Starter Rotation Slots as Defined by ERA-, FIP-, and xFIP- | Astromets Mind

Friday, March 20, 2015

Starter Rotation Slots as Defined by ERA-, FIP-, and xFIP-

Is Noah Syndergaard a future #1?

Defining rotation slots based on ERA-, FIP-, and xFIP-.

            When talking about baseball, it’s common for the discussion to turn to starters and rotations. Even though the distinction is often being made arbitrarily, people like to refer to pitchers as ‘a number 2 starter’ or ‘an Ace’ or even ‘a great 5th starter’ to help define roles within a rotation, and help compare strengths of rotations. It doesn’t have to be completely arbitrary though, as we can use metrics to define rotation slots. But which metrics are best to use? I am not the biggest fan of pitcher WAR, and this has been done before with pitching bWAR anyway – although they took a different approach than I will here. Also, I wanted to use inning independent statistics, and since ERA-, FIP-, and xFIP- are league and park adjusted, I think they were the best choices.
            Considering there are 5 starters per team and 30 teams, I wanted to consider about 150 pitchers for this group – such that the top-30 would be #1 starters and so on. This meant that I had to lower the inning minimum to 80 IP as a starter for the season, which is about half of a season’s worth of starts, just to get the group to include 150 starters. When I lowered the inning minimum to 70 IP, the group increased from 153 to 161, and the ERA- max in 2014 decreased from 173 to 135 (with FIP- and xFIP- following a similar pattern), which seemed more realistic. Almost all teams will need more than 5 starters during the season, and only about 88 have qualified for the ERA title per year in the 2000’s, but it was a little surprising that I had to lower the limit to 80 innings as a starter to reach 150 starters. Also, lowering the inning limit to 80 IP as a starter led to an average of 147.5 starters/season over the last 20 seasons, or 149 over the last 13 seasons. This speaks to how valuable it is just to have a healthy pitcher who can consistently keeps the team close for 5-6 innings/game throughout the season. I added an extra group to the normal 1-5 starter slots in Table 1, listed under ‘Rank – 1-5,’ which is for the ‘True Aces’ of the majors.


Table 1 – 2014 slots, minimum 70 IP as a starter.
Rank
ERA-
FIP-
xFIP-
1-5 (Ace)
48-61
51-66
56-68
6-30 (#1)
63-80
70-86
69-86
31-60 (#2)
80-92
86-94
87-96
61-90 (#3)
93-100
95-104
97-105
91-120 (#4)
100-114
104-113
105-110
121-150 (#5)
114-135
113-125
110-121



            As you can see, of the 150 best starters who threw at least 70 innings in 2014, more than 60% in the group were: within 20% of the league average ERA, within 14% of the league average FIP, and within 13% of the league average xFIP. All of the pitchers within the group of 4th starters are already below league average by these standards, and 3rd starters push the limits of being average. The 5th starters group stretches from the limits of what’s acceptable to the limits of a team’s patience – the ERA story paints a much worse picture, and it’s harder for a team to overlook, than the xFIP story. More data is needed to get a clear picture, which is why I look back at the last 20 seasons in Table 2 (13 seasons for xFIP-). If you’re interested, an image of the distribution curve for ERA-, FIP-, and xFIP- is shown in Figure 1.
            First, let’s take a quick peek at where Mets starters ranked within these groups last year. The Mets had 5 starters who threw at least 137.1 IP in 2014, so they all easily qualify for the dataset. Jacob deGrom easily led the way in all three groups, and he qualified for the #1 starters group in each statistic (78 ERA-/76 FIP-/82 xFIP-). Jon Niese falls into the 3rd starter group by ERA- and xFIP- (both 98), but drops to the 4th starter group by FIP- (105). Zack Wheeler just missed the 3rd starter cut by ERA- (102), was a 3rd starter by FIP- (101), and reached the 2nd starter group by xFIP- (95). Dillon Gee had the lowest FIP- (129) and xFIP- (114) among Mets starters, and fell into the 5th starters slot for ERA- (115) and xFIP-, but missed the cut by FIP-. Finally, Bartolo Colon was a 5th starter by ERA- (118), but a 3rd starter by FIP- (102) and xFIP- (100). In summary, the Mets had the following: 1, 3, 4, 5, 5 by ERA-; 1, 3, 3, 4, 5 by FIP-; 1, 2, 3, 3, 5 by xFIP-.

Figure 1 – Shape of the distribution curve for ERA-, FIP-, and xFIP-.

For this graph, ERA-, FIP-, and xFIP- are on the y-axis, and the major gridlines on the x-axis separate the starter slots. Aces fall to the left of the dark line, and guys who should be on the verge of being cut fall to the right of the red line. Notice that within those lines the slope is pretty steady.


Table 2 – ERA- and FIP- slots based on data from 1995-2014 (3,000 pitchers), xFIP- slots based on data from 2002-2014 (1,950 pitchers), minimum 70 IP as a starter.
Rank
ERA-
FIP-
xFIP-
Top 3.3% (Ace)
(25)-63
(28)-69
(52)-74
3.3-20% (#1)
63-82
69-87
74-89
21-40% (#2)
82-93
87-96
89-97
41-60% (#3)
94-104
96-104
97-103
61-80% (#4)
105-117
105-113
104-110
81-100% (#5)
118-144
113-132
111-123


            I put the lowest total in parenthesis because that was just the lowest from this group, and an Ace could potentially be even better. Notice that xFIP has a smaller spread than FIP- and much smaller than ERA-, which is what we saw in Table 1. The picture becomes a little clearer when you have more data, and we can make a few rough guidelines from Table 2:

      1)   Aces are 60 ERA-, 70 FIP-, and 75 xFIP- or better,
      2)   #1 starters are 80 ERA-, 85 FIP-, and 90 xFIP- or better
      3)   #2 starters are 90 ERA-, 95 FIP-, and 95 xFIP- or better
      4)   #3 starters are 105 ERA-, 105 FIP-, and 103 xFIP- or better
      5)   #4 starters are 115 ERA-, 115 FIP-, and 110 xFIP- or better
      6)   #5 starters are 140 ERA-, 130 FIP-, and 120 xFIP- or better.

Of these rules, the upper limit put on 5th starters ERA- is probably the hardest to accept, as it sounds too high, and is a big jump over the #4 limit. But, for a Mets starter in 2014, a 140 ERA- was equal to just under a 5.00 ERA, and that’s not so high that a team who believed in the starter (or was paying him a lot) wouldn’t let him keep pitching.
            Of course, there is more to being a starter than just having a good ERA (or FIP or xFIP) when you pitch, you also have to be able to pitch for your team every 5 games, and deep enough into starts to not keep draining the bullpen, for an entire 162-game schedule. This list is nearly half starters who wouldn’t qualify for their seasons ERA title – either because they were injured, in the minors, or just stunk and were let go. Still, nearly 75% of the starters in Table 2 went at least 130 IP, which is about 2/3 of the season, and a valuable total for their teams. While these rotation slots are not based on an ideal sample of 150 pitchers with 162+ IP, they are based on actual results, so the guidelines are still useful as at least a rough estimate. At the very least, this should make you rethink the value of a guy who can throw 200 innings, even if he’s pitching at a rate 20% worse than league average.


  • 0Blogger Comment
  • Facebook Comment
  • Disqus Comment

Leave your comment

Post a Comment

comments powered by Disqus
submit to reddit