Overnight, the PECOTA standings were released by Baseball Prospectus. You can get to them from the tweet below and see that they have the Chicago White Sox with 83 wins this season. As much as we love the number 83, you’re probably in a frothy rage looking at it right now. Well, don’t smash your computer just yet.
Let’s look at what PECOTA means and what it does before we start to fill the streets like a scene from The Purge. PECOTA stands for Player Empirical Comparison and Optimization Test Algorithm. What it does is below and directly from the BP site.
The above excerpt is extremely important when you think about the 2021 White Sox, their past performance, and the volatility of small data sets in statistical models. Think of a card counter playing blackjack. At the beginning of the deck, they make average sized bets waiting for more information. As cards get used up from the deck, if those cards are small they begin to be able to predict the likelihood of face cards, tens, and aces to pop up. The 2021 Chicago White Sox are an unshuffled deck of cards right now. And it can go many different ways.
As mentioned, these projections are based on past performance. The White Sox have only been an above average baseball team for one year. That year had over 100 less games than any other year. Of which, only four returning position players played 50 or more. This lack of information exists for most teams, but it particularly hurts teams with younger, unproven players. Not to mention aging veterans.
PECOTA breaks up Pitchers and Batters, so let’s start with the Batters where we have a lot of volatility. As mentioned previously, we have very little 2020 data to work with and the position players for 2021 are going to look different than what the Sox had then. We’ll take a look at some of these to get an idea of what lands the Sox at 83 wins:
Anyone who looks into these rankings past the win totals will notice that Vaughn is projected to have a slash line of .220/.289/.335 in 400 PA. This tells us that they are projecting Vaughn to have a majority of the DH plate appearances, which I think we can agree is a reasonable assumption. But before you get mad at the poor computer for spitting out that slash line, think about what it has to work with: no 2020 stats, 245 minor league plate appearances, and 745 NCAA plate appearances. You’re asking for a projection of 400 major league PAs based on 990 PAs below AA. That is a huge ask. This does NOT, however, mean Vaughn will be bad. It means, it is unknown. It is difficult to project/predict. When that happens, it is best to err on the side of caution.
Nick actually gets a pretty nice projection. But what caught my eye is that he projects to have less PAs than most of the other starters. Based on injuries in previous seasons, the projection is reducing the amount of games he is expected to play. This assumption really changes the team’s ability to get on base. Pretty much any replacement listed here is going to be ~50 points lower in OBP. We can all root for Madrigal to be healthy and assume he’ll make better decisions as he learns, but past performance dictates otherwise. Hence, the projection limits his time on the field and increases the time of worse players.
Bench & Bench Mob
One of the great fears of the White Sox rebuild, is that the harvest won’t provide enough reserves for that harsh winter. I don’t know what I just said, but I said something. Anywho, I wrote about the bench, and according this BLOG HEREthe average AL Pennant winning bench of the last decade gets ~23% of the plate appearances. This one is rolling in about 18%, so I’d say we are probably due for some more of the guys you don’t fancy playing a little bit more.
Wait though, we didn’t use that measure in the blog, we only used the “Bench Mob” which is technically just the top 4 players. In order of plate appearance the top 4 are Collins, Garcia, Mendick and Engel. This is your bench mob, they are projected to hit….
The standard per the bench mob article for (aka the average for AL Pennant winners of the last decade) was .248 / .315 / .388. Not good enough.
Willy Mo Peña comp? Come on Baseball Prospectus, now we are a little upset. But it is understandable that computer projection alone, based on limited stats, does not agree with us on Robert being a world beater.
However, Robert actually did hit .233 / .302 / .436 in the shortened 2020 season while playing everyday. He was being tested by MLB Pitching for the first time, and both smashed and struggled at times. So for his PECOTA projection to be .234 / .298 / .427 OR almost exactly the same…I gots to scratch my head. Maybe they forget to do his projection and just told an intern “Just make it close to last year, nobody is gonna notice“. I hope I don’t get 2 seasons of bad luck Lewis Brinsons for making this joke.
Adding Lance Lynn is a nice upgrade, but leaving the 5 spot open for interpretation is going to hurt when it comes to expecting an accurate representation of your high expectations for the team. Now that said, if we look at DRA-, they are pretty kind to the bottom of the rotation. DRA- is Deserved Run Average minus. It tries to calculate the amount of runs the pitcher deserved to give up including fielding and park effect metrics to do so, then comparing them to the rest of the league. 100 is league average and then much like ERA, lower is better. For example, they expect Liam Hendriks to be pretty damn good, so he projects to a 66. They expect all of our 4-5 starter options to be below average, so they range from 103-106.
Kopech, who is a great talent, is projected to pitch the 5th most innings on the team. So how do you project the 89 IP of Michael Kopech? He has 14.1 IP in the majors and those came in 2018. This basically gives him no past performance. It’s rare you see a pitcher this young have 2 full seasons without data, but also be considered the 5th starter. The system is basically making a guess here and giving him a below average DRA- of 105. Can Kopech shoot the moon? Sure. Should a projection model expect that? No.
PECOTA is looking at Lance Lynn to be a lot more like the one we saw in 2020 than the one we saw in 2019. That’s a reasonable assumption based on the player’s most recent results and the fact that he is an older pitcher. However, we do know that in the shortened season there were a lot of trade rumors and uncertainty. It’s possible that those had an effect, but not reasonable to expect an algorithm to adjust much at all for that.
It’s been made known that Crochet will likely be working out of the bullpen this year. And whether or not you agree with that career progression for him, you probably do agree that he looked pretty good in the short amount we saw of him last year. That’s something that will excite us as fans, but not really sway a model that needs to make a prediction about future outcomes based only on 6 IP in the majors, 132 in college, and 0 in the minors. It’s the equivalent of predicting someone’s weight loss after the first day of their diet.
Anywho, Rick Giolito promised me that Cease was going to stick all this stuff where the sun don’t shine on me. All my talk about how the guy can’t hit the broad side of a barn and low and behold PECOTA comes screaming out with 68 BB in 124 IP for old Dylan. That’s 4.9 BB per 9 innings pitched. That would be an improvement from last year’s 5.25 BB, but not what I was promised. I think I want my money back.
The prize of the White Sox off-season!! I wrote a bit of about the risks of having a Liam Hendriks on the roster HERE. Most of the risk is surrounding over use as he was a bit of an iron man in 2019 with 85 IP and even in the 2020 shortened season he threw 25 1/3 IP, which would extrapolate to 68 2/3 IP over a full season. PECOTA legit shocked me projecting him for only 56 IP.
Overall, PECOTA treats the White Sox as most would when there are a number of unknowns, it pushes them to the middle. Outliers on either side rounded out in an effort to be more precise. The White Sox 40 man roster surely has question marks and when you have those pushing up into the 26 man roster and then up into the starting 9, you cannot expect to have favorable projections. That said, if you find a sportsbook that has the White Sox season wins at 83, let us know. We’ll be quick to place a gross amount of money on that over.
-BeefLoaf and Chorizy-E
Not sure if you have heard but we BROADCAST the podcast LIVE on our YouTube Channel! We even take questions! Make sure you subscribe for alerts when we go live.