Mediating projections, Part I: Julio Teheran

Baseball fandom — or any fandom, really — is about disputes. Cheering for our favorite teams and players comprises part of the experience, but arguing with opposing fans and their teams and players makes up the rest. Pointless debates over everything from the Hall of Fame to the steroid policy to the playoff format shape us as spectators.

Ostensibly, the sabermetrics movement has eliminated much of the debate in baseball discourse. It theoretically emphasizes empiricism and objectivity to remove the opinionated elements of the sport's analysis. By doing so, it attempts to make peace among the diverse followers of this great game.

That hasn't really happened, of course, as disagreements still permeate every sphere of the sabermetric blogosphere, and will for the foreseeable future. WAR, the defining metric, comes in three different forms — Baseball-Reference's rWAR, FanGraphs' fWAR, and Baseball Prospectus' WARP. This creates no small amount of confusion and discord. (Here at BtBS, Bryan Grosnick has done his best to solve the three-metric quandary, with WAR Indexes that weight the three equally.) Additional differences exist in every area, dividing us and inhibiting our analysis.

One of said areas is also one of sabermetrics' most controversial fields: projections. They don't work as well as we'd like, they aren't particularly transparent, and worst of all (in light of the previous rant), they often contradict each other. To the unassuming baseball fan, there's nowhere to turn.

This predicament irks me, so I'd like to attempt to solve it. In the weeks to come, I'll look into the major discrepancies between some projection systems' view of players. I'll scrutinize each facet of the case fairly (#gottahearbothsides), and pick the one I arbitrarily objectively deem more likely to come to pass. Hopefully, this exercise will remove some of the deliberation on the issue and allow us to focus more on what really matters: the game.

I'll start by examining Julio Teheran, whom Steamer and ZiPS view quite differently:

Projection	IP	K%	BB%	HR%	BABIP	LOB%	ERA	FIP	WAR	WAR/200
Steamer	182.0	20.3%	6.3%	3.2%	.284	74.3%	3.82	4.10	1.3	1.4
ZiPS	204.0	21.5%	5.7%	2.5%	.276	76.8%*	3.22	3.37	4.3	4.2

*ZiPS doesn't provide strand rate projections; I estimated this using the data it does provide and assuming Teheran has a hit-by-pitch rate in line with his career norm.

Basically, they disagree on everything by fairly small amounts, and the net result is a huge disparity in projected production — the former thinks he'll be a middling role player, while the latter sees him as a budding star. Which should we believe? I'll break it down piece by piece.

Strikeouts

Teheran's first full season was in 2013 as a 22-year-old. In that campaign, he fanned 22.0% of the batters he faced — an exquisite figure, notably above the major-league starter average of 18.9%. Last year, that regressed a bit to 21.0%, while the MLB mark rose to 19.4%. This year, will it bounce back (as ZiPS predicts), or will it continue to fall (as Steamer predicts)?

I'd say the latter, because his peripherals support the decline. Mike Podhorzer's expected strikeout equation puts his 2013 and 2014 numbers at 22.1% and 20.8%, respectively. His strike, looking strike, and swinging strike rates stayed fairly stable throughout both years, but he suddenly accrued fewer foul balls: Whereas two years ago, his 28.8% F/Str came in at 27th in baseball, his 27.0% F/Str last year ranked only 69th.

Looking into his repertoire, we see that Teheran's four-seam fastball had the highest foul rate of any of his pitches in both 2013 and 2014, and he threw it less often in the latter year.

The quality of his four seamer (as measured by Foul%) didn't decrease that much, but the quantity of it overall did. That gives him fewer foul balls, which gives him fewer strikeouts.

But hey, if it's this simple, why wouldn't Teheran simply throw more four-seamers again in 2015? He made these changes for a reason — the velocity of the pitch dropped:

It still has decent bite, but it's not what it once was, and it probably won't come back. That means that Teheran will most likely sustain this negative trend; thus, Steamer's K% seems like the more probable outcome.

Walks

For a pitcher to avert free passes, he must throw the ball in the zone, or get the hitters to swing when he doesn't. Teheran has predicated his low walk rate on the former. While his O-Swing% has hung right around the MLB average of ~30%, he owns a sumptuous Zone% of 52.6% since 2013 — the seventh-highest in the majors.

Teheran's ability to pound the plate shows itself in his expected walk rate. His 5.7% xBB% mirrors his 5.8% BB%. With a sizable strike rate and in-play strike rate, he's certainly deserved the free pass level he's posted.

Additionally, those strikes haven't come as the result of luck. Based on his O-Swing% and Zone%, we'd expect 32.9% of his pitches to be balls, a number slightly below his actual 33.1% rate of balls (of the unintentional variety).

Nothing about this walk rate screams "fluke" to me; barring a meltdown — which, in fairness, can happen at any time to a hurler — it should remain as low as it has been. It seems that ZiPS has the more reliable projection here.

Home runs

This is one of the trickier elements of Teheran's case. For his career, he's allowed 48 home runs to 1,769 batters, giving him a HR% of 2.7%. Steamer thinks that'll rise significantly; ZiPS, not so much. The presence or absence of long balls can make or break a pitcher's season, so this is pretty significant. With which system should we side?

Well, on the one hand, Teheran gives up a lot of fly balls — which, paradoxically, means he may be less prone to dingers. Matt Swartz's research with SIERA showed that, among other things, pitchers with high fly ball rates have lower HR/FB rates. Moreover, Teheran plays in a park that generally suppresses home runs; Turner Field has never had a HR factor above 99. So this seems to endorse ZiPS' optimism.

On the other hand, there is some more concrete data: batted ball distances and angles, changes in which account for a lot of the variability in a pitcher's home run rate. Teheran hasn't fared too well in those regards, and the result has been — or should have been — more home runs. In 2013, he should have allowed 27.005 home runs, when he actually allowed 23; in 2014, he should have allowed 24.754 home runs, when he actually allowed 22. Those marks ranked 2nd and 1st in baseball, respectively, so they were rather atypical.

Adding those extra home runs in with the ones he's actually given up, we get a total of 54.759; dividing that by the 1,769 hitters he's faced creates a home run rate of...3.1%, closer to Steamer's projection. Obviously, that whole butterfly thing detracts from the validity of this, but the larger point remains: He'll probably give up more dingers in 2015.

Batting average on balls in play

Teheran's batted-ball profile lends itself to a low BABIP. Since 2013, only 36.2% of the balls hit off him have stayed on the ground — the ninth-lowest number in baseball in that span. Ground balls go for hits more often, so Teheran's .277 BABIP makes some sense. But can it stay that minuscule?

Whether it does or not likely depends on a factor outside Teheran's control: the defense behind him. It ranked 5th in baseball in UZR (and 7th in DRS) from 2013 to 2014, and 2015 might see it fall off a little. Jason Heyward, one of the best defenders in baseball, plays for the Cardinals now. Nick Markakis, whom Atlanta brought in to replace him, will allow many more balls to fall in for hits; as a fan of the Orioles, I can testify to the frustration that'll bring. Justin Upton left as well, and although he's hardly a strong fielder, Jonny Gomes and Zoilo Almonte won't turn any heads in his place.

Of course, this club still has Andrelton Simmons, whose résumé with the glove speaks for itself at this point. Keep in mind, however, that a shortstop doesn't field many fly balls, and those are the type that Teheran tends to garner. A poorer outfield defense behind him could entail a higher BABIP, which Steamer foresees. I won't argue that.

Strand rate

Over the past two seasons, only nine pitchers can top Teheran's 78.3% LOB%. At first glance, this struck me as a total fluke, as he appears to pitch the same across all situations:

Situation	TBF	K%	uBB%	BABIP	HR%	wOBA
Bases Empty	1034	21.6%	5.4%	.284	2.4%	.292
Runners On	624	21.3%	5.2%	.264	3.0%	.297

The strikeout and walk rates remain the same, and the higher home run rate offsets the lower BABIP. But these stats don't tell the whole story.

Teheran picks off runners. He's quite good at it — with 13 of them over the past two years, he leads the majors in pickoffs. It's so dangerous that his divisional rivals deemed it illegal. Eliminated runners can't come around to score, and sustaining this ability can therefore get Teheran out of jams.

This also has ripple effects elsewhere. If a base runner knows the pitcher could cut him down, he'll think twice before trying to swipe second. Not coincidentally, Teheran's limited base stealers in the aforementioned span; runners have taken off 4.9% of the time they could have, lower than the 5.6% leaguewide number.

Team changes could make that mark fall further this season. Evan Gattis and Gerald Laird — who caught 23 and 17, respectively, of Teheran's 63 games from 2013 to 2014 — won't return (the former's in Houston, and the club likely won't re-sign the latter). With -1 and 2 rSB, respectively, in their time in Atlanta, neither threw out runners particularly well. By contrast, Christian Bethancourt, their probable replacement, has a sterling defensive reputation; while he may be a bit raw, he'll still represent an upgrade over Teheran's former catchers.

Teheran already controls the running game rather well; an upgrade behind the plate may allow him to do even better. These two together give him the ability to strand runners consistently, and I see no reason why it should go away. No one can maintain a 78.3% mark, but a 76.8% — which ZiPS projects — seems reasonable.

Conclusion

So three of these go to Steamer, two to ZiPS. By my totally sound methodology, that means Teheran will get a little worse from last year, but not too much. Something in the neighborhood of 2.5 fWAR — a respectable but not outstanding number — will probably result from this. In what should be an ugly season for the Braves, he'll provide a ray of hope for the future.

Until humans evolve into gaseous forms, like the mighty robots, petty squabbles will likely define our time on this earth. Nevertheless, I'll continue to do my part to alleviate that as much as possible. Perhaps eventually, we won't need to quibble with the projections, because they'll be flawless; then again, what fun would that be?

. . .

All data courtesy of FanGraphs, Baseball-Reference, Baseball Heat Maps, and Baseball Prospectus.

Ryan Romano is an editor for Beyond the Box Score. He also writes about the Orioles on Camden Depot and on Camden Chat that one time. Follow him on Twitter at @triple_r_ if you enjoy angry tweets about Maryland sports.