ARLINGTON, TX - MAY 29: Danny Duffy #23 of the Kansas City Royals throws against the Texas Rangers at Rangers Ballpark in Arlington on May 29, 2011 in Arlington, Texas. (Photo by Ronald Martinez/Getty Images)
A couple of weeks ago, on my podcast The 20-80 Report, we were discussing prospects likely to tumble down prospect lists during the 2012 season. One player who was mentioned was Kansas City Royals LHP Mike Montgomery. I generally agreed with the idea that Montgomery’s stock might decline this year—after all, Kevin Goldstein notably ranked him outside the top 101 prospects already, and he’s struggled for two straight years. A third consecutive inconsistent campaign in the upper minors would likely cause more and more prospect hounds to be skeptical.
In the middle of the discussion, however, I remembered something that caused me to blurt out the following:
“One thing I will say about Montgomery in the positive sense is that I remember a while ago kind of looking for…like…downside comparables for him. So I’m sitting there thinking ‘Well, let’s see if I can find a lefthander who throws 92 as a starter and is not very good, is like a back-of-the-rotation guy.’ And there isn’t any.”
Of course, the initial search I referenced there was far from scientific—I merely scanned some FanGraphs velocity leaderboards and tried to find a hard-throwing lefthanded starter that didn’t strike me as particularly impressive. Besides, I did that sometime in 2010, so it was out of date. But with my memory jogged, I decided to investigate the matter more fully. And as a result of said investigation, I have come up with what I refer to as the “Duffy Rule.” The best wording for it that I could construct is as follows:
Any lefthanded pitcher with enough talent to be given 100 innings in a season as a starter in a major league rotation, and who can maintain an average fastball velocity greater than 92 mph in those innings, is almost certain to be successful.
There’s some ambiguity in that wording, so before I delve into what leads me to believe that the above holds true, allow me to clarify the meaning here.
“Any lefthanded pitcher with enough talent to be given 100 innings as a starter in a major league rotation”—This is an equivalent to the sabermetric saying of “If you or I took a major league mound, our BABIPs would not regress to .300.” Do not take the rule to mean “Any lefthander that can average 92 mph as a starter will be successful in the major leagues.” The pitcher has to have enough talent for a team to stick him in their rotation for at least about half a season. To pick one example, Tigers lefthander Andy Oliver has made seven starts in the majors and averaged 93.9 mph, but he has a 7.11 ERA and 6.44 FIP—he didn’t have the talent baseline to cut it, or at least he hasn’t gotten there yet.
“Almost certain to be successful”—Success doesn’t mean the same thing to everyone, so it’s important to explain what I’m using it to mean here. In this case, it’s something like “average or better.”
Or, to break the rule down another way: A lefthander that meets the IP and velocity qualifications is just about guaranteed to be effective enough to stay in the rotation. Not only that, he’s almost certain to be better than a #5-quality starter, and is highly likely to be overqualified as a #4.
In a way, that’s all pretty intuitive, right? Power lefthanders are arguably the most prized commodity in baseball, so it makes a lot of sense that there’s some statistical backing for why they would be coveted. Conversely, though, that’s some strong language I just threw around. A “rule?” “Just about guaranteed?” “Almost certain?” Just how airtight is this idea?
The FanGraphs pitch velocity tables go back to 2002, now a full decade of data. In that decade, there were exactly 48 instances of a lefthanded pitcher throwing 100 innings as a starter and averaging 92.0 mph or greater in those innings (NOTE: any relief appearances in these seasons are excluded from the data presented in this article). They are:
Brett Anderson (2009, 2010)
CC Sabathia (2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011)
Clayton Kershaw (2008, 2009, 2010, 2011)
Cole Hamels (2010)
Danny Duffy (2011)
David Price (2009, 2010, 2011)
Derek Holland (2009, 2011)
Erik Bedard (2006)
Francisco Liriano (2010)
Gio Gonzalez (2011)
Johan Santana (2004, 2005, 2006)
Jon Lester (2008, 2009, 2010, 2011)
Jorge de la Rosa (2008, 2009, 2010)
Manny Parra (2008)
Matt Harrison (2011)
Oliver Perez (2004)
Randy Johnson (2002, 2003, 2004, 2005, 2006)
Ricky Romero (2011)
Scott Kazmir (2005, 2007)
Zach Britton (2011)
As you might expect, there are some premium names on this list—Sabathia, Kershaw, vintage Santana, late-career Randy Johnson, David Price, etc. But there are also some guys who have frustrated at times—Perez, Liriano, Kazmir, Britton, Duffy, Parra, etc.
Here’s a quick rundown of the average stats of these 48 seasons:
Those are pretty fantastic numbers, on the whole. On average, these pitchers generated a ton of strikeouts and swinging strikes, and they performed at well-above-average levels. Of course, that shouldn’t surprise—after all, 10 of the 48 seasons were by Sabathia alone, Kershaw adds another four, Santana another three, and before you know it, the average is very much weighted down by seasons of obviously great pitchers.
So let’s do something else—let’s look at what the worst performance in each of those numbers was.
14.6% K (Zach Britton, 2011)
13% BB (Clayton Kershaw, 2009)
1.56 K/BB (Zach Britton, 2011)
6.23 ERA (Derek Holland, 2009)
5.17 FIP (Derek Holland, 2009)
4.58 xFIP (CC Sabathia, 2004)
4.55 SIERA (CC Sabathia, 2004)
5.73 tERA (Derek Holland, 2009)
137 ERA- (Danny Duffy, 2011)
118 FIP- (Danny Duffy, 2011)
112 xFIP- (Danny Duffy, 2011)
6.3% SwStr (Brett Anderson, 2010)
There are four seasons that jump out as troublesome here. The first is Derek Holland’s 2009, a rookie year that saw him post an ugly ERA. That was mainly due to a HR/FB ratio spike in the hitter’s haven of Arlington, and that number would regress to the mean for the lefthander in following years. Therefore, xFIP saw that season as solid, pegging Holland as deserving of a 4.31 ERA, while SIERA had him at 4.25.
Another one that jumps out is Zach Britton’s 2011, due to his low strikeout rate. It was by no means a dominant season, but Britton posted a 4.61 ERA, 4.00 FIP, and 4.12 xFIP, numbers all close to league average.
Danny Duffy’s 2011 has the distinction of being the only season of the 48 that ERA, FIP, and xFIP regard as being below average—the rule is named after him due to his status as that exception.
You might notice that those three seasons are all rookie years. The lone veteran season is Sabathia’s 2004, his fourth year in the big leagues. But notice the numbers Sabathia’s on here for—a 4.58 xFIP and 4.55 SIERA. That’s the worst in those two categories across the 48 seasons! If you believe those numbers to be more accurate gauges of true talent level than ERA, tERA, and FIP, then they get to the heart of the Duffy Rule right there.
So, a) two of the most popular ERA-estimator metrics say that no pitcher deserved an ERA above the mid-fours, and b) while there’s a bad number here and there, there is no one season that you can single out and decisively say “Yeah, he was terrible that year.” Again, Duffy’s 2011 was the only season that had an ERA-, FIP-, and xFIP- over 100—and even it came with a 4.46 SIERA. If we take an average of those three numbers for each of the seasons, only four come out over 100—Duffy’s 2011, Britton’s 2011, Holland’s 2009, and Price’s 2009. All of those were rookie years, where a lot of the issues can be chalked up to early adjustment problems. If we take an average of ERA, FIP, xFIP, tERA, and SIERA, only two seasons—Holland’s 2009 and Duffy’s 2011—come out above 4.50. Any way you spin it, over 90% of these seasons have come out above average, and the few that don’t a) haven’t been too far below average and b) were rookie seasons.
So, that all looks good, but I used a lot of hand-waving and qualifiers there, didn’t I? Yeah, Holland’s 2009 wasn’t that bad if you wave away the HR/FB. Duffy’s 2011 had a decent SIERA. Britton’s 2011 had a good groundball rate. All the seasons had a redeeming number somewhere. Looking through the data, I began to wonder, heck, maybe most seasons have one somewhere. After all, in order to stick around for 100 innings in a season, a pitcher has to do something right. So I looked at all lefthanders that threw 100 innings in a season as a starter and averaged below 92 mph. Here are the average numbers from that group:
The first thing that jumps out among this data is that several of these averages are close to the worst numbers among the 92+ lefties. The K%, xFIP, and SIERA all aren’t far off. All three of the – metrics are above 100.
Looking deeper into it, one can unearth some pretty staggering numbers. A whopping 169 of the 327 sub-92 lefties (51.7%) had a SIERA above 4.55, the worst mark in the 48 power lefty seasons. An even-more-staggering 197 (60.2%) had xFIP-‘s of 101 or higher; Duffy’s 2011 was the only power lefty season above 103. Seventy-one (21.7%) saw their ERA/FIP/xFIP/tERA/SIERA average end up above 5.00, while just two of the 48 harder throwers had one above 4.50 (Another 94 were in between 4.50 and 5.00). I could go on, but I think I’ve made the point—the worst seasons from a 92+ lefty starter in the last decade would be about average for a 92- lefty starter in the past decade.
That leads to the next obvious question: Is this just limited to lefties? What about righties who throw over 92? Well, here are their averages (across 217 seasons):
These all fall pretty comfortably in between the groups of lefthanded pitchers. Obviously, there are plenty of elite seasons, but this group also contains guys like Daniel Cabrera, Tyler Chatwood, Jason Berken, Jorge Sosa, Ramon Ortiz, A.J. Burnett, Todd Wellemeyer, Luke Hochevar, Kyle Davies, Wade Davis, Jason Davis, Brad Penny, Kris Benson, and Claudio Vargas—guys who either never put things together or had down years mixed in with the good ones. Of the 217, 45 (20.7%) had SIERAs worse than the worst (4.55) of the power lefties, including 14 seasons above 5.00.
Again, it just seems like 92+ lefties are somehow insulated from these instances of extreme failure. This idea continues to gain steam if one delves more into the careers of some of the less-heralded pitchers on the 92+ lefties list.
The perfect example is Oliver Perez. Perez pitched nine seasons in the majors. He averaged just 88.0 mph with his heater in his final year (2010). Every other year, he was between 90.0 and 91.5—except 2004, when his fastball averaged 93.0. That season, Perez posted a 2.98 ERA, 3.45 FIP, 3.62 xFIP, and 4.5 WAR. He never had a FIP or xFIP below 4.00 (or even 4.20) in any other season, and he never even was able to come up with half the WAR total.
Manny Parra, like Perez, cobbled together a pretty effective season in his one 92+ year, racking up 2.3 WAR in just 166 innings and racking up 147 strikeouts. The next year, he lost a tick, and he posted a quarter of the WAR in nearly as many innings. Pitchers like Ricky Romero, Gio Gonzalez, and Matt Harrison saw a boost in their performance in 2011 as they got over 92 for the first time (though that could just as easily be chalked up to maturation). And, of course, Francisco Liriano has been a completely different pitcher at 92+ (2005-06, 2010) than 92- (2008-09, 2011), very similar to Perez.
This suggests there are basically two outcomes for a lefty starter who throws 92+: Either a) he becomes very, very good, or b) he ultimately loses the velocity. The worst sustained career of throwing 92+ as a lefty starter in the past decade is probably Jorge de la Rosa’s 2008-10 run, where he amassed 8.1 WAR in 436 2/3 innings.
So, what does this all mean? Well, it led me to come up with the Duffy Rule, so there’s that. But what are the implications of the rule?
Obviously, it portends success for guys like Duffy, Britton, Gonzalez, and Harrison going forward if they can keep their velocity above 92. It also gives me a lot of reason to believe in somebody like Felix Doubront, who looks like he may well get 100 innings this season and average 92+.
It’s not a finding to take lightly—there’s a fair amount of data that backs this up, as I’ve laid out. That said, we’re dealing with a population of just 48 92+ lefties, and that creates some problems. It’s tough for me to get behind any of this with too much conviction in light of the small number of existing cases. But 48 is not zero, and it’s not ten. My intuition tells me that if there was going to be a reasonably high percentage of power lefty starters that ended up as just inconsistent back-of-the-rotation guys (like sub-92-mph Oliver Perez, or Jonathan Sanchez), we would have at least seen one or two of them by now. It’s been a decade and we haven’t even come across one yet (At least, not unless Duffy or Britton hold their velocity and continue to struggle)! Perhaps that intuition is incorrect, and there are statistical tests that could reveal this as just an odd random variation, but it seems statistically significant to me.
It’s very likely that, at some point, there will be an exception to this rule—if nothing else, some rebuilding team will give a power lefty in the Andy Oliver mold a shot and stick with him, and he’ll have a ton of trouble getting outs despite throwing very hard. Chances are, some rookie will be rushed up to the majors at some point and just won’t be prepared—as I discussed earlier, many of the least impressive seasons of the 48 came from rookies, several of whom would go on to improve. As the population of available test cases grows, there will be more variation in the results—that’s just the way statistics works. In the meantime, though, the data we have seems to point to the fact that there’s something special about being a lefthanded MLB starter who can pump his fastball into the mid-90s, even more than the scouting community would have you believe. What is that something special? Your guess is as good as mine.