Editor's note: the series from cdb on MLB performance outliers continues! You can check out their first post explaining their process here! And if you feel inspired to write something yourself, please check out our Fanpost section!
Time to start digging below the level of WAR. Fangraphs WAR for position players is built from three components, baserunning, fielding, and batting. These three components are aggregated as runs. More runs = better. So you can think of 'runs' as 'runs saved' for fielding, and 'runs created' for batting and baserunning. In addition to those performance characteristic, FG also uses three adjusters, also in units of runs: Positional adjustments, League adjustments, and defining Replacement Level (in runs). A full description of WAR can be found here: https://library.fangraphs.com/war/war-position-players/. Def runs are based on positional means, so if a player plays multiple positions, then the Def value reflects a total value over all positions.
It is important to keep in mind that these 'runs' are not actually 'runs'. That is, these metrics are not counting on field events to tabulate real runs. They are estimated runs given the events which occur on the field. In a bases empty situation, a single score exactly the same number of runs as a triple - zero. But a triple is worth more 'Batting Runs' because it is easier for that player to score on subsequent plays.
I wanted to look primarily at the distribution in the performance related 'runs' metrics, baserunning, fielding, and batting. Here is a series of histograms for each, looking at these values scaled to 600 plate attempts. Note here that these units are now effectively runs per 600 PA, not WAR, and that there is about 10 runs to a WAR unit.
There is a much wider distribution for Batting than Fielding, and for Fielding than Baserunning. This makes a good deal of sense: baserunning can only matter after you have gotten on base, and no one gets on base more than half the time, which already constrains the number of run scoring opportunities which are due purely to baserunning. Fielding is clearly important, but only when a ball is hit or thrown to that fielder. One can also make the argument that fielding runs are slightly less valuable than baserunning or batting runs: it is impossible for good defense to score runs, and scoring runs is necessary for a team to win. It is always possible for good defense/pitching to mask poor offense, and for good offense to mask poor defense/pitching. But imagine a scoreless game going into the ninth. run prevention can prevent a loss, but cannot enable a win on its own, offensive runs are necessary. In future posts i want to delve deeping into how these run metrics are built, but for now remember :
1. These are metrics estimating runs, not actual runs scored/prevented.
2. Order of impact, as estimated by FanGraphs: Batting > Fielding > Baserunning.
3. These distributions are roughly normal, centered at zero, where zero represents 'average'
4. Any player's WAR metrics are impacted not only by the performance on the field, but also by three adjusting values representing 'replacement level' runs, 'league adjustment' runs, and 'positional adjustment' runs. We are going to ignore those for this post.
As we think about total player value from a perspective on contribution to a winning team, this all means that a speedster on the bases cannot contribute many 'Runs', really. there is a ceiling to that productivity which is lower than Fielding, which is also lower than Batting. At least as estimated by Fangraphs. These distributions are estimated from models built on data, so one would hope that they reflect reality, if imperfectly.
For now, we are going to look at the unicorns for each of the three on-field Runs components. Note that using the data these data, we can re-calculate WAR from the formula as described in the above WAR linked glossary from Fangraphs, we simply sum all the 'Run' components and divide by 10.028 runs per win (the exact value varies year to year).
Now on to the unicorns. This is going to become a long post, but since we are looking at the relative contributions of these three performance metrics, Batting, Fielding, and Baserunning, it seems prudent to put all three of these into a single post.
NOTE: I have opted to look at this on a rate basis - per plate attempt - to get a better idea of who is performing the best on a rate basis, rather than who both performed well and played a lot. This means that we will probably find a few more unusual unexpected names, but also keep in mind that this data is filtered so that we have a minimum of 200 Plate Attempts. 200 PA is somewhat arbitrary, but there is no obvious breakpoint above 200 that would lead me to beleive 200 is a bad cutoff to use. 200 PA should ensure that we avoid a september callup's worth of data.
In this table i have listed the elite hitters, with 'BattingPer600' representing the number of batting runs per 600 plate appearances. This table looks remarkably like the table we saw in the WAR unicorns table from my prior post. I am going to go ahead and make the editorial decision to not spend too much time talking about this table - these are all familiar names, and i think that maybe the only surprise to me here may have been Byron Buxton, if it weren't for the fact that he also popped up in the prior WAR/600 unicorn table. Buxton appears to be a pretty elite hitter. i should clarify. he appears to have been a pretty elite hitter. His 2021 season was something special, though with relatively few plate attempts - he put up an OPS north of 1.000 in 254 plate attempts. It has since fallen to .832 in 2022, and .732 in 2023. When he hits the ball, he hits it hard, but when you are running a batting average of .239 (2023, 347 PA), you can't expect elite offensive numbers.
- Reverse Unicorns:
What stands out to me here is that there are actually a few players, despite atrocious offensive production, actually managed to put up a non-negative WAR value. These are the names that we may end up seeing in the Fielding and Baserunning Unicorns table, so i am going to just leave this table here for you to ponder.
These players, well, i can't say i know much about the vast majority of them. Perhaps this is unsurprising. Defense is not valued as much as batting - see histograms above. Note the maximum value here - Austin Hedges 2023 season. About 45 runs saved per 600 PA, far below what we observed for batting value. Note that i am keeping this per 600 PA, even though these are defensive runs saved, for the sake of consistency. This could have the effect of inflating the defensive numbers a bit - defensive specialists are more likely to be subbed in after an at bat to ensure the deficiencies their in their bat are minimized - they will tend to receive fewer plate attempts per inning played, on average. I also note that there are relatively few players on this list with more than 400 plate appearances. There are a few players who made the unicorn list in more than one season - Austin Hedges and Tyler Flowers - each of whom you will also find on the Batting reverse unicorns table. Defensive metrics are notoriously noisy, but clearly there is enough stability to demonstrate that these playes are not being randomly assigned to the unicorns list - Fielding runs has some predictive capacity.
Also of note: Byron Buxton makes this unicorn list as well. His 2019 season for Fielding has him accumulating nearly 2.5 WAR (about 10 runs to a win) over a full season. He only played for about half that amount though. I think that he is the only player that has made both the batting and fielding unicorns list. This is pretty remarkable, as he is now listed as a DH by baseball savant, after many good to great years in center for the twins between 2016 and 2022. Look at the similar hitters listed on Baseball Savant, and you will find some familiar names, including Giancarlo Stanton, Jorge Soler, and also NL Central names inlcuding Kyle Schwarber and our very own Nolan Gorman. It appears that Buxton has been used largely as a platoon player, consistently seeing twice as many plate attempts vs Righties than Lefties. However, it is not clear that this has been warranted, as his OPS are quite similar over the years, irrespective of pitcher handedness. his career OPS falls at only .768. His unicorn status on the batting chart seems a bit anomolous - maybe he is our TON - tantalizing tools, flashes of brilliance, ultimately unrealized potential. His draft status certainly suggests this - taken in the 1st round, second overall pick, by the Twins in 2012.
- Reverse Unicorns:
Couple of Cardinals on this list, including 2023 Knizner, and 2018 Munoz. Catcher defense is notoriously difficult to evaluate, so make of this what you will. If Willie McGee says he is the captain of the team... hmmm... i guess that doesn't really refute a negative Def runs statistic. In 2022, Robinson Chirinos was put up an impressive -43 runs per 600 over 220 PA, and has the distinction of being listed twice on this list. Again, evidence that this metric has some predictive capacity. I also find it interesting that on this list we find zero seasons with a total WAR over 0.5. I am actually pretty surprised by this. Bad defense, in theory can be more than offset by good offense - this is bad defense which is, at best, barely offset by good offense + good baserunning, in every case.
Baserunning is frequently assumed to be largely dependent on speed. When looking at the relationship between Fangraphs baserunning runs created vs the player speed metric, slightly less than half the variance in baserunning runs can be explained by player speed (r = 0.666 - yes, the number of the devil, or a less demonic R^2 = 0.44). That said, they are clearly strongly related. Note in this table the max baserunning runs produced is under 15 per 600 PA. Compare that to batting runs created. That said, the Cardinals are represented well here, with Tummy's 2019 season and Bader's 2018. Unfortunately, these values are each 4+ years old - last year's team was certainly much slower than the 2019 vintage.
In looking thorugh this list, you see a wide range of total WAR values for these players/seasons. All the way from Jose Ramirez's 8 WAR 2018 to a negative WAR value from Mag Sierra (former cardinal, noted speedster). I am going to spend a few words on Jarrod Dyson though - the oldest player on this list, and he derived nearly all of this season's WAR from his baserunning (at 10 runs per WAR, approximately 0.9 of his 1.3 WAR is baserunning). Dyson was drafted in the 50th (!!) round of the 2006 amateur draft by the Royals, and debuted in 2010. He peaked at 2.7 WAR in 2016, with a total of 55 accumulated baserunning runs between 2010 and 2021 (2833 plate attempts). He built a decent career out of speed and defense, having accumulated 55 baserunning runs, 65 defensive runs, and a brutal -27 batting runs over that timeframe (yes, that is negative 27). During this timeframe he earned a nice $18M. 50th round draft pic, earned enough to last a lifetime and then some, and produced some 13.6 total WAR for the teams he played for - seems like a win/win. Almost never helped his team with the bat (max of ~0.4 WAR in 2013), but made up for it with consistently good baserunning and defense (positive runs for each EVERY year).
- Reverse Unicorns:
oh, albert.... we still love you!
That said, we actually see six seasons where the worst baserunning performances still accrued more than 1 WAR on the season. This contrasts to fielding reverse unicorns, where our best seasons was 0.5 WAR. This does make some sense, given the magnitude of fielding runs is greater than that of baserunning - bad fielding can negatively impact your total WAR more than bad baserunning.
Bottom line: it becomes easier to understand why teams value batting so highly. While fangraphs is only one model, it is a model which some professional teams actually pay money for, so it is a pretty good one. A strong bat can completely mask bad defense and baserunning. Great defense and baserunning combined has a difficult time masking a bad bad. below average hitters can put together a decent career based on baserunning and defense - see Jarrod Dyson as an example - but they will rarely be the players receiving 8 figure contracts when they hit free agency.
Moving forward, i have a few paths i want to trod:
1. looking at team level trends in batting vs defense vs baserunning. spoiler - there are trends.
2. looking at the distrubution of the proportion of WAR that comes from these three areas.
3. i kinda want to look at WAR and its components by position - this is going to require some more coding, as it isn't reported in the table i retrieved. Ditto for salary.
4. start digging deeper into the more fundamental components which go into batting, baserunning, and fielding. there are literally more than 300 metrics to play with here, so i will not be hitting them all, but i will probably start by looking at some of the components of defense, as this is a great deal of mist and mystery surrounding it, relative to batting, and since it matters more than baserunning.