clock menu more-arrow no yes

Filed under:

Which Team Component Correlates Best with Team Quality?

New, 156 comments

What’s the relationship to winning for offense, pitching, defense, and baserunning?

Divisional Series - Atlanta Braves v St Louis Cardinals - Game Four Photo by Jamie Squire/Getty Images

Last week, I took a look at a question that’s sure to be important to the Cardinals entering 2021. Namely, how good does an offense need to be for a team that emphasizes pitching and defense? Here’s what I found to be true for teams with top-10 pitching (by FIP-minus) and top-10 defense (by DEF):

...every extra point of [team] wRC+ is worth 7/10ths of a win. A team with a 100 non-pitcher wRC+ would expect a pythagorean winning percentage of .548, or 88.8 wins. Or at least, 50% of a .548 pythagorean expected performance could be explained by their wRC+.

Of course, the Cardinals in 2020 didn’t exactly qualify on the pitching front. If they want to be a successful pitching and defense team in 2021, they’d need to start by enhancing their pitching. Then they need to improve their team wRC+ by seven points from 93 to 100 to get to the threshold described above. Researching that question led to another question for this week. What exactly is the relationship between a team’s hitting, pitching, defense, baserunning, and overall team quality?

We’ll collect the non-pitcher wRC+, FIP-minus, DRS (Defensive Runs Saved), BsR (FanGraphs’ baserunning metric), and pythagorean record for every team from 2003 to 2019. Each is the FanGraphs version of the stat. I used 2003 as my cutoff because DRS isn’t available before that, and I suspect it’s the most accurate metric for team defensive quality. In the very least, it’s as good as any of them, particularly when taken in the aggregate for a full team over a full season. I omitted 2020 because of the absurd sample size and wild variance in team health. It’s hardly a normal season and won’t tell us much about teams moving forward.

I used wRC+, FIP-minus, BsR, and DRS in a multiple linear regression to predict a team’s pythagorean record. It came back with a .85393 r-squared. That’s... very good? I think? It makes sense, too, since so many of these stats are created with overall team quality in mind. If you’re curious, here’s the formula.

Predicted Pythag = 0.51441 + (0.00427 * wRC+) + (0.00036 * DRS) + (0.00068 * BsR) - (0.00441 * FIP-minus)

Using that as a prediction formula gives me an average difference of 3.4 pythagorean wins between predicted and actual for the 510 teams in the sample. Now we have a baseline to see how a team’s predicted wins will change based on how they increase or decrease in each of the four categories (wRC+, FIP-minus, BsR, DRS). Here’s how that works out:

  • One point of wRC+ (say, going from a team wRC+ of 99 to either 98 or 100) is worth .6917 wins over 162 games.
  • One point of FIP-minus is worth .714 wins.
  • One Defensive Run Saved is worth .058 wins.
  • One point of BsR is worth .1102 wins.

Obviously and as to be expected, pitching and hitting have the biggest impact. The surprising part, however, is that improving team pitching actually outweighs hitting by a little bit. It’s not a massive gap, but it’s still something. Improving a team’s FIP-minus by 20 points would be worth 14.29 pythagorean wins. Doing the same for a team’s wRC+ would be worth 13.83 pythagorean wins over the course of a full season. Those are extreme cases representing humongous improvement for any one team from year to year, and at its max, it still only comes out to half a win of an edge for pitching.

Defense and baserunning lag far behind, but they’re not insignificant. We don’t have to look far for a real example. The Cardinals had 22 DRS and 13.0 BsR in 2018. The following season, those numbers were up to 91 and 14.7. The defensive improvement was worth almost 4 wins exactly, and the baserunning improvement got them... well, not much. It comes out to about .18 pythagorean wins. Add them up, though, and you’ve added 4.2 pythagorean wins. Now consider that the 2018 squad had 88 pythagorean wins and missed the playoffs. The 2019 squad made the playoffs with 92.

Those little margins in defense and baserunning can add up. It’s especially true of defense because a lot of defensive improvement can be made simply by leveraging the right defenders into tight spots, better positioning, and shifting when appropriate (not to be confused with simply shifting more). In other words, there are wins to be had on defense even without spending to get them on the open market.

Wild Card Round - Houston Astros v Minnesota Twins - Game Two
Having a player like Jose Berrios instead of Dakota Hudson would have represented a 3 point improvement in team FIP-minus for the 2019 Cardinals
Photo by Brace Hemmelgarn/Minnesota Twins/Getty Images

How Much Difference Does One Player Make?

All of this raises the question of how much any one player can improve a team’s overall fortunes by improving a team’s wRC+ or FIP-minus. I’ll focus on those two specifically since they offer the most impact.

Let’s take the 2019 Cardinals, but we’ll replace Marcell Ozuna’s production at the plate with 2019 J.D. Martinez. We’ll assume that the production in the field and on the bases were the exact same as what Ozuna provided, and we’ll prorate Martinez’s batting line to Ozuna’s plate appearance total. That represents a bump of 30 wRC+ from Ozuna’s 109 to Martinez’s 139 over 549 plate appearances. Doing so would lift the team wRC+ about 3.4 points. That’s a decent bump.

How the 2021 Cardinals get there is up to them. They don’t have to find a single player who is 30 points better than their 2020 counterpart. They can find multiple players who improve by 10 wRC+, or two regulars who improve by 20 wRC+, or... you get the picture. I just thought it’d be instructive to have an example.

We’ll do the same 30 point increase on the pitching ledger. We’ll swap out Dakota Hudson’s 174.2 IP and 114 FIP-minus with Jose Berrios and his 84 FIP-minus, again prorating his numbers to Hudson’s 174.2 IP. Doing that shaves the team FIP-minus down by 3 points. It’s pretty comparable to the hitting scenario. A 30 point improvement from a regular will improve the team total by about 3 points, compared to 3.4 for a hitter.

What Does This Mean for the Cardinals?

I chose the 30 point jump because it’s the kind of big improvement you might see in a larger, but not necessarily top shelf, free agent acquisition. Given the likelihood of the Cardinals sitting out most free agent signees on the top shelf this year, most people would be surprised if they found a single 30 point upgrade in either wRC+ or FIP-minus.

However, they have several reasonable bounceback candidates. For instance:

  • Dylan Carlson, 65 wRC+
  • Tyler O’Neill, 70
  • Kolten Wong, 92 (ok, not this one)
  • Tommy Edman, 90
  • Paul DeJong, 86

Every one of those players is capable, and even likely, of boosting their wRC+ by 10 points. In some cases like Carlson and O’Neill, a 30 point swing to the positive shouldn’t surprise anyone.

Pitching is a little murkier because of injuries and some overperformance in 2020, so I won’t belabor that too much. Just know that the list of potential improvers on the pitching side is much shorter, bordering on non-existent. The two exceptions are Jack Flaherty (94) and Daniel Ponce de Leon (129).