clock menu more-arrow no yes

Filed under:

Michael Fulmer and rookie ERA minus FIP

Can second-year improvements cancel out natural regression?

MLB: Detroit Tigers at Houston Astros Thomas B. Shea-USA TODAY Sports

Thanks to an excellent rookie season in which he won the American League Rookie of the Year and finished tenth in the AL Cy Young race, Michael Fulmer has joined the ranks of the most promising young talent in baseball. Fulmer finished 2016 with a 3.06 ERA and would have won the AL ERA crown with a decent final start of 2016. Instead, he gave up three runs and made it just 3 13 innings, falling three innings short of qualifying for the ERA crown and 0.06 runs behind Aaron Sanchez. In doing so, he missed out on becoming the first rookie pitcher since Mark Fidrych in 1976 to win an ERA crown.

Despite the outstanding rookie season, there were some folks who had Fulmer as a bit of a question mark heading into the 2017 season. In the modern era, baseball fans love to dig deeper than just ERA and wins, so nearly every stats-friendly fan knew about Fulmer’s relatively pedestrian strikeout rate and — even more frequently cited — the gap between his FIP (3.76) and his ERA (3.06). While Fulmer finished third among AL pitchers with as many innings in ERA, he was outside the top ten when it came to FIP (and his xFIP [3.95] ranked 17th in the AL alone).

While FIP (and xFIP) are hardly the be-all and end-all when it comes to conversations regarding pitchers, they have certainly proven to be more predictive than ERA when it comes to future performance, and there’s a reason we put a decent amount of stock into these two metrics.

So what has Fulmer done for a follow up to his excellent rookie season? He’s been even better in 2017. His ERA now sits at an even 3.00, and that’s after allowing a season-high five runs in his last outing, an outing that may have seen an inflated run total due in part to pitching to contact with a big lead (the Tigers led 10-1 by the end of the third inning). Fulmer had met the qualifications for a quality start in each of his first ten starts before Friday’s five-run, seven-inning outing, a total only matched by Clayton Kershaw and Dylan Bundy this season (both of whom have started more games than Fulmer). Fulmer’s ninth-place AL ERA rank isn’t quite as lofty this season, but that is due more to some fluky early-season ERAs that are yet to come back to earth (looking at you, Ervin Santana) than any slide from Fulmer.

If the AL leaderboard is sorted by FIP, Fulmer now ranks fourth, and he is only 0.01 behind Lance McCullers in third place. One could make a pretty strong case for Fulmer being among the top three pitchers in the American League this season.

So what happened to that regression for which Fulmer was due?

Well, the simplest explanation is that the season is still young. It is barely June and there are several months remaining in which Fulmer could see his ERA begin to creep up to the 3.75 range that his 2016 results suggested was his true talent level for last season.

The other, very possible, explanation is that Fulmer has simply improved in his second full season in the big leagues. Fulmer is just 24 years old, and he has made a grand total of 37 starts in his major league career. Fulmer saw his ERA decrease at the second year of each stop along his minor league career, and that pattern is continuing now at the big league level in 2017.

So with the June-5-small-sample-size caveat out of the way, let’s try take a look at a few charts to gain some historical perspective. Is there evidence that rookies who outperform their FIPs tend to take a step forward in their respective second seasons? Is that jump enough to cancel out the supposed regression for which they were due, or is there’s no real rookie secret to beating FIP?

Before getting started, it’s important to note that John Choiniere, when writing for BtBS in 2015, did a study asking: how does a pitcher’s age affect their FIP over- or under-performance? This study was based on a hypothetical raised on the always mentally stimulating Effectively Wild podcast, and Choiniere came to the conclusion that pitchers in the age 31-35 age bracket tend to slightly outperform their FIP compared to pitchers at other stages of their career, but the difference wasn’t stark.

This study will differ in that it will focus specifically on rookie eligibility rather than age. It will also be dealing with small-N samples rather than large-N samples as Choiniere did in his piece. It would be fascinating to see some large-N analysis on this question, but for the sake of this piece, there will be 10-12 players per chart with modifications set to specifically look at the question of rookies have who outperformed their FIPs and what they have done in their follow-up seasons.

Here’s the first chart. For this group of ten, the parameters were set at rookie starting pitchers since 1947 who threw at least 100 innings and had a FIP at least 50 percent higher than their ERA.

Post-integration Rookies with Large ERA-FIP Differences

Season Name Rookie ERA Rookie FIP Rookie ERA-FIP 2nd Year ERA 2nd Year FIP 2nd Year ERA-FIP
Season Name Rookie ERA Rookie FIP Rookie ERA-FIP 2nd Year ERA 2nd Year FIP 2nd Year ERA-FIP
1948 Gene Bearden 2.43 3.89 -1.46 5.10 4.70 0.40
1958 Red Witt 1.61 2.92 -1.31 6.93 5.11 1.82
1967 Rickey Clark 2.59 4.02 -1.43 3.53 3.41 0.12
1973 Steve Rogers 1.54 3.22 -1.68 4.47 3.34 1.13
1990 Scott Erikson 2.87 4.39 -1.52 3.18 3.76 -0.58
1992 Cal Eldred 1.79 2.81 -1.02 4.01 4.38 -0.37
1994 Bobby Munoz 2.67 4.03 -1.36 5.74 6.29 -0.55
2002 Denny Stark 4.00 6.11 -2.11 5.83 5.62 0.21
2006 Jered Weaver 2.56 3.90 -1.34 3.91 4.06 -0.15
2011 Jeremy Hellickson 2.95 4.44 -1.49 3.10 4.60 -1.50

This is a somewhat strange group. They are the outliers who managed to outperform their FIP by the greatest margins, and therefore were due for the heaviest regression in their second seasons. The numbers bear that out, as not a single one of the ten managed to improve their ERA, and only three managed to post an second-year ERA better than their rookie-year FIP.

As noted above, this is a somewhat strange group, however. Their rookie-year FIP actually had a negative correlation with their second-year ERA, as did their rookie-year ERA and second-year ERA. It’s a group that outperformed their FIP by a far greater margin than Fulmer and has only three names from this millennium.

Here’s a chart that focuses more on the modern era. These 10 rookie pitchers from the past decade threw at least 100 innings and had a FIP 25 percent higher than their ERA. (Junior Guerra was left off because he has thrown less than 20 innings in his follow-up 2017 season so far.)

Large ERA-FIP Difference Rookies, Past Decade

Season Name Rookie ERA Rookie FIP Rookie ERA-FIP 2nd Year ERA 2nd Year FIP 2nd Year ERA-FIP
Season Name Rookie ERA Rookie FIP Rookie ERA-FIP 2nd Year ERA 2nd Year FIP 2nd Year ERA-FIP
2007 Kyle Kendrick 3.87 4.94 -1.07 5.49 5.55 -0.06
2007 Jesse Litsch 3.81 5.14 -1.33 3.58 4.29 -0.71
2008 Armando Galarraga 3.73 4.88 -1.15 5.64 5.47 0.17
2009 Randy Wells 3.05 3.88 -0.83 4.26 3.93 0.33
2009 J.A. Happ 2.93 4.33 -1.40 3.40 4.32 -0.92
2011 Guillermo Moscoso 3.38 4.23 -0.85 6.12 4.5 1.62
2011 Jeremy Hellickson 2.95 4.44 -1.49 3.10 4.60 -1.50
2012 Miguel Gonzalez 3.25 4.38 -1.13 3.78 4.45 -0.67
2013 Chris Archer 3.22 4.07 -0.85 3.33 3.39 -0.06
2013 Tony Cingrani 2.92 3.78 -0.86 4.55 5.37 -0.82

Jeremy Hellickson is the lone repeat on the chart, and this time he is surrounded by a lot more recognizable names (maybe you’re a big Red Witt stan, but I know this was the first I’d ever heard of him). This time we actually have a positive correlation between the rookie-year FIP and second-year ERA, but it still isn’t much. In fact, the 0.15 correlation between rookie-year FIP and second-year ERA is significantly lower than the admittedly still low 0.50 correlation between rookie-year ERA and second-year ERA. However, it is interesting that the ERA-ERA correlation is higher than the FIP-ERA correlation when going from rookie season to second season. This is a 10-pitcher sample, with the biggest factor being they’re selection as FIP outperformers, but it goes in line with the premise that rookie pitchers might be able to make up for the difference between their rookie FIP and ERA by making the typical improvements a second-year players makes.

Here’s a final chart. It is admittedly set with parameters to specifically include Fulmer, but since he’s the focus of the article, it makes sense to have him in at least one of the charts. These are 12 rookie pitchers 25 years of age or younger who threw at least 150 innings, had an ERA+ of at least 135, and had a FIP at least 20 percent higher than their ERA.

Michael Fulmer-specific Parameters

Season Name Rookie ERA Rookie FIP Rookie ERA-FIP 2nd Year ERA 2nd Year FIP 2nd Year ERA-FIP
Season Name Rookie ERA Rookie FIP Rookie ERA-FIP 2nd Year ERA 2nd Year FIP 2nd Year ERA-FIP
1968 Jerry Koosman 2.08 2.70 -0.62 2.28 2.67 -0.39
1968 Stan Bahnsen 2.05 2.64 -0.59 3.83 4.20 -0.37
1970 Wayne Simpson 3.02 3.93 -0.91 4.76 4.45 0.31
1975 Dennis Eckersley 2.60 3.63 -1.03 3.43 2.72 0.71
1976 Mark Fidrych 2.34 3.15 -0.81 2.89 2.50 0.39
1977 Dzve Rozema 3.09 3.95 -0.86 3.14 3.71 -0.57
1980 Britt Burns 2.84 3.41 -0.57 2.64 3.44 -0.80
1987 Mike Dunne 3.03 4.05 -1.02 3.92 4.73 -0.81
1990 Kevin Appier 2.76 3.32 -0.56 3.42 3.06 0.36
2006 Josh Johnson 3.10 3.99 -0.89 7.47 4.58 2.89
2013 Jose Fernandez 2.19 2.73 -0.54 2.44 2.18 0.26
2016 Michael Fulmer 3.06 3.76 -0.70 3.00 2.97 0.03

While the parameters had to be set to almost Stefon levels of specificity to get Fulmer in there, it does leave an interesting group of pitchers. Names like Jerry Koosman and Jose Fernandez (rest in peace) jump out immediately, but a whole book could be written about this motley crew. For the purposes of this study, however, the focus should be that the rookie-year FIP to second-year ERA correlation (0.53) is now higher than the rookie-year ERA to second-year ERA correlation (0.47).

That goes in the opposite direction as our last group and goes to show how small the sizes are of the samples with which we are dealing. Since this is the final chart — and the lone one with Fulmer actually appearing in it — let’s take this one step further, though. Looking over the chart above, one number sticks out like a sore thumb: Josh Johnson’s second-year ERA. It’s massively higher than his FIP, but it came in only 15 ⅔ innings pitched thanks to an injury-riddled 2007. Since his innings total is so low, it should be accounted for by weighing his ERA-FIP less than the others, some of whom threw 200+ innings.

When the total amount of innings thrown in both the rookie season and second season are weighted, here are the collective totals:

Weighted Fulmer-specific Chart

Weighted Rookie ERA Weighted Rookie FIP Weight Rookie ERA-FIP Weighted 2nd Year ERA Weighted 2nd Year FIP Weighted 2nd Year ERA-FIP
Weighted Rookie ERA Weighted Rookie FIP Weight Rookie ERA-FIP Weighted 2nd Year ERA Weighted 2nd Year FIP Weighted 2nd Year ERA-FIP
2.63 3.37 -0.74 3.32 3.44 -0.12

The classic Halfway Between Two Conclusions. On the one hand, the second-year ERA (3.32) is nearly identical to the rookie-year FIP (3.37), which would suggest that the regression monster comes for us all. On the other hand, this group of pitchers did manage to beat their second-year FIP by a collective 0.12 — not inconsequential.

It would certainly be interesting to see someone spin off of this idea and do this sort of study for a large-N sample, maybe looking at all rookies ever, but in Fulmer’s case, looking at these smaller groupings of similar pitchers can be informative. It will be interesting to see if Fulmer is able to continue this stretch of excellent pitching throughout the 2017 season, and to see if 2017 rookies like Kyle Freeland, Ty Blach, and Antonio Senzatela can continue outperforming their FIP to give us another data point in the charts above.

All stats are current through June 4.


Jim Turvey is a fresh, new face to Beyond the Box Score. He also writes for DRays Bay, Call to the Pen, RotoBaller, and Insider Baseball. You can follow him on Twitter @FantasyBaseTurv.