Finding the Gaps in Guesstimated vs. Projected Production

Ben Godar stole my article. I had the VEB equivalent of a “towel on the beach chair” by putting up a title (I think it was “Something Something PECOTA Something”) in our writer’s dashboard, when Ben came along and wrote the article out from under me! Stupid jerk.

But, then he crushed that article, finding the exact algorithm of hatred that PECOTA has put into its projections for the Cardinals.

If you haven’t done so, go read his article. In truth, Ben and I coordinated a little on this (occasionally us writers work together.) He provided the essential background information on this subject that I’m going to skip over. So, go read it.

Seriously, go read it. Right now.

Now that you’re back, my goal here is to take PECOTA’s embedded hatred and see how it compares with the other projection systems. Does PECOTA stand alone or do all computers just hate the Cardinals, Matrix-style?

So far, Fangraphs has projected win percentages for us. ZiPS provides overall team WAR according to Depth Charts’ playing time allotment. Those, plus PECOTA’s estimated win totals, provide three data points for Cardinals projections:

2020 Cardinals
PECOTA - 80 wins
Depth Charts win% - .491, 35.3 WAR
ZiPS Depth Chart - 35.9 WAR

That’s about as consistent a data set as you’ll find from three independent projection systems. The baseball mathematicians have slide-ruled the Cardinals to 80 wins and about 36 WAR.

Everyone wants to freak out about that and either throw the projections out or use them to prove their narrative about the front office. Don’t fall into either trap. Before we go any further, let’s just compare those projected WAR totals to the actual production of the club from ‘17-’19.

2020 projected WAR: 35.5
2019 actual WAR: 22.4 hitting + 15.5 pitching = 37.9
2018 actual WAR: 25.8 hitting + 14.1 pitching = 39.9
2017 actual WAR: 26.3 hitting + 15.7 pitching = 42

So, 2017 was our best team recently? Uh, no. That’s not really how this works. There is a correlation between team WAR and win totals, but, as you can see, there is also variance. That’s why it is best to understand both projections and team WAR as a range.

PECOTA places the Cardinals 50th percentile wins at 80:

Can you see that little hump to the right side of the 80-win line? That bump signifies that while 80 wins are the average simulated total, the sims spit out 80-85 win seasons more frequently than they do 80 or below. A large number of 81-85 win seasons where a few things go right for the Cardinals are being pulled backward by a long tail of 70-80 win seasons where several important things go wrong.

This club has an old core — Carpenter, Goldschmidt, Wainwright, Molina, Fowler — supported by a lot of unproven youth — Hudson, Edman, O’Neill, Carlson, Thomas. Simulations factor in injury and collapse for older players at a higher rate then they do players in their 20’s.

The projections are telling us that if the core hitters on the Cardinals can stay relatively healthy, they have a good chance at beating their 50th percentile win totals. But, if they don’t, things are going to get ugly very fast.

Understanding what those projections are really saying helps us to find the gaps between the Cardinals disappointing projections and our expectations of the club.

I’ve been perusing the comments here at VEB and Twitter and it seems to me that fans believe this club should be about the same as it was last year, give or take what happens in left field. Let’s put those sentiments into a range and name it what it is: a guess.

Fan guesstimated range: 88-90 wins
Simulated range: 80 wins, with an 81-85 win “mode” (most frequent).

Using our WAR projections and actual performance, let’s try to find the difference between these two. Where are the gaps between the Cardinals’ guesstimated and projected production?

John LaRue did a great job of addressing some of these questions in his recent article and I’m going to pick up pretty much right where he left off.

Photo by Dilip Vishwanat/Getty Images

Let’s start with left field. Many fans argue that between O’Neill, Edman, Thomas, and Carlson the club can find enough production to approximate what the departing Ozuna provided. Projection systems aren’t quite so optimistic. However, projections for young players are volatile, and the skillset of these players is almost certainly being undervalued.

You might have noticed, though, that John wisely didn’t center his analysis exclusively on left field. The more critical gains need to come from elsewhere.

If you’re looking for the gaps between the club’s guesstimated win total and simulated projections, look no further than the core of the Cardinals roster meeting their career expectations. The club has built the roster expecting certain key players to produce at or near their career levels. Simply put, simulated projections do not believe those players can come close to meeting club expectations.

Below, I’ve listed WAR totals that are a guesstimate of what the Cards are expecting from several core players based upon career production. Those guesstimates are then compared against projections to determine each player’s “production gap”.

Photo by Rob Tringali/SportsChrome/Getty Images

Matt Carpenter

Expected production - 4 WAR
Depth Charts - 1.9
ZiPS - 2.6
Production Gap - +1.4 - 2.1 WAR

Paul Goldschmidt

Expected production - 5 WAR
Depth Charts - 3.5 WAR
ZiPS - 3.1 WAR
Production Gap - +1.5 - 1.9 WAR

Dexter Fowler

Expected production - 2.5 WAR
Depth Charts - .8
ZiPS - .8
Production Gap - +1.7 WAR

Carlos Martinez

Expected production - 3.5 WAR
Depth Charts - 1.4
ZiPS - .7
Production Gap - +2.1 - 2.8 WAR

You can see how quickly the production adds up. If just those four players perform at or near career levels, the Cardinals could add 8.5 WAR to their totals. That would push the 2020 Cardinals to their highest WAR totals in this study, and could translate to an expected range of 7-10 wins added. That also doesn’t include any other production gained from tweaking the way the simulations project playing time or production (and there are obvious issues in both the offense and the rotation).

PECOTA recognizes this potential. This is why the high end of their win simulations extends into the 95 range. IF all of the Cardinals veteran players perform at or near their career numbers and nothing else major goes wrong, the team could be exceptional.

That doesn’t happen very often. It’s more common for players in their mid-30’s to continue to decline. It’s also common for young, unproven players — even good to great prospects — to flame out rather than blow up.

Hopefully, that helps put the 2020 projections in their proper perspective. Our guesstimate or “gut feeling” that this Cardinals team should win 88-90 wins is reasonable to us because we believe that our prospects should be pretty good and our veterans should bounce back. Some of that will definitely happen this season. But, some won’t. Unbiased observers feel that uncertainty creates a club with a wide range of landing points, the most frequent of which is mediocrity.

This spring, attention will remain focused on the big questions of the offseason: left field, the back of the rotation, and the search for a closer. The season probably will not be won or lost based on any of those issues. Projections think those issues will work themselves out in a satisfying way.

Our attention should probably be focused on the places where the season truly hinges: the production gaps found with Carpenter, Goldschmidt, Fowler, and Carlos Martinez. Let’s hope these players stay healthy and rediscover some of their youth. The Cardinals will need them.