Searching for underrated MLB free agents

Free agency in MLB is an interesting beast. We know that, for the most part, the broad pattern of spending by teams will align with a fairly predictable pattern. We’re able to throw out the mythical “$/WAR” figure in large part because that figure actually aligns to what teams pay free agents, at least as far as meaningful, non-reliever contributors go. But, that figure is an average, collected from an aggregate of free agent spending. Not every deal will align — some players will be undervalued, and some players (and their agents) will be better at extracting far more from teams than they’re worth.

It’s hard to know, given the infant (toddler?) stage of this offseason, exactly which free agent will fall into which of the above categories. Once the players sign, analysis is also straightforward, if time-consuming: model out the player’s forecasted WAR, compare it to the player’s salary, and see whether the deal is fair, team-friendly, or (way) too much for the player in question. Different models may disagree on the player’s expected production, which sometimes changes the answers, but it’s still a pretty facile exercise. Here, though, I’m doing something different.

Specifically, I’m trying to look at which free agents are currently underrated and overrated. Since I can’t really know exactly what teams are thinking, I’m relying on something that may not be the next best thing, but could serve our purposes here nonetheless: top 50 free agent lists. To date, I’m aware of three of these lists: Fangraphs (Ben Clemens, and crowdsourced), MLB Trade Rumors, and ESPN (Kiley McDaniel, $). Because the Fangraphs list has two different estimates, this gives me four sets of values. (Are there other lists I’m missing? If so, let me know!) With these lists in hand, I use the following methodology to evaluate the predictions therein.

Methodology

First, it’s important to clarify that here, I am focusing solely on total contract dollars. Why? Because in the recent past, we’ve seen something between “a few disparate cases” and “a trend” (in other words, a mixed bag) wherein some teams will offer some players extra years, without changing the total contract dollars you’d expect for the player when modeling their contract. In other words, teams are diluting the AAV owed to the player, potentially for luxury tax purposes, at the expense of “paying” the player through a year that seems likely to be underwhelming and/or useless. Because this is likely going to happen in at least a few deals this offseason, looking at AAV doesn’t really make sense.

With that in mind, what I do is simply take the four total contract estimates and average them. That’s the “predicted” side of the equation. If a player didn’t appear on one top 50, but was an honorable mention, I gave the player a predicted contract value of $4 million (MLBTR) or $5 million (ESPN), based on the midpoint between $0 and the lowest contract predicted among the top 50 in that publication. If the player didn’t appear at all in a publication, then I just omitted that publication from the average, in case they forgot about the player’s existence. Then, I need something to compare it to. For this, I develop a very simple model that does the following:

• It estimates 2023 performance in one of two ways: 1) Steamer’s current projections for the 2023 season, with playing time adjusted to reflect the current assumptions on Fangraphs’ Depth Charts; and 2) an IWAG projection for the 2023 season for the player (including the player’s point estimate orojection for playing time).
• For each of these two projections, the corresponding contract is modeled as follows: the player is assumed to receive a contract that lasts up to the point where they cross the 2 WAR threshold. A player’s projected production declines as follows: no change from the projected WAR through age 29; a loss of 0.25 WAR in age 30; a loss of 0.5 WAR per season in ages 31-34; a loss of 0.625 WAR in age 35; a loss of 0.75 WAR per season in ages 36-39; a loss of 0.875 WAR in age 40; and (thanks, Justin Verlander, ya goober), a loss of 1.0 WAR per season in age 41 and further on.
• To estimate a dollar value for the contract, I initially used the following: for players whose 2023 projection is 2.0 WAR or above, $8.5 million per total WAR accrued over the contract; for players whose 2023 projection was below 2.0 WAR, $6.5 million per WAR accrued (by definition, this WAR was accrued only in 2023). However, in my initial review of results using these figures, it became very clear that the four prediction sources were using something, perhaps subconsciously, closer to $9 million per WAR. So, I changed $8.5 million to $9 million per WAR for the “regulars,” and $6.5 million to $7 million for the “role players.” This seems fairly minor, but it made a big difference in calibrating the results.
• With those two contract values (one being Steamer, and one being IWAG) in hand, I just averaged the total contract dollars, and compared to the average of four predictions described above.

To belabor the point, let’s look at Andrew Heaney. His four predicted contracts are really different: Ben Clemens had him at $12M/1, the Fangraphs crowdsource at $20M/2, MLBTR at $42M/3, and ESPN also at $20M/2. The average contract value is $24M. Steamer projects Heaney for 2.6 WAR in his 2023, age-32 season, which means that his total contract is 2.6 WAR in 2023 and 2.1 WAR in 2024 (less 0.5 going from 32 to 33). 4.7 * 9 = $42.3M. IWAG definitely sees Heaney as a 2.7/200 guy, but figures he will only stay healthy for about 100 innings, so its 2023 projection for him is 1.4, which means IWAG’s contract estimate is just 1.4 * 7 = $9.8M. (IWAG is usually far closer to Steamer, but Heaney is a good example because the contract estimates also vary widely.) The average is $42.3M and $9.8M is basically $26M, so in this case the “delta” we get is right around +$2M. In other words, this exercise sees Heaney as worth ever-so-slightly more than the average of the four predictions.

Big ol’ note: ideally I would’ve included ZiPS in some way, but there’s no 2023 “full-throated” ZiPS that incorporates 2022 yet, so instead you’re stuck with IWAG serving as a sometimes-corroborator, and sometimes-way-more-sanguine projector vis-a-vis Steamer. See the “calibration” section at the end if you care about those sorts of considerations; I think this exercise is decently calibrated for being a fairly quick thing I threw together.

Alright, now that I’ve wasted your time for over 1,000 words without results, so if you skimmed all of the above because tl;dr, you can read about actual players below.

---

Without further ado, the “five” (not really five) most underrated free agents, by this methodology, where “underrated” actually means “warrant a bigger contract than seems to be estimated by the four predictions we currently have” are...

5. Justin Turner

Turner’s predicted contracts are all pretty much the same: from $11M/1 to $14M/1. And, honestly, if you only had Steamer doing the lifting here, there wouldn’t be too much to discuss: Steamer seems him as a 2 WAR player for 2023 in his age-38 season, which is a few bucks higher than the prediction, but not egregiously so. On the flip side, though, consider that Turner already came in at 2.4 at age 37, and threw up 4.2 at age 36. The big change between those two seasons wasn’t his hitting, but rather that the Dodgers DHed him for about half his playing time. Of the 20-run difference for him between 2021 and 2022, two are because he missed some time, and six can be explained by just playing DH so much (the six runs is already adjusted for the hit he would’ve taken by playing more 3B; it’s nowhere near as steep as the DH penalty). If he had clocked a 3-win season in 2022, would the predictions still be for a one-year deal? Yeah, maybe.

Really, the reason why Turner is an underrated free agent is because of the combination of his fairly conservative projection still warranting a bigger payday than is being predicted, and the fact that he’s got room for even more production with a fairly easy way to get there. Maybe age 38 is when he finally falls off the cliff, but he hasn’t yet, and it looks like the expectation is that the possibility is so real that it needs to be priced in.

Note: the Dodgers declined a $16M/1 option on Turner on November 10, which you figure kind of caps what he might get on the market... but that doesn’t mean he isn’t underrated anyway.

Scornful comparison: Nelson Cruz got $15M/1 last offseason, coming off back-to-back 2.0 fWAR seasons and heading into his age-40 campaign. Yet, Turner is supposed to get less than that, coming off 4.2 and 2.4, while two years younger?

4. Dansby Swanson

Swanson is another interesting case, because like Turner, and unlike the guys that are going to appear further down, Steamer really doesn’t think much of him. In fact, if you use the methodology described above, the prescription for a contract using Steamer’s 3.3 WAR forecast for 2023 is essentially $99M/4, which is way, way below the predictions, which are all in the $140M-$155M range.

Yet, Steamer seems like a pretty tough, low-end assessment of Swanson’s value. It figures that the man with the fantastic hair is good for just a 104 wRC+ point estimate for 2023, when Swanson’s xwOBA has been above the 70th percentile among regulars in three of the last four seasons. Steamer also figures that Swanson will (again, on a point estimate basis), play below-average defense; by OAA, Swanson hasn’t had a below-average year since 2017.

The four predictions culminate in an average of $147M for Swanson. That suggests essentially a 4 WAR player signing a five-year deal to cover his age 29-33 seasons. If you take Swanson’s 2022 (6.4 fWAR) and subtract his entire defensive value, i.e., not just him playing well in 2022 relative to other shortstops, but take away the entire shortstop positional adjustment as well, he becomes a 4 WAR player. Throw in him being an average shortstop and just maintaining his 2022 hitting, which featured his third-lowest xwOBA of the past four years, and he easily clears 4 WAR. Add in some positive defense at short, and there’s a lot of upside there if all it takes is one of those forecasted deals.

Scornful comparison: Trevor Story and Javier Baez each got $140M/6 with an opt-out last offseason. Story was coming off a 2.8 fWAR year with recent xwOBAs in Swanson’s range, despite playing at Coors Field (which boosts xwOBA, too). Story’s 6+ fWAR, defensively-driven season was two years before he hit free agency. Baez outhit his xwOBA a bunch to get to 4.0 fWAR in his walk year, he too had a couple of 6ish fWAR seasons, one defensively-driven, multiple years before he hit the market.

3. Clayton Kershaw

Yeah, we’re getting into the fairly blatant ones now. The kicker is, Kershaw has already agreed to a $20M/1 deal with the Dodgers. It’s a clear win for them, though I’m sure he doesn’t mind.

Steamer has Kershaw projected for 3.2 WAR, which would be his lowest full-season total since his rookie season. Yet, even if he were to provide 3.2 WAR on a one-year deal, that would be worth around $29 million, even if you ignore that a guy projected for 3 WAR now probably warrants a second year. And if Kershaw’s 2023 looks like his 2022 (3.8 fWAR?). Hey, Casey Close and Excel: the Dodgers don’t need this kind of surplus value assistance, thanks.

Side note: you may be thinking to yourself that Kershaw’s storied injury problems make $20M/1 more fair than is presented above. But look: Kershaw finished 92nd among starters in innings in 2022, and 78th among starters in innings between the 2021 and 2022 seasons combined. He’s also 21st in fWAR among starters for 2022, and 17th in fWAR among starters for those two seasons combined. In a universe where the Dodgers aren’t scooping him up for essentially the Qualifying Offer amount, you line up around the block to get that sort of production, even/especially if he can do it in so few innings that someone else can add value on top of that.

Scornful comparison: Noah Syndergaard got $21M/1 last year, after basically missing two whole years, and being not much better than Kershaw’s 2022 (4.3 fWAR in 2019, 4.2 in 2018) before that.

2. Carlos Rodon

Okay, what about a starter that hasn’t yet re-signed on a sweetheart deal with his most recent team? Rodon’s contract predictions are all in the $120M-$140M range; given that he’s entering his age-30 season, the average of $131M is akin to him entering 2023 at 4 WAR. Thing is, he was worth nearly 5 fWAR (4.9) in “just” 132 ²⁄₃ innings in 2021, and then blew right past that with 6.2 fWAR in 178 innings in 2022. Steamer conservatively projects him for 4.5 WAR, which likely prices in to a substantial extent the fact that he barely pitched in 2019-2020.

In short, Rodon is on this list because buying into the predictions for his contract suggests that the possibility of catastrophic injury wields an outsize impact. Unlike Turner and Kershaw, age doesn’t really factor in — this is based on a track record that Rodon has been able to rewrite, to some extent, over the last two years.

Scornful comparison: Max Scherzer got $130M/3 heading into his age-37 season after amassing 13.7 fWAR in his last 419 innings. Rodon is being predicted to get around $130M heading into his age-30 season after amassing 12.1 fWAR in his last 353 innings.

1. Brandon Nimmo

Nimmo was, to be fair, the most obvious candidate for this spot even without doing the exercise itself. The predictions have bandied about contracts in the $100M-$120M range, which seems kind of weird — it suggests he’s something like a 3.5-4.0 going forward, when the most recent time he’s been at or below that range on a rate basis was 2019 (which was preceded by a 4.8 fWAR season in 2018).

The Fangraphs writeups by Clemens and David Laurila note that Nimmo has not been particularly durable — he missed big chunks of both 2019 and 2021. He’s also outhit his xwOBA a bunch, and he has this weird thing where he’s physically very fast, but sucks on the basepaths. Combine that with his counter-zeitgeist, OBP-focused profile and the fact that he was a not-particularly-exciting outfielder before apparently finding a groove in center field in his late 20s, and you can see a bunch of little reasons why his contract might get dinged. But in sum, it’s not clear that these dings are justifiable. Both Steamer and IWAG have Nimmo at 4.5 WAR for next year; it takes a lot of dings to explain going from the $160Mish that production warrants down to taking a 30 percent haircut.

Scornful comparison: There aren’t a lot of great ones here, but Nick Castellanos got $100M coming off 3.5 fWAR in his last 600 or so PAs, and 6.3 in his last 1,500 or so. Nimmo has 5.4 in his last 700 or so, and 11.7 in his last 1,500ish. That seems to warrant a lot more than $100M, or even $110M. (Nimmo detractors, however, have J.T. Realmuto’s pre-2021 deal to point to — it was $115M for an even more productive player who also got his value in not-entirely-straightforward ways.)

---

The man out of time, space, and contract modeling: Aaron Judge

The really big guy out of Sacramento warrants a bit of his own discussion here. I think he kind of breaks this exercise. The reality is that guys at the very top of the market have to contend with something more than market rates — they have to contend with MLB owners and Front Offices being reticent to leap multiple rungs up the compensation ladder. There’s discussion about this in Russell Carleton’s great book, The Shift, but the general idea is that Judge’s insane, 11.4 fWAR walk year creates a situation that players and teams haven’t really navigated before, and that Judge’s earnings are going to be limited by more than the fact that he’s hitting free agency in his 30s and Father Time being undefeated.

To that end, I think there’s room to discuss whether the predictions for Judge’s ultimate deal, which are all in the $300M-$335M range, are only a little off due to market-setting reticence, or perhaps far more substantially affected by non-player considerations. That said, it’s hard to describe Judge’s free agent candidacy as “underrated,” because, well... 11.4 fWAR in a walk year. Still, he’s a singularly weird case for this offseason.

---

Notes on Calibration

This exercise would be (in my mind, anyway) considerably less interesting if the contract estimation models and heuristics were way off of what free agent buyers were predicted to spend. Fortunately, with just a bit of adjustment to the $/WAR framework, as described above, I think they align pretty well.

For this exercise, I assessed 60 players; the only players appearing on a top-50 list that were left off were guys like Drew Rucinski or Michael Lorenzen, who either didn’t have a recent stateside track record leading to useful projections, worked on a hard-to-value swingman capacity, or both.

Across all 60, using Steamer’s projections and the methodology described herein, player contract predictions were inflated relative to the modeled contract by $7M on average, which doesn’t seem like a lot until you realize that $7M * 60 players = a net predicted overspending of $420M. IWAG serves as a counterpoint — player contract predictions compared to the modeled contracts using IWAG’s 2023 estimates and the corresponding aging assumptions were off by less than $1M on average, or a net predicted underspending of under $40 million — however, this is driven almost entirely by the fact that IWAG has no idea how to handle Aaron Judge going forward given that his most recent season was insane. In addition, 13 of the 60 “top” free agents are relievers, which have enjoyed a well-known, pre-established pattern of nonsensical spending that drives a lot of the “overspending” suggestion.

If you remove the relievers, and average the Steamer and IWAG-based contracts, you are left with aggregate overspending of about $42 million across 47 players, which seems reasonable-ish. That’s predicated on the (silly) idea that teams should collectively reallocate some funds from a bunch of not-so-great bets and put it more towards Judge and some of the guys mentioned here, but I think it works okay in aggregate.

If you made it this far, stay tuned for the flip side piece about overrated free agents using the same methodology.