First, lets be positive, assume that the “big four” all return to something like pre-2009 form, and that Johan Santana, David Wright, Carlos Beltran, and Jose Reyes average 5 WAR each. A team of replacement level players can be expected to win something like 47-50 games, so that would jump us to 67-70 wins. So to reach 95 Pythagorean wins, the rest of the roster has to generate about 25-28 wins. If we assume 28 to be conservative, that’s 28 WAR / 21 remaining roster spots = an average of 1.33 WAR per additional player. We can do better though. Seven of those roster spots go to relievers, and five to bench players. Neither position figures to have many 1.33+ WAR players.

18 players made up the group of what was at any time seven RP roster spots in 2009, and they totaled 533.2 IP. We’ll throw out Nieve, Figueroa, and Misch since the vast majority of their innings were as SP. The rest of the group totaled exactly 1.0 WAR (That’s totaled, not averaged. Yuck. Perhaps its slightly more, since I can’t distinguish Parnell’s SP WAR from RP WAR, and he was much more effective as a reliever, but still, yuck). Divided by seven (for the seven roster spots occupied by a reliever at any given time), that’s about 0.14 WAR per spot.

If we run the same exercise for 2006, when the Mets actually had a very good bullpen, we get a total of roughly 6.9 WAR, or just under 1 WAR for each of the seven roster spots. Obviously, still less than 1.33 WAR per roster spot, but much closer.

Even before we get to the bench, we can see the impact the difference of a decent bullpen, any roster’s most volatile unit, can make. If the pen totals just 1 WAR, combined with the “big four”, that leaves 14 remaining roster spots to account for 27 WAR. The rest of the roster needs to average just a tad under 2.0 WAR per roster spot. Without at least two or three more legitimate star players (4+ WAR), this will be virtually impossible after accounting for the five bench spots. If the bullpen instead gets you about 7 WAR, then the other 14 roster spots need to get you only 21 WAR, or a much more reasonable 1.5 WAR per roster spot.

The bench is a bit more difficult to define in terms of discrete roster spots. We can all agree that the bench exists, but how do we account for platoon players? How about guys who flirt back and forth between the bench and the everyday lineup? Older players who need rest? For the sake of argument, I’m going to go slightly conservative, and assume the average bench roster spot should produce about 0.3 WAR over a full season, which is only slightly better than a replacement player you might call up from the minors.

Working under this assumption, in the “strong bullpen” case, even if the five bench spots average 0.3 WAR, that leaves 19.5 WAR left to generate from the nine remaining primary players (4 SP, 5 starting position players), which breaks down to roughly 2.2 WAR per spot, or roughly average production over a 162 game season.

In the “weak bullpen” case and an average of 0.3 WAR per bench spot, the rest of the roster needs to generate about 25.5 WAR, which spread among 9 players is 2.83 WAR per spot, requiring multiple additional players who are considerably above average. This isn’t impossible as a general thought, but very unlikely given the team’s current roster and organization’s spending budget for the offseason.

Now lets take this a step further. Of the nine remaining roster spots not covered yet, we have players at many of those positions guaranteed or virtually guaranteed jobs. We can assume either Jeff Francoeur or Angel Pagan will likely be the starting RF, and between them, we have reason to hope for a legit 2 WAR player (perhaps better, and a little spare change for the bench maybe, but the 2 WAR estimate seems “fair” at best, so lets not go crazy adding extra decimal points to the bench). Luis Castillo may take a bit more optimism, but lets give him a repeat of his solid 1.5 WAR season. Omir Santos was a reasonable 1.0 WAR catcher in part time duty, so if we combine him with an equal catcher, and subtract 0.3 WAR since at any given time one of them will occupy one of the “Bench” roster spots, that’s 1.7 WAR. Big Pelf was 1.8 WAR, again, lets be generous and bump him to 2.0 WAR. Lets also assume that Ollie bounces back, and even give him a little positive value, 0.5 WAR. That leaves us with this fairly generous estimate:

C – 1.7 WAR
2B – 1.5 WAR
RF – 2.0 WAR
SP – 2.0 WAR
SP – 0.5 WAR

Total – 7.2 WAR
Average – 1.44 WAR
Remaining roster spots – 4 (LF, 1B, SP, SP)
a. Remaining wins to 95 with a 1 WAR bullpen – 18.3
b. Remaining wins to 95 with a 4 WAR bullpen – 15.3
c. Remaining wins to 95 with a 7 WAR bullpen – 12.3

Matt Holliday has been 5.6 WAR, 6.2 WAR, and 7.6 WAR in each of the last three seasons, respectively. So if he were added, we could assume conservatively 5 more WAR with three roster spots remaining. So a, b, and c above would look like this with Matt Holliday signed to play LF:

a. 13.3 WAR remaining to 95
b. 10.3 WAR remaining to 95
c. 7.3 WAR remaining to 95

Three roster spots to go. If we assume a John Lackey/Aaron Harang-esque SP addition, that’s about another 3 WAR. Now we’re at:

a. 10.3 WAR to 95
b. 7.3 WAR to 95
c. 4.3 WAR to 95

And we have 1B and SP left to fill. There are options in place for both positions, namely Daniel Murphy for 1B and John Maine and Jon Niese for SP. Since at this point the Mets do have six positive value starters, we can even be generous and add another 1 WAR in replacement SP innings. Still, none of these three figures to be more than average, or about 2 WAR given a full or nearly full season of playing time, and that’s being very generous to all three. It could very well be enough….if the Mets could somehow piece together a 7 WAR bullpen again. That’s by no means impossible given the volatility of bullpens year to year and a few cheap and shrewd relief acquisitions, but its hard to imagine the organization having enough money for a big impact acquisition at this point. With a more mediocre bullpen, in the 3-4 WAR range, which seems quite plausible, the team will likely be right back somewhere in the 88-90 win range they spent 2007 and 2008 in. With a 1 WAR bullpen, they’ll likely be at or around .500, but anything better cannot be said without an incredibly startling improvement from a player or two already in the organization.

In conclusion, I’d like to stress that my conclusion is first and foremost NOT about how “bullpens win championships”. Simply that relief pitching, more than any other aspect of a baseball roster, is subject to the law of small numbers, and even a bullpen expected to produce about 4 WAR still has a decent chance of producing either 1 WAR or 7 WAR. But with some shrewd pickups, minor tweaks, and a little luck, this roster is not so far out of the realm of contention that we should already give up on 2009. It will require likely require at least one “star” addition to the roster and another “near-star” addition to the roster, in this example, Holliday and Lackey/Harang, and then even a bit more. And it will also require us to own up and understand that having Murphy at 1B AND Maine/Niese as the “#4” SP probably isn’t going to cut it, but one or the other might suffice. We don’t need a start 1B or the best #4 starter in the league, but at least one of those spots need to be filled by an average or slightly above player. And of course, all this is assuming “the core” can be what the were before 2009, which at this point, is a large unknown, but hardly an unreasonable assumption. In fact, for roster construction, its probably a necessary assumption, since if two of those guys fall way short again, it does seem virtually impossible that the team will be seriously competitive unless they are replaced with an equal or better player.

One more note: Its important to distinguish between the WAR of a player, say David Wright, and the WAR of a roster spot, say starting 3B. If I assign a specific player to a specific position and give them a WAR total, I’m generally doing my best to factor injury/attrition into the mix, in which case, for the given roster spot that player occupies, I’m assuming a replacement level replacement, or a 0 WAR player. All of this math is a little fuzzy for this reason, since lots of the elements I’ve labeled constants or limited outcome variables are actually non-limited outcome variables. But, I still think as an exercise, something like this proves somewhat insightful for what we should expect/hope to happen in the coming months to return our favorite team to respectability.