That Greg Holland, who was signed on Opening Day to a one-year, $14 million contract which also cost the St. Louis Cardinals a second-round draft pick, has improved from his start to the season is about as predictable of a baseball event as one can find.

Through May 25, after which Greg Holland found himself on the Disabled List, ostensibly with a right hip impingement though more than a few sources speculated that this was more an excuse to keep him away from the MLB club for a while, Holland had a 9.45 ERA. If Greg Holland entered a game for an inning, it was mathematically likely that he would surrender at least one run. His 5.96 FIP, 7.18 xFIP, and 6.72 SIERA didn’t bode much better. Sure, Holland’s ERA was unsustainably, ludicrously bad, but his peripherals still indicated a pitcher who was suddenly very bad at pitching.

Since returning last Tuesday, however, Holland has looked unhittable in his dominance. It has amounted to four innings, sure, but the eye test seemingly confirms what his tremendous numbers indicate—that Greg Holland (four innings pitched, five strikeouts, one hit, and zero walks) is extremely back.

Now, Holland isn’t the unhittable bullpen wizard that he has been in June—he has been superhuman in obvious ways which we have never seen if extrapolated to a full season. But this looks like somewhat convincing evidence that Greg Holland is better than he was earlier in the season. It’s a low standard, but it’s something.

It is typical of relief pitchers to fluctuate in performance. Holland is being a typical reliever. In a seemingly backwards way, exposing Holland as a typical reliever reinforces the argument that teams shouldn’t be signing relievers in the first place.

In the modern era of relief pitching (I’m going to draw the line semi-arbitrarily at 1979, the first managerial season of Tony LaRussa, widely accepted as the architect of modern bullpen management), there have been 208 reliever seasons in which a pitcher accumulated at least three Wins Above Replacement–for any other position, three is a good-not-great mark, but for a reliever, it’s something that only happens 5 1/3 times per season (yes, there’s strike-shortened seasons in there; yes, you should probably mentally adjust the fractional amount up a little bit; no, I’m not going to do it for you). 141 pitchers did it at least once–Greg Holland is among that group. Only 43 pitchers did it two or more times. Only 13 pitchers did it three or more times. Only 4 did it four or more times. Keith Foulke did it five times, and Mariano Rivera ruined the curve by doing it ten times.

Let’s compare that to starting pitchers in the same time frame with the Wins Above Replacement threshold being set at 5.8 WAR so as to achieve as close to the same number of unique pitchers who have achieved the mark (in this case, it is 139 rather than 141). 56 starters have had multiple such seasons. 34 had at least three. 20 had at least four. 10 had at least five. Clayton Kershaw and Roy Halladay had six; Pedro Martinez and Curt Schilling had seven; Randy Johnson and Greg Maddux had eight; Roger Clemens had 11. Great starters more consistently remain great starters than do great relievers, who, when they aren’t Mariano Rivera, tend to have much shorter peaks.

If Greg Holland had come back to the Cardinals and been outright terrible in relief, it would of course be worse for the Cardinals as a team, but one could also chalk it up to an exceptional bit of bad luck. Perhaps that Greg Holland was hurt; perhaps that Greg Holland had suddenly, inexplicably become toast. That the Cardinals front office missed these red flags isn’t exactly becoming, but mistakes happen on a micro level regardless of the front office. But if Greg Holland is simply, as most relievers are, an inconsistent entity, it underscores the silliness of investing so heavily in one’s bullpen.

The best reliever in baseball this season per Baseball Reference WAR is Adam Ottavino, a pre-free agency former starter (with the Cardinals!) whom the Colorado Rockies claimed off waivers in 2012. The best reliever by FanGraphs WAR is Josh Hader, a second-year 24 year-old Milwaukee Brewers reliever who was exclusively a starting pitcher in the minor leagues in 2016 and 2017. Yes, Aroldis Chapman is among the top relievers, but much of the upper reaches of the list are also-rans who are, at least for now, very good.

On the Cardinals, depending on who you ask, the top relievers have been Jordan Hicks, a minor league starter, and Bud Norris, a low-tier free agent signing who was a long-time starting pitcher in the Majors. Neither of these players required a major financial commitment and were essentially lottery tickets purchased by the club. Many like them didn’t work out; they did.

Of course, by the time Greg Holland was signed, the Cardinals had exhausted their free agent options. If Holland maintains his recent pace throughout the rest of the season, he will salvage his early awfulness to be above Replacement Level, which would make him more valuable than, well, banking the money. But the money could have been spent on, say, adding infield depth. Or signing a legitimate backup catcher. Or buying out Mike Matheny’s contract. Of course, none of these things were actually going to happen, but they could have.

At the end of the day, the Greg Holland signing was based on faulty logic–that a team needs a “proven closer” when dominant closers by and large cycle through and burn brightly for short periods of time. Whether one blames Mike Matheny or the front office for the acquisition, the Cardinals bought into a stale mantra by believing that a once-great closer will likely be something other than a player with a very short shelf life.