To much fanfare last week, Baseball Prospectus rolled out a new metric- DRC+. Of course, in Cardinal Nation, there were other, more pressing matters to discuss last week. With the Goldschmidt acquisition almost fully exhausted, now is a good time to dig in to BP’s new toy and how it relates to the Cardinals. If you’re exhausted by acronyms, DRC+ stands for “Deserved Runs Created” and the plus means it’s been placed on a 100 scale, just like other similar metrics like wRC+ and OPS+. You can learn more about it here, and here, along with several other articles they’ve written about it in the last week.
A quick summary is in order. Luis Torres from Beyond the Box Score provides a great summary of the differences between DRC+ and one of the previous gold standards, wRC+ from Fangraphs:
I have been a big fan of wRC+ for years. It is adjusted for league effects, park effects, and the scoring environment, so you can use it to compare players across eras. It is easy for anyone to understand. But of course, as with all stats, it had its shortcomings. It did not adjust for quality of competition. The standard deviation for it is never given, i.e. how different do two separate wRC+ values have to be in order for it to be significant? It does not tell us how much of the player’s skill went into that value. The park adjustments assume that players play all their games at home. This is not a problem when dealing with parks that are neutral or close to it, but it becomes one when going to the extremes.
DRC+ attempts to address all of these things. I’ll leave it to actual mathematicians to gauge how effectively it addresses these things, but I have faith that Baseball Prospectus has done their due diligence. The inclusion of the standard deviation for each hitter is very helpful. To quote one of the creators of the metric, Jonathan Judge:
DRC+ is willing to be uncertain. As with our catcher framing and DRA metrics, DRC+ has specific uncertainty bounds. Mike Trout, for example, in 2018 had a DRC+ of 180, with a standard deviation (plus or minus) of 13 points. So, we are pretty confident Mike Trout was somewhere between 167 and 193, and skeptical about him being in the range beyond that, but are happy to admit he very well could deserve something other than exactly 180. Being honest about uncertainty is an important part of understanding what it means to be accurate.
With that said, let’s take a look at how the 2018 Cardinals differed from wRC+ to the new DRC+. I’ve also included some players who are already traded, the newest Cardinal (Goldschmidt), and Bryce Harper. They’re sorted by the net difference. The more positive the difference, the more wRC+ has underrated the player by DRC+. The more negative the difference, the more wRC+ has overrated the player by DRC+.
DRC+ minus wRC+, 2018
One last step- here’s the same group of players with the standard deviation and confidence range for their DRC+:
DRC+ and Standard Deviation Range
From here, let’s look at some of the more intriguing cases and offer some explanations.
Bader vs. Fowler
The numbers in the table that scream at you the most are the DRC+s for Harrison Bader and Dexter Fowler. By DRC+, these two were equal hitters, which is an amazing concept if you watched them play during 2018. Bader became a fan favorite while Fowler had the worst season of his career. How on earth did that DRC+ result come about?
Quality of opposing pitcher isn’t the culprit, as they faced fairly comparable opposing pitchers. If anything, it favors Bader, who faced slightly tougher competition. The sample sizes are fine for both, with each receiving semi-regular playing time. Full seasons might have smoothed some of the data out, but it’s not like we’re talking about Carson Kelly or Patrick Wisdom and their less-than-100 plate appearance seasons. The ballparks don’t seem to especially favor one or the other. Fowler didn’t get to play in Colorado, while Bader did. But Fowler got more playing time in Cincinnati. The other ballparks seem fairly similar at first blush.
In looking for what makes up the result, we’re left only with the amount of skill used by each player that DRC+ finds reliable. In this case, Fowler gains value through walk rate (11.4% to Bader’s 7.3%) and more balls put in play (22.5% K rate vs. Bader’s 29.3%). Bader was also the recipient of a .358 BABIP, while Fowler’s was disastrously low at .210. Even if you assume there’s skill involved with BABIP- and you should- those are fairly extreme ends of the spectrum. Amongst players with 300 plate appearances, Bader was 17th highest and Fowler was third lowest.
In both cases, we’re talking about players whose performance fell on the opposite ends of the extreme of their standard deviation. Bader’s wRC+ (106) is within the single standard deviation for his DRC+, but just barely, landing right at the top edge. Fowler, on the other hand, falls below his single standard deviation with a 62 wRC+.
The Small Samples: Carson Kelly, Francisco Pena, Tyler O’Neill, Patrick Wisdom
The obvious point here is that these four have some of the highest standard deviations, with only Bader having more variance in confidence level. For Kelly, there appears to be four factors at play. One of those is the small sample. Additionally, Kelly faced a tougher batch of pitchers compared to his teammates, and his BABIP was an impossibly low .143. Finally, the bulk of his playing time happened in pitcher’s parks- 17 plate appearances in St. Louis, 12 in Petco Park, 4 each in PNC Park in Pittsburgh and Comerica Park in Detroit. All of it was compounded by the small sample size.
Pena’s BABIP wasn’t especially high (.278) but it came with a meager .068 ISO, a nasty 4.2% BB rate and 30.3% K rate, and his quality of opposing pitcher was a little worse than average (his opposing pitchers were a little below average in total average allowed). He’s just barely off the standard deviation- his low range is 33 in DRC+ and wRC+ had him at 32. It appears that DRC+ thinks his BABIP would have eventually evened out a little.
That’s a similar culprit for O’Neill and Wisdom, each beneficiaries of unusually high BABIP. Again, with the small samples for each player, the results get potentially noisier. With that in mind, O’Neill is past one standard deviation, but it’s not a number that jumps out at you. He had a 114 wRC+, and 104 was the top-end of his single standard deviation in DRC+. Wisdom is a bit more extreme but it’s 58 plate appearances. For O’Neill specifically, DRC+ dislikes his extraordinarily high strikeout rate.
Harper vs. Goldschmidt
There are two conclusions to be drawn here from looking at the DRC+s of Bryce Harper and Paul Goldschmidt. First, they’re both upper echelon hitters. Harper’s DRC+ was 13th best in baseball in 2018, and Goldschmidt was tied for 14th with Anthony Rendon. They’re separated by 0.3 DRC+. (side note: Matt Carpenter was 11th)
And that leads to the second conclusion. If you wanted the Cardinals to add the quality of Bryce Harper’s bat to their lineup, that’s exactly what they did. Obviously, they can add both if they want. They aren’t mutually exclusive. Still, the larger point remains- they added a hitter who was exactly as productive, by DRC+, as Bryce Harper.