How Deep are these Cubs?

Over the course of last two weeks, Fangraphs published its 2019 preview in the form of what they call ‘positional power rankings.’ It, along with their Playoff Odds, are a treasure for someone looking for analytical information regarding a particular MLB team.

Digging through the preview is useful for any number of reasons: It shows a team’s positional strengths and weaknesses. It condenses player projections into digestible buckets. It includes commentary from smart people who are particularly smart when it comes to the subject of baseball.

It also gives, taken as a whole, a fairly detailed look at a team’s depth chart. While the playing time projections aren’t supplied by organizations themselves, they’re a well-educated guess, and this particular means of presenting the depth chart offers hints as to how a backup’s production at one position might be worth more than that same production at another (for example, third basemen tend to be better hitters than shortstops, meaning a backup infielder, if their defense is equal between the two positions, is more valuable compared to a replacement shortstop than a replacement third baseman).

After reading through them, I decided to see how they’d serve when employed to evaluate a team’s depth—or, more specifically, evaluating how much weight is placed on the shoulders of individual members of a team.

WAR is a great tool for evaluating a baseball player’s value relative to the average replacement player at their position(s). What I wanted to do was compare teammates’ WAR projections against one another to evaluate a player’s value in the context of their team’s current roster. I had no idea how long this would take, or how effective it would be. So I tried it. And while I still don’t have a perfect answer to how effective it is, I have a better idea than I did, because I tried it with the team where I know the most context: the Cubs.

To measure a player’s value in the context of their team’s current roster, I created a metric called Wins Above Expected Replacement (WAER). Meaning, this metric is intended to represent how many more games a team would win with this particular player available, compared to how many they would win with only the player’s expected replacement (or replacements).

The idea behind this exercise is to identify how important each player is in the context of their roster. Practically, my hope was that this information could be used to label how damaging an injury would be, but more vaguely (and less morbidly), it could demonstrate which players a team is counting on to carry the weight. It’s supposed to be a measure of indispensability, or inversely, dispensability.

To measure this, I went through, player by player, removed them from the lineup, and extrapolated Fangraphs’ projected WAR totals for that player’s expected replacements.

For example:

Were Kris Bryant to retire today, before the first pitch, and were the Cubs to make trade to replace him for the entirety of 2019, someone would have to fill in his projected 651 plate appearances. According to Fangraphs’ playing time breakout, Daniel Descalso is currently projected to take roughly 58% of non-Kris Bryant plate appearances in the Cubs’ third base lineup slot, with David Bote receiving 28% and Ian Happ receiving the remaining 14%. Extrapolated across those 651 missing plate appearances, this means Descalso would pick up an additional 372, Bote would pick up an additional 186, and Happ would pick up an additional 93. Since Descalso has a projected WAR of 0.1 in those 58% of non-Kris Bryant plate appearances (28 PA’s in total), his new projected WAR is 1.4 (0.1 x 400/28 = 1.4, rounded). Since Bote and Happ each have a projected WAR of 0.0 in their combined 42% of non-Kris Bryant plate appearances, their new projected WAR is still 0.0. With Bryant, the Cubs have a projected WAR of 5.5 at the third base position. Without him, it drops to 1.4. His WAER, then, is 4.1.

As you’ve already probably noticed, there are flaws in this method. With this example specifically, it’s unreasonable to think Bote and Happ would be worth a combined 0.0 WAR in 300 total combined plate appearances. This happens because the data from Fangraphs cuts off at one decimal place, and the sample size of projected plate appearances for each of these guys is small. This is certainly a flaw, but the overall impact is limited, because these rounding errors tend to cancel each other out. While Bote and Happ’s WAR’s are probably rounding down to 0, it’s likely that Descalso’s is rounding up to 0.1. Still, the impact is there, and clearly the WAER figures upon which I ultimately landed have a confidence interval of at least a few tenths of a win due to rounding alone.

It’s good to note the confidence interval, because another systemic flaw here is those of Fangraphs’ own projections. All of this is based upon projections that are, true to definition, only estimates of what may happen. They’re also only projections of quantifiable statistics. They don’t capture any intangibles, which, depending on your view, might be a good thing or a bad thing, but at the very least bears mentioning.

These figures only go one layer deep, which creates limitations. Certain pairs of players, such as a platooning set of outfielders, may be more valuable in tandem than the sum of their individual WAER. Similarly, replacements may produce more efficiently in a small sample size than a large one, and I lack the knowledge about ZiPS and Steamer (the two projection systems from which Fangraphs is pulling their numbers) to know whether this is incorporated in them.

Obviously, I had to set limitations. The Kris Bryant example is an easy one, because he only appears as a third baseman in Fangraphs’ playing time projection, but for other players, their value is spread across multiple positions. To rectify this, I capped position players’ total plate appearances at 651, capped catchers’ total plate appearances at 500, and capped the five members of the primary starting rotation’s innings pitched at their projected total (in essence, if Yu Darvish were to disappear, these numbers assume his starts will be split among all options not named Lester, Hendricks, Hamels, or Quintana). Lastly, I only did these calculations for players with a projected total WAR, when adding the raw figures across all positions, of 0.2 or higher. Conveniently, this left me with 25 players. Inconveniently, not all of those are on the Cubs’ active roster to begin the season.

Which brings me to the final flaw I’ll note before giving you the numbers and accompanying observations: The Cubs don’t agree perfectly with these projections. They have their own, and they’re looking beyond immediate production, and when they are thinking of immediate production they’re occasionally thinking of it in the very short term, all of which I say to explain that I’m not disagreeing with their decision to start Ian Happ’s season in Des Moines.

Without further ado (1,000 words of ado was enough), here are the numbers:

Here are some things that jumped out to me:

Kris Bryant and Anthony Rizzo are really important.

This goes without saying, but asked to take a guess, I would have guessed that Bryant was the more important of the two. His raw WAR figure is higher, but Rizzo, according to WAER, is slightly less replaceable (though neither is at all replaceable, making last year’s 95 wins all the more impressive).

Mike Montgomery is valuable.

This might sound confusing, given that he has a negative WAER, but part of why no individual starting pitcher’s absence would cost the Cubs more than two wins is that Montgomery is a solid replacement in the rotation (Tyler Chatwood’s vast potential helps here too).

This shows a flaw in WAER that probably should have been self-evident to me: It’s good at evaluating the value of everyday players, but it’s terrible at evaluating backups, because the very thing it’s measuring is depth at a particular position.

The Cubs are strong up the middle—or at least interchangeable.

Javy Baez’s 0.3 figure is a black eye for this fledgling metric, but digging deeper, it’s more apparent why. Fangraphs expects some regression from him this year, and the Cubs have options when it comes to their middle infield: Ben Zobrist, Descalso, Bote, and, yes, Addison Russell are capable of contributing positively. Is Javy Baez an incredible baseball player? Yes. Clearly, this number doesn’t capture that. But it does capture that middle infield is an area where the Cubs are fairly deep, or will be should Russell return to the mix following the completion of his suspension (and as always, I add the caveat that the well-being of the survivor of Russell’s transgressions, and of all survivors of domestic violence, is infinitely more important than how many games the Cubs win this season).

Victor Caratini is more important than he should be.

Caratini is not going to be an all-star, but without him, you get a lot more of Taylor Davis behind the plate. Yes, Contreras would pick up a few games, but even so, the Cubs need Caratini given their current roster. Given that he posted a 65 wRC+ last year (that’s 35% below the MLB average hitter, if you’re unfamiliar with the metric), this is more an indictment of the roster than praise for Caratini.

The bullpen might be able to handle losing one guy.

The Cubs don’t have anyone so valuable in the bullpen that their absence would create a massive hole. As with Caratini’s importance, that may be more a function of lacking a strong top line than having a lot of depth. After all, it only means that parts are interchangeable, especially since this method handles the absence of, say, Steve Cishek by splitting his innings roughly evenly among Morrow, Strop, Edwards, Montgomery, Brach, Barnette, etc. Which might be fine with one absence, but creates big issues if more than one arm is down, which tends to happen in bullpens.

This is probably an observation more about bullpens in general than just the Cubs.

Fangraphs likes Ian Happ.

This, paired with their distrust of Albert Almora Jr., create one of the most striking figures in the above table: WAER’s claim that the Cubs would be nearly a full win better without Almora.

Looking more closely, I see that Fangraphs expects Almora to continue the regression at the plate that began last season while improving somewhat defensively, while they expect Happ to stay consistent at the plate as well as in the outfield (Happ got stronger defensive marks in 2017 than 2018, but I’d venture at least some of that has to do with him playing more second base in 2017). None of this is a glowing endorsement of the Cubs’ situation in center field. Something to keep an eye on.

***

In total, I don’t love this WAER metric. It isn’t precise enough to be useful in comparing players against each other unless they’re (1) different by an order of magnitude or (2) in no way a replacement player themselves. It requires context to avoid prompting unreasonable conclusions, and metrics like that are more easily spun. Still, it seems at least moderately useful in evaluating a team’s depth. I won’t use it again, but now we’ve tried it, and should we ever want to answer a similar question, we’d do a better job of it now that we’ve completed this exercise.