2004 were better, for one very obvious reason…
I’m switching to fWAR for this part of the analysis. This is simply because Fangraphs has better tools for splitting production into rotation and bullpen components, than Baseball Reference. But whichever metric you choose, the 2004 team was the better side, though the gap is smaller by bWAR. In fWAR, however, the 2004 Diamondbacks ranked 13th in the league, with a total value of 10.6 fWAR; the 2021 team came in dead-last, at 4.0 fWAR. The relievers for the two seasons were similar in (lack of) production, both coming in below replacement level. 2004 were -0.8, while 2021 were -1.2. So, by a process of elimination, it’s the rotation which was responsible for almost all the different, so let’s start there.
There’s a case to be made that Randy Johnson was the best pitcher ever to play on a terrible team. He made 35 starts, averaged fractionally over seven innings per start, and had a 2.60 ERA. That combination made him worth 9.6 fWAR – only one National League pitcher (Ben Sheets, 8.5) was within three wins of the Big Unit that season. I took a look at the 11 other teams to lose 110 games in the past sixty years, and who was the best pitcher on each by fWAR. Here’s the full list:
- 2021 Diamondbacks: Merrill Kelly, 2.4
- 2021 Orioles: John Means, 2.5
- 2019 Tigers: Matthew Boyd, 3.2
- 2018 Orioles: Alex Cobb, 1.5
- 2013 Astros: Bud Norris, 1.9
- 2004 Diamondbacks: Randy Johnson, 9,6
- 2003 Tigers: Nate Cornejo, 1.6
- 1969 Padres: Al Santorini, 2.0
- 1969 Expos: Mike Wegener, 1.8
- 1965 Mets: Jack Fisher, 2.4
- 1963 Mets: Al Jackson, 1.3
- 1962 Mets: Al Jackson, 4.0
You have to go back almost the full sixty years to find a pitcher who was even one-third as valuable on as bad a team, as Johnson was. He was as productive as the four most recent “aces” on 110-loss rosters, COMBINED. There is a very credible – indeed, almost inarguable – case to be made that Johnson was unjustly robbed of a sixth Cy Young award. Let’s play the old Pitcher A/B/C game among the trio who received first-place votes that season, though this one is largely obvious:
- Pitcher A: 2.98 ERA, 214.1 IP, 218 SO, 1.157 WHIP, 145 ERA+, 5.7 fWAR
- Pitcher B: 2.60 ERA, 245.2 IP, 290 SO, 0.900 WHIP, 175 ERA+, 9.6 fWAR
- Pitcher C: 3.49 ERA, 237.0 IP, 206 SO, 1.145 WHIP, 124 ERA+, 6.5 fWAR
Yes, A is Roger Clemens, who won by getting 23 first-place votes. B is Johnson, who got eight first-place votes. And C is Roy Oswalt, who got one first-place vote and forms a strong case for drug-testing BBWAA voters. The reason for this unfortunate miscarriage was it took place back in a time where voters regarded W-L record as a credible metric. Clemens was 18-4 and Oswalt 20-10, while Johnson was a mere 16-14. Worth noting: across those 14 losses, his ERA was a hideous… um, 3.64. The league average ERA in losses that year? 8.17. Hell, Johnson’s ERA was 18% better in his losses than the overall league average. [Clemens in his losses: 5.11; Oswalt in his losses: 6.67]
Of course, those 14 losses was almost entirely not Johnson’s fault. In all but two of those, the Arizona offense scored two runs or fewer in support. with a total of just 22 runs – an average of 1.57 per game. The five no-decisions were perhaps even less his responsibility. Johnson had an ERA of 2.19 in those, and got an average 2.80 runs of support. Overall, the last starting pitcher to have an ERA+ of 170 or greater, and still lose 14 or more games, was Dazzy Vance of the 1930 Brooklyn Robins (17-15, 2.61 ERA). And they finished eighteen games above .500 for the year. But here’s one final stat to prove Johnson’s dominance. He was worth 9.6 fWAR. Every other D-back pitcher combined that year was worth 1.0 fWAR.
But we shouldn’t forget Brandon Webb, who was worth 2.6 bWAR. While little more than a quarter of Johnson’s value, it would have made the D-backs #2 in 2004, the best pitcher on virtually every other 110-loss team in the expansion era – including the 2021 D-backs. But the drop-off thereafter was painfully steep, with only one person passing 0.3 fWAR (knuckleballer Steve Sparks was worth 0.8 in 18 starts). The 2021 D-backs didn’t have anyone at the level of Johnson, obviously, but Kelly, Madison Bumgarner and Zac Gallen were all better than most of the 2004 rotation. Below, is a chart showing all the starters for both seasons, along with their fWAR, in descending order of starts.
The win percentage splits between the rotation and bullpen were very similar. Starters won .298 in 2004, compared to .303 this year; for relievers, the equivalent numbers were .354 and .349. The main difference was how the number of decisions broke down. In 2004, the starter took the W or L 114 times, compared to only 99 this season. A significant factor is the average length of start: in 2004, the Diamondbacks’ starters threw a total of 96.2 more innings than in 2031. The bullpen can’t get a decision if they’re not in the game. The increased work for the current relief corps shows up in a clear Win Probability difference by the bullpens: 2004 were -245%, while 2021 came in at a whopping -915%.
Rather than ERA or fWAR, this is perhaps the best metric to use for comparing the two relief corps. Similarly to the chart above, the one below shows the number of games and Win Probability by all the relievers used this year [I have excluded the five position players who pitched from this table – it was quite long enough already!] However, in this case, it has been ordered by WP rather than games.
It’s interesting to note that 2004 had both the best reliever (Mike Koplove) and the worst (Matt Mantei) across the two seasons. But otherwise, what stands out, much as with the starting pitching, is the “long tail” of the relievers this year. 23 different pitchers ended up with negative win probability for Arizona this season. The 2001 Diamondbacks didn’t even USE 23 different relief pitchers. Almost half of those (eleven) were worth -50% or worse, compared to only four who fell into that category from 2004. If there were times this season where it felt like the Diamondbacks were throwing whatever they could find at the relief wall, to see what stuck… the table above seems to concur.
The problem is, with only seven relievers posting positive WPA – and that includes Tyler Clippard, who won’t be back – I’m hard pushed to say the team found much that will be of use in 2022 and forward. There are a few names who might be usable: Noe Ramirez, Sean Poppen, Rule 5 pick Brett deGeus and even Joe Mantiply, whose numbers outside of WP were pretty good. But it’s not a roster which, at this point, seems likely to strike fear into opposing hitters. The quest for power bullpen arms will likely continue.