-Star break last season, Kendry Morales was one of the game's most feared sluggers. He ranked fourth in the majors in home runs (19) and RBIs (59) and batted .330 with a .989 OPS in the season's traditional "second half."
With that knowledge, might you push Morales a few spots higher in your rankings, following the long-held belief by countless fantasy baseball owners that a second-half surge often portends greater things to come the next season?
Here's some news for you: It's not always the case.
The rationale behind heavily incorporating second-half stats into the next year's projections might seem sound: They're the most recent data with which to evaluate a player, and when it comes to statistics, you want the most recent set, not a group from the distant past. For example, nobody relies significantly on 2006 stats for their 2010 evaluation purposes.
But the problem with relying too heavily on second-half stats is that, for one, they represent a small sample size -- or, at least, smaller than a full season of statistics. Sure, they beat a one-month split, but anything shorter than a full 162-game schedule can skew the data. For instance, who's to say that a player's second-half surge isn't just as likely to identify him as a slow-starting, hot-finishing performer as it is to identify him as a possible next-year breakout? What if the first-half/second-half splits were a direct result of an injury?
To get a sense of the relevance of second-half statistics, which are readily available and quoted throughout the industry these days, it's worth examining whether there's a strong correlation between finishing one season strong and maintaining that performance the next. Everyone seems to cite second-half stats, but shouldn't we look back afterward to see whether those same players kept it up?
First, in order to determine what precisely constitutes a "second-half standout," let's set a couple of ground rules for the study:
• Only statistics from 2004 to 2008 are included, providing a fair, five-year analysis. Why no 2009? Because the 2010 season hasn't happened yet, so 2009 second-half studs haven't yet had the opportunity to carry over their strong finish.
• Hitters had to have at least 400 plate appearances and pitchers at least 150 innings in their careers before their second-half outburst. Anything less than that and you're talking about youngsters and prospects, and their second-half data might be all there is to analyze.
• Hitters had to have at least 200 plate appearances and pitchers at least 50 innings during that standout second half. No cheapie second-half surges here; we want to avoid those guys who hit .400 with just 50 at-bats.
• Hitters had to have an OPS in that second half that was at least 100 points higher than these two variables: 1) that season's first half; and 2) their career OPS entering that season's All-Star break. For example, Carl Crawford took a .755 OPS into the 2005 All-Star break, which had upped his career OPS to .723. He then posted an .864 OPS in the second half of 2005, so he would qualify for the study. Why OPS? Because it's one of the more all-encompassing offensive statistics, as opposed to picking something like batting average, which is tougher to calculate in smaller sample sizes, or overall fantasy value, which could be skewed by unrealistic surges in stolen bases.
• Pitchers had to have a WHIP in that second half that was at least 0.15 lower than both their first-half number and their career number entering that season's All-Star break (just like with hitters). Again an example: Jorge De La Rosa took a 1.63 WHIP into the 2008 All-Star break, giving him a 1.72 career WHIP. His 1.33 WHIP in the second half of 2008 was easily more than 0.15 lower than both. And just like with OPS, I'm picking WHIP because it's generally regarded as the fairest of the primary Rotisserie categories for judging a pitcher's performance (certainly more so than ERA).
With those prerequisites in mind, there were 174 instances (108 from hitters, 66 from pitchers) included in the study. That's still a somewhat small sample size, all considering, but certainly enough to draw some conclusions regarding "second-half studs" going forward.
Improvement continued: In only 11 instances, the players improved their OPS (hitters) or WHIP (pitchers) from their second-half number in their subsequent seasons, including such big-name players as Matt Holliday (2005-06), Tim Lincecum (2008-09), Jake Peavy (2004-05) and Joey Votto (2008-09). This group represents 6.3 percent of all players in the study.
Repeat performances: In 15 instances, the players remained constant, registering an OPS within 50 points (hitters) or WHIP within 0.075 (pitchers) of their previous year's second-half number. These represent 8.6 percent of the players in the study.
Couldn't match the second half, but still improved: There were 15 cases in which the players fell short of their previous second-half OPS by more than 50 points (hitters) or WHIP by more than 0.075 (pitchers), but they had an OPS 100-plus points higher (hitters) or WHIP 0.15-plus lower (pitchers) the following season than their career-to-date number entering the previous All-Star break. Essentially, this means their second-half performances were absurdly good. I'd still call it a positive to be in this group because while they did regress from those second-half numbers, their follow-up performances remained noticeably improved when compared with their career standards. This represents 8.3 percent of the overall total, and includes such (previous or present) fantasy studs as Adrian Gonzalez (2006-07), Travis Hafner (2004-05), David Ortiz (2006-07), Aramis Ramirez (2006-07) and Johan Santana (2004-05 and 2005-06).
Returned to past form: There were 63 cases of players' OPS declining by more than 50 points (hitters) or WHIP increasing by more than 0.075 (pitchers) from their previous year's second-half number, but their subsequent season's OPS was still within 50 points (OPS) or 0.075 (WHIP) of their career number entering the All-Star break of that previous year. This group represents a whopping 36.2 percent, the largest percentage of these groups.
Regression: For a hitter to be placed into this group, his OPS compared to that second-half surge had to drop by at least 100 points, and his OPS compared to his career numbers entering the previous All-Star break had to drop by at least 50 points. For a pitcher, his WHIP that subsequent season had to rise by at least 0.15 compared to the second-half surge and .075 compared with his career number entering the All-Star break in question. So how many times did this happen? Forty times, or 23.0 percent of the total. Yes, 23.0 percent! That's nearly four times as many as the number that improved, and roughly the same number (23.2) from the first three categories combined.
Retired: Brad Radke retired following a splendid second half of 2006.
The neutral area: This includes the remaining 29 instances in which a player regressed noticeably from his second-half performance yet still produced slightly better than he had previously in his career (between 50-100 points in OPS for hitters and between 0.075 and 0.15 points in WHIP for pitchers when compared to the career numbers entering the previous All-Star break). An improvement? Yes and no, which is why they're neutral. When you draft a player based on his second-half stats, you expect him to approach those stats. These players clearly didn't, though they did show a slight uptick compared with their career numbers before that second-half surge. So this group, which consists of 16.7 percent of the players in the study, shouldn't be considered success stories, nor should they be considered failures. They're firmly in between.
The scorecard: In the study, 41 occurrences (23.6 percent) of the players fell in the first three groups (improvement continued, repeat performances, couldn't match the second half but still improved), which should be considered successes. The neutral grouping, which also would include Brad Radke, then consists of 30 instances (17.2 percent). So that leaves 103 cases (59.2 percent) in which the player either returned to past form or got worse.
You read that right: Approximately six out of every 10 players in the study failed to convert their second-half success into the same performance the subsequent year. In other words, even a flip of a coin would provide better odds.
The age factor
A follow-up thought: Perhaps a player's age might impact those odds?
It's a fair point, and as such I also analyzed the data accounting for players' ages as of April 1 of the year of their second-half surges. The chart below shows how players slot into the above categories, broken down by age group:
|21-23||4 (21.1%)||2 (10.5%)||4 (21.1%)||3 (15.8%)||2 (10.5%)||4 (21.1%)|
|24-26||6 (11.1%)||7 (13.0%)||6 (11.1%)||18 (33.3%)||7 (13.0%)||10 (18.5%)|
|27-29||0 (0.0%)||4 (9.1%)||4 (9.1%)||16 (36.4%)||13 (29.5%)||7 (15.9%)|
|30-32||1 (3.3%)||1 (3.3%)||1 (3.3%)||14 (46.7%)||7 (23.3%)||6 (20.0%)|
|33-35||0 (0.0%)||1 (5.9%)||0 (0.0%)||8 (47.1%)||6 (35.3%)||2 (11.8%)|
|36-43||0 (0.0%)||0 (0.0%)||0 (0.0%)||4 (44.4%)||5 (55.6%)||0 (0.0%)|
Now here's where things get interesting. It seems that the younger the player, the better his chances at extending his second-half development into the following season. That's especially true for players in the ages 21-23 group, of whom 52.6 percent improved (the first three chart classifications) and only 26.3 percent returned to past form or regressed.
Of course, that's really no surprise. After all, players who reach the majors by those ages tend to be the most special talents -- your Miguel Cabreras, Tim Lincecums and Justin Uptons. Sustained growth should be expected from them.
But it's that next age group -- the 24- to 26-year-olds -- that should surprise you. And I say this because it actually surprised me, too. One might think that 24- to 26-year-olds might be just as apt to extend their second-half growth patterns into the following year, but the numbers don't suggest that. Only 35.2 percent of players from that group landed in the first three positive ranges, while 46.3 percent returned to past form or regressed. In other words, right around the time you'd typically expect a player to reach his "prime years," history suggests there's no greater likelihood of him extending his second-half outburst into the following season.
That's disconcerting if you're targeting Morales, who enters this season at a ripe 26 years of age. And sure enough, a closer look at his 2009 splits indicates that his BABIP was .359 and home run/fly ball percentage 22.6 after the All-Star break, compared to .303 and 14.4 before it. That hints some of his performance might have been luck-driven, and it's why I'm not as apt to trust his breakout as I might Billy Butler's. Butler batted .314 with 13 home runs, 55 RBIs and a .925 OPS after the All-Star break in 2009, when he was only 23 years old.
And what about players even older than that?
Of the 27- to 29-year-old age group, only 18.2 percent fell on the plus side of the debate. By comparison, 65.9 percent lose the proverbial "coin flip." Seems that players at that age, which many consider the "prime years," are no more likely to maintain or build upon a hot second half than a player five years his senior.
Speaking of those seniors, here's a most disconcerting finding, in the possible event that you want to defend a veteran player coming off a second-half surge:
not 2nd half but improved
|Younger than 28||32 (34.0%)||44 (46.8%)||18 (19.1%)|
|28 and older||9 (11.4%)||59 (74.7%)||11 (13.9%)|
That's a steep enough split -- not to mention a healthy enough sample size -- that I'm inclined to completely toss out the window the second-half numbers of a veteran such as Derrek Lee, who batted .336 with 18 homers, 54 RBIs and a 1.092 OPS after last year's All-Star break. Now, the gut instinct of most people might have been to already weigh the 34-year-old's unexpected surge as worth less than they might the 24-year-old's, but in the event Lee's numbers caught your eye and have you thinking big things for him in 2010, have one reaction: They're irrelevant.
That might be the appropriate response to almost every exciting second-half split tossed in your direction this season, save for those of the under-23 players, who might have been obvious candidates for continued development anyway.
Unless, of course, you enjoy gambling on odds weaker than a simple coin flip.
Tristan H. Cockcroft is a fantasy baseball analyst for ESPN.com and a two-time champion of the League of Alternative Baseball Reality (LABR) experts league. You can e-mail him here, or follow him on Twitter @SultanofStat.
FANTASY TOP HEADLINES
- Hard Count: Make choices you can live with
- Matthew Berry's Love/Hate for Week 16
- Week 16 flex ranks: Trust your Bears
- Kaiser: More playing time, more value
MOST SENT STORIES ON ESPN.COM
2011 Fantasy Baseball Draft Kit
2011 Projections and Profiles
• Top 300 | AL-only | NL-only | Cheat sheets
• Catcher: Rankings | Preview
• First Base: Rankings | Preview
• Second Base: Rankings | Preview
• Third Base: Rankings | Preview
• Shortstop: Rankings | Preview
• Outfield: Rankings | Preview
• Designated Hitter: Rankings | Preview
• Starting Pitcher: Rankings | Preview
• Relief Pitcher: Rankings | Preview
• Top 50 prospects for 2011 | 51-100
• Tristan H. Cockcroft's Top 250 keepers
Talented Mr. Roto
• Cockcroft's dream team
• 2011 sleepers and busts
• Closer sleepers
• Jason Grey's plant-my-flag players
• Should you buy in: 20 youngsters
• Eric Karabell's "Do Not Draft" list
• Jason Grey's "10 I'm not buying" list
• Notable offseason moves | All moves
• Kings of Command: Pitching sleepers
• The real value of the Player Rater
• Rankings summit recap
Strategy/Primers• How best to apply the BABIP stat
• Deconstructing ERA