Ancient History

Former A's and Cardinals manager Tony LaRussa is generally credited with popularizing the current methodology for bullpen usage. Specifically, he and pitching coach Dave Duncan have been (incorrectly) pointed at as having invented the modern-day "closer" role.

From this somewhat gushing retrospective of LaRussa's career:

It was La Russa who paid obsessive attention to how certain batters did against pitchers and vice versa and therefore adjusted his strategy accordingly. And it was La Russa who elevated the role of situational relief pitching into a whole new dimension in this year’s playoffs by calling for his bullpen a record 75 times, and winning the World Series despite no starter going more than five innings in the National League Championship Series against the Milwaukee Brewers.

Ever since the early 1980's, major league clubs swiftly popularized matching up lefty and righty relievers over one-inning stints (or even shorter), until now we see every club in the majors following the same formula. For thirty years, seven- and eight-man bullpens see pitchers who go one inning (on relatively rare occasion four outs or two innings) unless the "long-man" or "mop up guy" has to come in to relieve a pitcher pulled early after giving up so many runs that the game is out of reach in the early innings. The long man is usually the guy not good enough to crack the back of the starting rotation, and is usually only called in to pitch in low-leverage losing efforts.

But back in the 1970's and 1980's, bullpens were constructed a little differently. Starting pitchers typically would aim for complete games. The average length of a pitcher's start was as high as 7.1 innings as recently as 1975. After the starting pitcher came out, instead of turning the bullpen over to three pitchers to pitch one inning each, it was common to hand the ball to a bullpen ace, or Fireman, who would pitch the rest of the game.

Two such pitchers included Rollie Fingers, who during his Cy-Young- and MVP-winning 1981 season appeared 45 times for the Milwaukee Brewers, but pitched one inning or fewer in only seventeen of those games. He totalled 78 innings that season, and so averaged over five outs per appearance. He had a 1.08 ERA during that season, earning 28 saves. Similarly, Rick "Goose" Gossage appeared in nine All-Star games and averaged over five outs per appearances as a relief pitcher also. Both Fingers and Gossage were enshrined into the Hall of Fame after dominant careers as Firemen.

But LaRussa's success in managing his bullpen based on batter splits caught on, and Firemen largely (actually, entirely) disappeared.

What does this have to do with the 2016 Astros?

Nothing. Except, the 2016 Astros have a modern-day Fireman -- a bullpen ace who comes in and shuts down opponents for two or three innings during close games. Not a mop-up long reliever. He's no Brian Moehler, or even a Scott Feldman.

His name is Michael Feliz.

A Developing Dominant Pitcher

Feliz's call up to the big leagues last summer was met with some surprise. The then-21 year old started the season at high-A Lancaster, but was called upon to provide long-relief depth as a young, powerful arm. Though he's struggled through a few rough outings earlier this season in mop up duty (blowouts in New York and Seattle), Feliz has made the most of his opportunities in May, shutting hitters down across 11.2 innings and earning a more prominent role in the bullpen.

Feliz's short time in the majors comes with small sample size caveats all around. He's logged just 28 major league innings and 20 this season; with so little experience, it's impossible to say whether or not trends that have led Feliz to better outings in May will be the norm for the rest of the season.

First off, Feliz has been throwing harder since the beginning of the season. This culminated in a fastball average of 97 mph in a scoreless, three inning outing against Cleveland where he picked up his first major league win. That average has plateaued over his last four outings, suggesting that Feliz's velocity increase may be the norm going forward.

Feliz has predictably generated more swings and misses in May with increased velocity. Whiffs on his fastball have jumped from 7% in April to nearly 13% in May, even though his fastball usage has stayed constant. In a bullpen without many power arms, Feliz's progressing velocity may have already made him the go-to relief option as a power strikeout pitcher, evidenced by Josh Fields' demotion a last week.

Feliz's slider has also shown promise this month. As a three-pitch guy (he also throws a changeup) Feliz has relied on his slider about 25% of the time in his short major league career. His slider usage hasn't changed much in May; it's just been a different pitch.

Feliz has increased the velocity on his slider, throwing it at an average of 84 mph in May. However, pitch location also seems to be a big factor in the success of that pitch. In April, Feliz's slider ended up mostly in the strikezone. Though this didn't negatively affect outcomes against hitters (who only hit .167 against the slider in April), he wasn't getting a ton of swinging strikes on the pitch either. Here's Feliz's pitch locations on balls in play and strikes in April:

Feliz’s slider location has been a different story this month. Instead of locating it in the zone, he’s been throwing it out of the zone to hitters more often.

The whiff rate on Feliz’s slider has jumped to 28% this month, making it a great complement as a wipeout pitch to his powerful fastball.

Feliz' first appearance in 2016 was a messy mop-up effort in his first appearance of the season, a 30-degree game during which starter Collin McHugh was pulled with one out in the first inning, having yielded five earned runs to the Yankees. Feliz allowed six runs to score in that appearance, and made a rocky appearance in his second start as well.

But since then, Feliz has allowed only one run and zero walks in fifteen innings, bringing opposing batters to their knees in the late innings.

There are strong reasons for continuing Feliz in a relief role. For one, although he throws three pitches, his changeup is a below-average offering at present, grading out as being worth -2.34 runs per 100 thrown so far during 2016. Unless he develops it further (which would be gravy!), this limits the offerings that he can make to present batters with different looks. A limited pitch selection means batters can adjust more quickly during subsequent plate appearances.

This 2-1/2 pitch repertoire is compounded by the fact that batters in general become more effective with the more looks they get at a pitcher during a game. Research by Michael Lichtman shows that pitchers are more effective the first time through the order.

From this perspective, a pitcher like Feliz with two fantastic offerings and one "meh" one could be awesome during his first time through the lineup, then experience a severe drop in results afterwards. So at present, a long relief, high-leverage role is ideal for Feliz, as it maximizes his effectiveness by preventing batters from adjusting.

The Astros still see Feliz as a starter in the long run, but manager A.J. Hinch says (via Tweet from Brian McTaggart) that development of that changeout will be paramount for him to become a starting pitcher.

Modern Fireman

Feliz is a major league reliever who was developed as a starting pitcher. As such, he has the endurance typical of a starter, which allows him to pitch multiple innings as a long man.

As hinted above, the Astros have trended towards using Feliz in the role of a high-leverage, long-inning relief pitcher. This season, he has appeared in 10 games, and pitched 20 innings, an average of two innings every appearance. Last season, only one reliever topped 2 IP/G, Rockies' pitcher Christian Bergman (4.50 xFIP). Anthony Bass (4.03) was next-highest with 1.97 IP/G. Both players were used in a more traditional "mop-up" role, and did not reach the level of dominance being achieved by Feliz (1.25). You have to go back to 2002 with John Halama to find a reliever who averaged more than 2 IP/G and had a FIP under 3.20. There have only been 36 relievers in the last twenty years who averaged 2 IP/G, and not a one of them has been able to achieve anything approaching Feliz' 2016 dominance.

But if you go back further, you find a different story.

• 1976 Jim Crawford: 88 IP, 27 games (3.3 : 1), 2.87 FIP
• 1976 Reggie Cleveland: 81 IP, 27 games (3 : 1), 2.79 FIP
• 1989 Dennis Lamp: 112 IP, 42 games (2.7 : 1), 2.86 FIP

From 1970 to 1990, there were 90 qualified player seasons during which a reliever made zero starts, averaged more than 2 innings per games, and had a FIP below 3.50.

From 1991 to 2016, there have been...eight.

If Feliz continues in his current role, he will be the first return of a true fireman in a very long time. It's a different game now, and the Astros' use of Feliz is a throwback to an older brand of baseball, one that is long overdue in returning.