&#8220;This illustrates the large problems I have with the gospel of advanced stats and prediction systems. <a class="twitter-hashtag pretty-link js-nav" href="https://twitter.com/hashtag/Bluejays?src=hash"><s>#</s>Bluejays</a> rotation ranked 16th in <a class="twitter-hashtag pretty-link js-nav" href="https://twitter.com/hashtag/MLB?src=hash"><s>#</s>MLB</a>&#8221; &#8211; Mike Wilner
This was Mike&#8217;s <a href="https://twitter.com/Wilnerness590/status/846384682841374720">initial reaction</a> this week to the release of the first half of <a href="http://www.fangraphs.com/blogs/2017-positional-power-rankings-starting-rotation-16-30/">the starting pitching segment of FanGraphs&#8217; positional power rankings</a>. The Blue Jays&#8217; group of starting options did indeed rank 16th, and that certainly is jarring given their quality one through five. But does that make it a moment to rail against a straw man, or might there be something we can actually learn from this?
The whole point of the statistical revolution in baseball &#8212; which Wilner, among broadcasters, was a very early proponent of &#8212; was to let numbers challenge our assumptions. We don&#8217;t recoil when we see numbers we don&#8217;t like, we investigate. Either the numbers are trying to tell us something we&#8217;re missing, or we can use our understanding to make the next set of numbers better and the current set make more sense.
In this case it&#8217;s all of the above.
Projections, <a href="https://bluejaysnation.com/2017/02/15/making-sense-of-the-rather-grim-projections-for-the-2017-blue-jays-season/">as I&#8217;ve written about</a> already this off-season, don&#8217;t like certain members of the Jays rotation very much. I think we pretty much understand that by now. Marco Estrada induces bad contact with an elite-spin &#8220;rising&#8221; fastball, a cutter, and one of the best change-ups in baseball, but projection systems generally see the BABIP he induces as wholly unsustainable. They probably have a little bit of a point, even &#8212; Estrada&#8217;s skillset likely shines a little brighter than maybe it should because of a touch of luck and a very good defence behind him &#8212; but the ERA in his projection in FanGraphs&#8217; piece is more than a full run worse than what Marco has actually allowed over the last two seasons.
This is not a player that the projection system handles well, in other words. But a rare case like that, to me, is almost &#8212; alllllmost &#8212; a feature and not a bug. This is a pitcher who is producing value in an atypical way, and if a team can find out what makes him successful, maybe they can find other guys who are being overlooked that can be successful in a similar way. His ability to seemingly buck his projections leads us into new and different ways of thinking about pitching and how to get batters out. That&#8217;s one of the most interesting things about this stuff. Nobody worth listening to is looking at that number and accepting it as gospel.
The other guys in the Jays&#8217; rotation are a little less interesting in this regard:
<ul>
<li>Marcus Stroman&#8217;s 3.2 WAR projection is fine.</li>
</ul>
<ul>
<li>Francisco Liriano gets dinged because of age and because of the fact that he had an abysmal 2016 season (provided you don&#8217;t do what most Jays fans have been doing and ignore everything he did up to the point where he was traded).</li>
</ul>
<ul>
<li>J.A. Happ has a little too much J.A. Happ in his track record to make the system full-on believe he&#8217;s really a the guy we&#8217;ve seen since he got Searage&#8217;d after a mid-2015 trade to Pittsburgh (though his 2.6 WAR projection is hardly outlandish).</li>
</ul>
<ul>
<li>Aaron Sanchez doesn&#8217;t have enough in his track record for his projection to line up with his 2016 performance (though his is still among the top 20 projections for any starter, and right in line with Cleveland&#8217;s Carlos Carrasco and St. Louis&#8217;s Carlos Martinez, which are very reasonable comps.).</li>
</ul>
In fact, other than being especially light on Estrada (1.9 proj. WAR, compared to 3.0 fWAR in 2016 and 4.0 RA9-WAR), the projections for the Jays&#8217; five starters really aren&#8217;t so bad. They&#8217;re probably even pretty good!
It&#8217;s after we look at those guys where this all starts to go sideways for the Jays. And rather than grandiose dismissals, it&#8217;s here where we should probably be paying attention to what the numbers are actually saying. Because the numbers are definitely saying something. They&#8217;re saying the same thing that Mark Shapiro and Ross Atkins have been saying about the club&#8217;s pitching situation since the day they arrived. They&#8217;re saying that depth matters.
We generally refer to a team&#8217;s &#8220;rotation&#8221; as its best five pitchers, but we all know that those aren&#8217;t the only guys who&#8217;ll make starts for a club over the course of a season. And because they&#8217;re not just fucking around out here, FanGraphs&#8217; rankings aren&#8217;t just about projecting a team&#8217;s front five, they&#8217;re about projecting the entire starting staff, using &#8220;manually curated playing-time forecasts,&#8221; along with &#8220;projections for rate stats from ZIPS and Steamer.&#8221; (They do this because, as Dave Cameron explains in <a href="http://www.fangraphs.com/blogs/2017-positional-power-rankings-introduction/">his intro</a>, &#8220;while forecasting systems have been shown to do better than most humans at forecasting production, humans win out when it comes to allocating playing time.&#8221;)
The funny thing about projections so apparently awful as to make one question the whole enterprise of advanced stats itself is that, according to the numbers they&#8217;re giving us, the Jays&#8217; top five pitchers (by innings pitched) actually stack up rather well with the rest of the league. At 12.9 projected WAR, they&#8217;re just a hair behind the Astros (13.1), which would make them the ninth best team in the majors, as far as front fives go.
As much as we may want to slather these five Jays starters in praise, that&#8217;s probably exactly around where they should be, because the rotations ahead of them, while maybe not having the Best Fifth Starter In Baseball (TM), sure as hell ain&#8217;t bad: the Nationals (17.4), the Mets (17.1), Cleveland (16.2), the Cubs (16.0), the Dodgers (15.9), the Red Sox (15.6), and the Giants (15.4).
Hell, if you really think Estrada is alone in getting screwed and want to bump up the Jays&#8217; total by a couple wins, you just put them right up there with the best of the best!
Ahh, but this. This is why the Jays are ranked 16th by these projections &#8212; the ranking of all MLB teams by projected WAR from their sixth starter and beyond:
LAD 4.1 
PIT 2.7 
CIN 2.5 
NYY 2.2 
SDP 2.2 
TBR 2.1 
HOU 2.0 
LAA 1.9 
MIL 1.8 
TEX 1.8 
MIN 1.7 
BOS 1.6 
BAL 1.5 
CLE 1.5 
NYM 1.5 
STL 1.5 
MIA 1.4 
DET 1.3 
PHI 1.3 
ARI 1.2 
KCR 1.2 
CHC 1.1 
COL 1.1 
OAK 1.1 
ATL 1.0 
SFG 1.0 
SEA 1.0 
WAS 1.0 
CWS 0.9 
TOR 0.6
The Jays lose a half win to damn near everybody because of their lack of depth. They lose a win-and-a-half to the deepest teams. And when we consider that the difference between the top ranked team overall and the 20th ranked team is just eight wins, how they fell so far here starts to make sense.
But then when we remind ourselves that the projections for guys beyond the top five are a understandably a little shakier and more prone to rounding errors, and that the playing time adjustments maybe aren&#8217;t quite so flawless, and that the Jays&#8217; outstanding 2016 rotation ranked 25th according to this exercise last season, we can maybe even feel a little bit of comfort about the whole thing.
If they&#8217;re healthy, the Jays&#8217; starters can pitch as well as anybody. If their depth gets tested things could get difficult for them. I think we probably knew that already, but it&#8217;s nice to see it borne out by the projections. Hey, and we maybe learned a little along the way, too!

"This illustrates the large problems I have with the gospel of advanced stats and prediction systems. <a class="twitter-hashtag pretty-link js-nav" href="https://twitter.com/hashtag/Bluejays?src=hash"><s>#</s>Bluejays</a> rotation ranked 16th in <a class="twitter-hashtag pretty-link js-nav" href="https://twitter.com/hashtag/MLB?src=hash"><s>#</s>MLB</a>" - Mike Wilner
This was Mike's <a href="https://twitter.com/Wilnerness590/status/846384682841374720">initial reaction</a> this week to the release of the first half of <a href="http://www.fangraphs.com/blogs/2017-positional-power-rankings-starting-rotation-16-30/">the starting pitching segment of FanGraphs' positional power rankings</a>. The Blue Jays' group of starting options did indeed rank 16th, and that certainly is jarring given their quality one through five. But does that make it a moment to rail against a straw man, or might there be something we can actually learn from this?

The whole point of the statistical revolution in baseball -- which Wilner, among broadcasters, was a very early proponent of -- was to let numbers challenge our assumptions. We don't recoil when we see numbers we don't like, we investigate. Either the numbers are trying to tell us something we're missing, or we can use our understanding to make the next set of numbers better and the current set make more sense.

In this case it's all of the above.

Projections, <a href="https://bluejaysnation.com/2017/02/15/making-sense-of-the-rather-grim-projections-for-the-2017-blue-jays-season/">as I've written about</a> already this off-season, don't like certain members of the Jays rotation very much. I think we pretty much understand that by now. Marco Estrada induces bad contact with an elite-spin "rising" fastball, a cutter, and one of the best change-ups in baseball, but projection systems generally see the BABIP he induces as wholly unsustainable. They probably have a little bit of a point, even -- Estrada's skillset likely shines a little brighter than maybe it should because of a touch of luck and a very good defence behind him -- but the ERA in his projection in FanGraphs' piece is more than a full run worse than what Marco has actually allowed over the last two seasons.

This is not a player that the projection system handles well, in other words. But a rare case like that, to me, is almost -- alllllmost -- a feature and not a bug. This is a pitcher who is producing value in an atypical way, and if a team can find out what makes him successful, maybe they can find other guys who are being overlooked that can be successful in a similar way. His ability to seemingly buck his projections leads us into new and different ways of thinking about pitching and how to get batters out. That's one of the most interesting things about this stuff. Nobody worth listening to is looking at that number and accepting it as gospel.

The other guys in the Jays' rotation are a little less interesting in this regard:
<ul>
 	<li>Marcus Stroman's 3.2 WAR projection is fine.</li>
</ul>
<ul>
 	<li>Francisco Liriano gets dinged because of age and because of the fact that he had an abysmal 2016 season (provided you don't do what most Jays fans have been doing and ignore everything he did up to the point where he was traded).</li>
</ul>
<ul>
 	<li>J.A. Happ has a little too much J.A. Happ in his track record to make the system full-on believe he's really a the guy we've seen since he got Searage'd after a mid-2015 trade to Pittsburgh (though his 2.6 WAR projection is hardly outlandish).</li>
</ul>
<ul>
 	<li>Aaron Sanchez doesn't have enough in his track record for his projection to line up with his 2016 performance (though his is still among the top 20 projections for any starter, and right in line with Cleveland's Carlos Carrasco and St. Louis's Carlos Martinez, which are very reasonable comps.).</li>
</ul>
In fact, other than being especially light on Estrada (1.9 proj. WAR, compared to 3.0 fWAR in 2016 and 4.0 RA9-WAR), the projections for the Jays' five starters really aren't so bad. They're probably even pretty good!

It's after we look at those guys where this all starts to go sideways for the Jays. And rather than grandiose dismissals, it's here where we should probably be paying attention to what the numbers are actually saying. Because the numbers are definitely saying something. They're saying the same thing that Mark Shapiro and Ross Atkins have been saying about the club's pitching situation since the day they arrived. They're saying that depth matters.

We generally refer to a team's "rotation" as its best five pitchers, but we all know that those aren't the only guys who'll make starts for a club over the course of a season. And because they're not just fucking around out here, FanGraphs' rankings aren't just about projecting a team's front five, they're about projecting the entire starting staff, using "manually curated playing-time forecasts," along with "projections for rate stats from ZIPS and Steamer." (They do this because, as Dave Cameron explains in <a href="http://www.fangraphs.com/blogs/2017-positional-power-rankings-introduction/">his intro</a>, "while forecasting systems have been shown to do better than most humans at forecasting production, humans win out when it comes to allocating playing time.")

The funny thing about projections so apparently awful as to make one question the whole enterprise of advanced stats itself is that, according to the numbers they're giving us, the Jays' top five pitchers (by innings pitched) actually stack up rather well with the rest of the league. At 12.9 projected WAR, they're just a hair behind the Astros (13.1), which would make them the ninth best team in the majors, as far as front fives go.

As much as we may want to slather these five Jays starters in praise, that's probably exactly around where they should be, because the rotations ahead of them, while maybe not having the Best Fifth Starter In Baseball (TM), sure as hell ain't bad: the Nationals (17.4), the Mets (17.1), Cleveland (16.2), the Cubs (16.0), the Dodgers (15.9), the Red Sox (15.6), and the Giants (15.4).

Hell, if you really think Estrada is alone in getting screwed and want to bump up the Jays' total by a couple wins, you just put them right up there with the best of the best!

Ahh, but this. This is why the Jays are ranked 16th by these projections -- the ranking of all MLB teams by projected WAR from their sixth starter and beyond:

LAD 4.1
PIT 2.7
CIN 2.5
NYY 2.2
SDP 2.2
TBR 2.1
HOU 2.0
LAA 1.9
MIL 1.8
TEX 1.8
MIN 1.7
BOS 1.6
BAL 1.5
CLE 1.5
NYM 1.5
STL 1.5
MIA 1.4
DET 1.3
PHI 1.3
ARI 1.2
KCR 1.2
CHC 1.1
COL 1.1
OAK 1.1
ATL 1.0
SFG 1.0
SEA 1.0
WAS 1.0
CWS 0.9
TOR 0.6

The Jays lose a half win to damn near everybody because of their lack of depth. They lose a win-and-a-half to the deepest teams. And when we consider that the difference between the top ranked team overall and the 20th ranked team is just eight wins, how they fell so far here starts to make sense.

But then when we remind ourselves that the projections for guys beyond the top five are a understandably a little shakier and more prone to rounding errors, and that the playing time adjustments maybe aren't quite so flawless, and that the Jays' outstanding 2016 rotation ranked 25th according to this exercise last season, we can maybe even feel a little bit of comfort about the whole thing.

If they're healthy, the Jays' starters can pitch as well as anybody. If their depth gets tested things could get difficult for them. I think we probably knew that already, but it's nice to see it borne out by the projections. Hey, and we maybe learned a little along the way, too!

Recent articles from Andrew Stoeten

Breaking News