Retrospective Player Valuation: Mixed League Hitters

Welcome to my fourth annual look at retrospective player valuation at Baseball Prospectus. Over the next few weeks, I will be writing a series of articles examining how players performed from a fantasy perspective in 2016. This is the fifth article in a series of six. The first four articles in the series focused on NL-only and AL-only leagues. The final two posts examine mixed leagues, with this article examining hitters.

Before I dig in, here is a brief description of the charts below. (If you have been reading along for the entire series, note that there are some changes for the mixed league articles).

The $ value column is based on my Rotisserie-style, 5×5 formulas. It doesn’t exactly match anything in Baseball Prospectus’ PFM, but is derived using a SGP valuation model (something the PFM does offer). There are two important things to know about the values:

1) They are derived using the 210 best perceived hitters and the 135 best perceived pitchers (read most expensive or highest draft position) on Opening Day, not the best 210 hitters and 135 pitchers at the end of the season.

2) The values of the 345 most expensive players add up to $3,900. This is a fundamental difference from many pricing systems that use z-scores and assign the top 345 players an aggregate value of $3,900. While perhaps more accurate, dollar values derived from the best 345 players at the end of the season do not reflect how a fantasy team should behave in an auction environment or what these dollars truly represent,

Actual Rank lists where players ranked overall based on my dollar valuations. Since the NFBC rankings do not differentiate between hitters and pitchers, this column does not do so either.

The NFBC column lists the average draft position (ADP) as measured by the National Fantasy Baseball Championship (NFBC) drafts.

The LABR column lists the draft results of the LABR Mixed Auction, which gives us insight into what the fantasy “experts” were thinking back in February.

MG is yours truly, your heroic pricer and proud prognosticator since 2013. Another good reason to look back is to see if the fantasy expert you are following is good at what he or she does. It is easy to make predictions in March and never revisit those predictions or (worse) cherry pick the ones you got right and take a hollow victory lap. But how good are we at what we do? The rankings below were published in March at Baseball Prospectus as part of our Top 300 Draft rankings. I have always taken others to task for their predictions; now it’s finally time for me to face the music.

(This article contains valuations for select players. A complete valuation list can be found here.)

A theme surrounding much pre-season fantasy analysis is that hitters have become far more unpredictable than they have in past seasons. However, the NFBC drafters stuck with at least half of the Top 15 hitters from 2015.

Table 1: 15 Highest Drafted NFBC Hitters, 2016

Trout, Goldschmidt, Donaldson, Rizzo, Stanton, Cabrera, and McCutchen repeat on the Top 15 hitter picks in 2015 and 2016. As is always the case, the NFBC and LABR drafters are not making predictions, they are placing bets. There is a slight amount of variance with a few of these picks, but even the difference between Donaldson’s rank in NFBC and LABR isn’t particularly significant.

Trout and Goldschmidt didn’t quite finish 1-2 overall, but it is difficult to argue that a Top 10 finish is anything but an unqualified success. Even with the benefit of hindsight, Stanton is the only hitter in the top 15 who was probably too risky in this slot. McCutchen seemed appropriately valued, particularly by the NFBC drafters. The big story in 2016 was the impact that stolen bases had on fantasy value. Altuve and Betts finished 1-2, and while neither hitter was an elite stolen base threat, the combination of the steals they did provide along with their all-around statistical package put them at the top of the heap. Contrast this with Machado, who had a great year but could barely crack the Top 50 in large part due to getting shut out in steals.

Return readers may notice a few changes from last year’s mixed league, retrospective valuation tables. Both Tout Wars leagues have been removed from the table. They use on-base percentage, and while the fluctuation doesn’t matter much in aggregate valuation, it does have a significant impact on hitters at the top of the rankings. The Tout Mixed Draft saw some particularly sharp divergence from LABR and NFBC. Rather than try to capture these differences with yet another valuation column, I left them out this year.

If it wasn’t obvious from Table 1, the best 2016 mixed hitters fully display the impact of steals.

Table 2: Top 15 Mixed League Hitters, 2016

*Nunez is not factored into the average row at the bottom of this table. Villar is, with a ranking of 346 (which would be the top reserve pick in a 23-man roster, 15-team draft).

Each of the top eight hitters in 2016 stole at least 17 bases, with a range between 17 (Blackmon) and 62 (Villar). It’s worth pointing out how much batting average matters in Table 2. Nine of the 15 best hitters in 2016 topped .300, with only Dozier finishing below .285. While steals do matter now more than ever, it is important not to oversell how much the paradigm has shifted. Except for Marte, every hitter on Table 2 hit at least 16 home runs. I’m sometimes accused of overvaluing stolen bases, but you can’t get here without offering at least a little bit of something across the board. You can finish at the top without steals. Five of the 15 best hitters stole eight or fewer bases, with Arenado, Murphy, and Ortiz stealing five or fewer.

My rankings (the MG column) are a lot like both LABR and NFBC, with LABR doing a slightly better job. Apart from Villar and possibly Segura, there is consensus on nearly every hitter. As noted above, LABR, NFBC, and MG are all placing bets. I had Murphy ranked highest, but ranking him 100 instead of 151 would have been an impractical exercise. I wasn’t trying to predict where Murphy would finish but rather expressing a preference compared to the market that wasn’t completely out of line with his market value.

If you are familiar with other valuation systems, the “$” column may seem too low. As an example, here are the Top 15 as the PFM saw it in 2016.

Table 3: Top 15 Mixed League Hitters, 2016, PFM

When I pulled these data from the PFM, I was somewhat surprised. I expected there would have been more variance between the PFM and my prices. But even though there isn’t more than a five-dollar difference for any player in Table 3, it is not difficult to see how some of the differences are derived. The PFM doesn’t value speed nearly as much as I do, particularly with the SGP option turned “on”. Where the biggest difference comes, once again, is in batting average. Murphy falls all the way from 16^th overall to 36^th. The gap in valuation between my valuations and the PFM’s doesn’t matter nearly as much as I thought it did. But it does matter.

The reason that my values are different than the PFM’s and nearly any other mixed league valuation system is because I use an SGP model as the baseline for my formulas. For those of you who are new to pricing (or for those of you who are smart and don’t waste your time poring over this kind of data like I do), this means that the value of the 345 most commonly drafted hitters and pitchers adds up to $3,900. In nearly every other pricing system I have examined, the value of the best 210 hitters and 135 pitchers adds up to $3,900. This may sound like an esoteric difference, but the devil is in those pesky details.

Table 4: AL, NL, and 15-Team Mixed League Stats: Auction/Draft and Free Agents

Table 4 shows the 168 most expensive hitters in AL and NL only formats, the 210 most highly drafted hitters in mixed formats, and what the free agents did in each of these formats. If this sounds familiar, it is because I ran this table in 2015’s version of this article.

It is easy to see the difference between the undrafted talent in mixed versus in only formats. Using the best 345 players as opposed to the top 345 players drafted can alter the valuation landscape considerably in mixed leagues. There is nothing “wrong” with valuing players the way the PFM or other valuation systems do, but this isn’t measuring the value of a player at the time of the draft or auction, which is what I am primarily interested in when I perform this task at the end of every season. I’m interested in knowing what a player is worth at the auction or draft table, not what he is worth in November when the season is done.

Some balk at this. They argue that the 211^th best hitter in mixed leagues should be worth zero dollars. The PFM almost agrees. The PFM says the 211^th best hitter in 2016 was Eddie Rosario, who was worth exactly one dollar. This makes him interchangeable with Sean Rodriguez, Jarrod Saltalamacchia, and Bobby Wilson who were also worth one dollar. The PFM says that Lind earned $0.00. If your valuations are built around an open-ended universe where the 210 best hitters are rated using the benefit of hindsight, this is correct. Lind was not in the Top 210 and should not have been drafted, and was worth zero dollars. However, since I am using a draft valuation model, I reject this notion. Plug Lind into the draft rosters back in March and he was worth four dollars. I don’t believe that the PFM is wrong, but that my valuation model and the PFM are measuring two different things.

Theory aside, the next table is an excellent illustration of why going overboard on the top hitters isn’t logical.

Table 5: Top 15 Mixed League Free Agent Hitters, 2016

Both NFBC and LABR did a better job in 2016 than they did in 2015 at identifying some of these players, at least as reserve picks. Pick 435 was the cut-off in LABR Mixed (23-man rosters, 6-man reserve lists), so Martin, Davis, Prado, Owings, Gregorius, Napoli, and Pagan all made the cut in NFBC as key reserve picks, while five of the same hitters were nabbed on reserve in LABR. I missed on every player on Table 5, but then my preseason rankings stopped at 300. If I used my draft sheet from LABR instead, I would find one or two players in my Top 435.

Because of the variance in the free agent pool, finding trends is even more difficult than it is in mono formats. A simpler way of expressing this is that while every league is different, the differences are starker in mixed leagues. The shallower mixed leagues are, the less representative each league is.

Nevertheless, the lack of rookies in Table 5 is surprising. The expectation in mixed leagues is that you can snag a rookie from the free agent pool quite easily and get $10-15 of value. In 2016, there was Diaz and no one else. Table 5 is populated with many of the same, post-hype players who are often bargains in only formats. Nunez, Duvall, Napoli, Upton, Pagan…the list reads like a Who’s Who of players who once had value and/or significant expectations who were downgraded in 2016. This isn’t to suggest avoiding rookies entirely, but if you’re in the reserve phase of a mixed league draft or auction it is worth considering nabbing someone like Martin or Pagan as opposed to rolling the dice on Dalton Pompey, Rymer Liriano, or Jurickson Profar…players who were all selected earlier in LABR than the best free agents.

To be fair, the experts in LABR and the NFBC drafters followed these guidelines. There was a time where a rookie like J.P. Crawford would have been selected in LABR, but that time has passed. However, opportunity does exist in mixed leagues in 2017 to work the angles in two specific areas:

1) Low end stolen bases
2) Undervalued veteran bats

It will be worth looking at the NFBC results beginning next month to see if there truly is an opportunity in either of these areas for mixed league drafters in 2017 or if this was merely a one-year trend.