The Mets need to improve their bullpen. That is alarmingly clear to anyone who watched the team in 2018. They have some interesting arms but ultimately, they need to a better job of building a bullpen. How does a team do that though?

The easy answer is throwing a lot of money at it or trading for the big named reliever on the market. Sometimes that works out but sometimes, that also does not. For example, the Indians got Andrew Miller and it could be argued that he was the reason an undermanned team almost won the 2016 World Series. Other times, you have a team like the Rockies that spends more than $100 million on their bullpen only to have all of them pitch terribly.

In the 2018 season, the teams with the five best bullpens as per WPA/LI were the Astros, Athletics, Rays, Yankees, Cubs. A little further down the list were the Brewers, Padres, and Red Sox before a big drop-off.

So first things first, what is WPA/LI? Win Probability Added (WPA) measures how much a player increases or decreases his teams chances of winning on a given particular outcome. So say Seth Lugo comes into a game that the Mets have a 50% of winning. He subsequently strikes out the side and the Mets’ chances of winning jumps to 70%. His WPA would be 0.20. This is a cumulative statistic and builds over the course of a season.

Leverage Index essentially measures how important of a situation that the player is being used. So if Lugo was coming into a 15-0 game just to get some work, his leverage index would be low but if it was the late and the Mets had a one-run lead, his LI would be high. Leverage Index measures the average leverage that a player is in.

WPA/LI adds together all the situations that a player is involved in. It divides their WPA/LI and adds them all together over the course of a season. Let us stick with Lugo for this example. Say he comes into a bases loaded, no-out jam with the 2-3-4 spot of the lineup due up and it is a one run game in the ninth inning. The Mets probability of winning is probably 10% or so. Let’s say Lugo gets out of it and the Mets win, his WPA for that outing would be 0.90. The leverage index would also be five for arguments sake. So his WPA/LI is 0.18.

The higher a WPA/LI is for a player, the better they are. WPA/LI for a team just adds together the WPA/LI for all the players on the team. For the bullpen, it adds together the WPA/LI of all the members of the bullpen.

The Astros bullpen was led in WPA/LI by Collin McHugh, Tony Sipp, Ryan Pressly (who only pitched in Houston for two months), Will Harris, Joe Smith, and Hector Rondon. The only major signing in that group were Tony Sipp and Joe Smith. Both were signed for their strength against same side hitters. Even then, neither were bank breakers.

McHugh was a mediocre starter who was put into the bullpen to resounding success. Pressly was 29 years old and only had a good first half of 2018 under his belt. The Astros got him for relatively cheap and used their data to make him a reliever that had a 0.77 ERA, 1.49 FIP, and 32:3 strikeout-to-walk ratio in 23.1 innings for Houston. Will Harris was someone the Astros found in the minors and has been a solid reliever since they got him. Rondon was someone with a good fastball/slider combo who had success in the past but needed a change of scenery from Chicago. The Astros got him for a good discount and 3.20 ERA, 2.79 FIP, and 67:20 strikeout-to-walk ratio in 59 innings.

The Athletics essentially assembled a group of guys who threw really hard like Blake Treinen and Lou Trivino to make their bullpen. They are the epitome of finding relievers for cheap and maximizing their value. Same thing with the Rays, they have guys who have throw very hard and are used in situations that allow them to provide the most value to their teams chances of winning. Neither of these teams spend a lot on their bullpens but still manage to field good bullpens.

The Yankees gave Aroldis Chapman a lot of money but the other guys in their bullpen were all essentially failed starters who became dominant relievers. Dellin Betances, Chad Green, and Jonathan Holder are all the molds of guys who began as starters but were moved to the bullpen.

The Cubs did invest a lot into their bullpens and some moves have worked and some have not. Steve Cishek and Brandon Morrow were good adds but Justin Wilson was underwhelming. Cishek and Morrow were both high on the list of WPA/LI but so were Jesse Chavez and Jorge De La Rosa. Chavez was someone I identified in the offseason as a guy who could be a good reliever on a MILB deal. The Rangers signed Chavez and put him in their bullpen. He showed the ability to increase his velocity in shorter stints and the Cubs picked him up at the deadline. While pitching for the smarter Cubs (compared to the Rangers) Chavez posted a 1.15 ERA, 2.39 FIP, and 42:5 strikeout-to-walk ratio.

The Brewers were lead by a failed starter (Josh Hader), 2017 non-tender candidate (Jeremy Jeffress), and a guy they got for Yovani Gallardo (Corey Knebel).

The Padres got Brad Hand as a failed starter from the Marlins and turned him into an elite prospect in Fransisco Mejia. They also found a couple of gems on other teams scrap heaps like Kirby Yates and Craig Stammen.

The Red Sox had a big guy in Craig Kimbrel but also had guys they turned into solid relievers like Matt Barnes, Joe Kelly, and Ryan Brassier. Alex Cora is also a brilliant bullpen manager.

So what is the point to all of this? The answer is: don’t waste all your resources on the bullpen. Good relievers come out of nowhere and sometimes come from your own team. Guys who failed starters have shown the ability to become good relievers. Guys who throw hard suddenly develop a breaking pitch with the help of good coaching and become good relievers. If Jesse Chavez can turn into a solid reliever, there are good odds if you do your research that you can find more pitchers like him.

My message to the Mets is simple: fix the bullpen but do not waste all your resources on the bullpen. Last season, their best reliever was a guy who was an okay starter but had his stuff play up in shorter stints.