We’ve all heard it. This side of the word “sleeper,” it’s perhaps the most run-of-the-mill, average, ho-hum, and downright trite word used in all of fantasy sports. You know the one I’m talking about… say it aloud with me, as you wear a loathing look of disdain: “regression.” There’s no way he keeps this up. He’s playing above his head. The underlying stats and history don’t suggest he can sustain this kind of pace. The league will adjust. In short, the regression monster’s coming to get him!
But should that really be the end of our conversation? I mean, it doesn’t take any kind of substantial analysis to say that a player that unexpectedly performs well will regress to the mean of his career norms, or more importantly, the mean of our expectations. As esteemed Yahoo! fantasy sports writer, Scott Pianowski, so often harps on, regression should be the beginning of the conversation, rather than the end. In other words, how much is Player A going to regress? Should we be looking to deal him at the peak of his value or ride the train, even though we think it will be traveling at a slightly slower speed from here on out? Do other people anticipate a bigger regression than we do, and thus we should deal for Player A? What types of sample sizes should convince us?
To answer these questions on a per player basis, I think the best place to check for anticipated outcomes is the Remainder of Season (RoS) player projections provided by databases such as ZiPS and Steamer. By viewing the RoS projections from these sites, we have a starting place that does three things: 1) it provides the most accurate representation of what to expect moving forward, 2) it reduces the recency bias we have in our minds when we value a players based on their to-date statistics, and 3) it helps eliminate arbitrary sample sizes that often cloud our judgment, ex. “last 14 days stat line.”
– Mitchel Lichtman recently did an in-depth statistical analysis on why using computerized projections that heavily weight longevity and career numbers is more accurate than basing your projections on season-to-date statistics. I highly recommend you go check it out. His findings show that even if you were trying to predict a player’s upcoming September, it is more accurate to predict that month based on a player’s projections rather than on the stat line accumulated from April through August.
With player projections as our starting point, let’s continue the discussion on regression and take a look on what we can expect on two of 2014’s hot starters.
Dallas Keuchel
ZiPS RoS: 87 IPs, 4 Ws, 63 Ks, 4.62 ERA, 1.42 WHIP, 4.00 FIP
Steamer RoS: 90 IPs, 6 Ws, 65 Ks, 3.95 ERA, 1.34 WHIP, 3.75 FIP
Yahoo owned- 82% ESPN owned- 93%
Keuchel is pronounced “KYK-ill” just to throw a wrench in the works as we all ignorantly chant “COOOOOOOOOCCHHH” at the Astros box scores. This man was shrouded in the ole “Minute Maid veil of secrecy,” early in the season, as no one paid attention to any pitcher that had “Astros” across his chest. Well, fast forward to the dog days of summer and we’re all paying attention. What’s interesting about Keuchel is, he’s doing one thing in particular that few pitchers have ever done. Namely, his Ground Ball percentage sits at 63.4%. I’m sure by this point in the season, the earth worms in Houston are checking the daily lineups to see if Keuchel gets the nod, so they can get the heck out of dodge. Seriously, that’s an insanely high rate. Since 2008 only 2 pitchers have bested that number over the course of a full season: Tim Hudson in 2010- 64.1% and Brandon Webb in 2008- 64.2%. Their respective ERAs were 2.83 and 3.30 in those season. In short, a historically high ground ball rate covers a multitude of sins. Much has been written about whether or not Keuchel’s start is legit, but for the ground ball rate alone I do believe that Keuchel can sustain more of his current rate and regress less than the standard projections indicate.
The one area that could be of concern for Keuchel is his walk-rate. Over the past two starts he has registered eight walks. He does throw his sinker just bellow the zone often times. As Keith Law of ESPN pointed out, hitters could begin to lay off that pitch more and cause Keuchel’s walk-rate to increase.
Keuchel’s bread and butter pitches are his sinker (40%) and his slider (23%). Those pitches are solid and if he keeps mixing them effectively, he should be able to keep hitters off balance.
I do expect the current BB/9 of 2.26 to go up a tick as the season progress. Despite his ground ball tendencies, I also expect the HR/9 to normalize from his current 0.43 rate which is 6th lowest in the majors. But the highly used sinker and the ensuing ground balls should continue to do wonders for Keuchel ratios.
With the aforementioned projections as our starting line (instead of his season-to-date numbers), and considering I anticipate Keuchel outperforming those projections, I no longer need to use the word “regression.” Expect Keuchel’s Remainder-of-Season stat line to be:
Projection- 95 IP, 6 Ws, 71 Ks, 3.62 ERA, 1.28 WHIP
Jake Arrieta
ZiPS RoS: 73 IPs, 4 Ws, 65 Ks, 4.16 ERA, 1.38 WHIP, 4.09 FIP
Steamer RoS: 77 IPs, 5 Ws, 72 Ks, 3.96 ERA, 1.32 WHIP, 3.84 FIP
Yahoo owned- 52%, ESPN owned- 66%
Here’s an interesting case of a guy that’s really come on strong as of the last three weeks. Not just because of his performance, but more importantly because of his usage. There was a maddening pitch count placed on him earlier this season that made it nearly impossible to roster him in all but the shallowest of leagues. We just could not have any certainty of him pitching past the fourth inning. Those restrictions appear to have been lifted, as he’s pitched seven innings in each of his past three outings. The win opportunities will continue to be suppressed due to the poor run support he gets from the Cubs, but that statistic is the most variable, so it should always be the last thing we look at when evaluating a starting pitcher. While we started with the projections that more heavily weight his last several years rather than his last 3 starts, let’s take a specific look at his last 100+ innings. Since joining Chicago in the middle of last season, Arrieta owns the following line:
108 IP, 2.82 ERA, 1.11 WHIP, 8.4 K/9, 3.45 FIP
Lending voice to statistics: he’s been absolutely stellar since joining the Cubs. We talked about environmental factors last week and let’s not overlook them now. I could certainly see them contributing to his success, as getting out of Camden yards and the AL East generally boosts a pitcher’s pitchers confidence (particularly when it comes to giving up the long ball). To top it off, Arrieta’s clearly been changing up his pitching mix by relying far more heavily on his cutter. He’s never thrown it more than 15% of the time, and now he’s doubling that rate. As Eno Sarris of Fangraphs points out,
“450 innings into a career that has seen results that don’t follow his upside, and still sporting a fastball averaging nearly 94 mph, Arrieta has velocity to lose and command to gain. Now he’s got his walk rate down to a career best at least partly due to throwing his cutter more often in fastball counts.”
All in all, we should take into account his current projections, improved surroundings, improved pitch mix, and the ditched pitch-count. Add those factors together, and I think we have to consider him a top-40 pitcher going forward, and he should be rostered in all leagues down to 10-team mixers. It’s amazing how a guy can be nearly irrelevant and sporting a 4.84 career ERA, and then be a solid add in 10-team mixed leagues not even a month later.
Starting with the above projections and working accordingly with the information we’ve gathered, the following stat line should be an appropriate projection from here forward:
Projection- 94 IP, 5 Ws, 88 Ks, 3.55 ERA, 1.24 WHIP
In conclusion, I do expect these pitchers to regress from their current paces. I’m not reinventing the wheel when it comes to terminology. Ultimately though, we must juggle a wealth of factors to decide just how much that regression will be and not let that become a blanket statement for players we should avoid. We must take that statement further. How much regression we anticipate is what we must figure out in order to bid, sell, or hold on any given player. Both of these pitchers should be serviceable moving forward. Neither should perform as poorly as their current projections nor as highly as their season-to-date stat lines. Gauging from the above projections, you must judge how that plays into your specific league. Do opposing managers value Keuchel and Arrieta more or less than this? Now it’s your turn to decide what their “regression” means for you and yours.
Cover Photo courtesy of BeGreen90, Flickr
