Simulated season on Baseball Reference

[A. Cavatica]

Many of you undoubtedly know that Baseball Reference has been simulating an entire 2020 season.

https://www.sports-reference.com/blog/2020/03/baseball-reference-simulating-2020-season-with-out-of-the-park-baseball-21/

Obviously the records and individual stats are imaginary, but it does offer a perspective on how a shortened season could play out. 67 games is almost as long as the owners have been willing to go.

The division leaders at this point are the Yankees, Indians, As, Braves, Cardinals, and (.742!) Dodgers. The Mariners and Nats are nearly tied with the As and Braves. I personally think the Indians, As and Cards have overperformed but only the Mariners stand out as a Cinderella team. A playoff bracket of those 8 wouldn’t be a travesty.

The Sox took off like a rocket, regressed to .500, and are now sitting at 35-32, 8 games behind the overrated Indians.

The simulation has many flaws. Just looking at the Sox, Robert didn’t start the season with them and Madrigal still hasn’t been called up. Bullpen spots that would have gone to the winners of spring training battles were instead allotted to unsigned nobodies like Tony Cingrani. The batting orders are a joke; even Ricky wouldn’t bat Mazara second.

One thing, sadly, rings true: EE is batting .159 and Ricky is sticking with him.

[Dumpjerry]

Many of you undoubtedly know that Baseball Reference has been simulating an entire 2020 season.

https://www.sports-reference.com/blog/2020/03/baseball-reference-simulating-2020-season-with-out-of-the-park-baseball-21/

Obviously the records and individual stats are imaginary, but it does offer a perspective on how a shortened season could play out. 67 games is almost as long as the owners have been willing to go.

The division leaders at this point are the Yankees, Indians, As, Braves, Cardinals, and (.742!) Dodgers. The Mariners and Nats are nearly tied with the As and Braves. I personally think the Indians, As and Cards have overperformed but only the Mariners stand out as a Cinderella team. A playoff bracket of those 8 wouldn’t be a travesty.

The Sox took off like a rocket, regressed to .500, and are now sitting at 35-32, 8 games behind the overrated Indians.

The simulation has many flaws. Just looking at the Sox, Robert didn’t start the season with them and Madrigal still hasn’t been called up. Bullpen spots that would have gone to the winners of spring training battles were instead allotted to unsigned nobodies like Tony Cingrani. The batting orders are a joke; even Ricky wouldn’t bat Mazara second.

One thing, sadly, rings true: EE is batting .159 and Ricky is sticking with him.

Dave, I mean Cavatica, what do you mean the computer isn't accurate?

[A. Cavatica]

I apologize, Hal.

Seriously, computer simulation is my real-world job. A simulation is only as good as the data and assumptions that go into it, but a simulation can still be worthwhile in aggregate even if many individual data points and assumptions are wrong. If they played a 65 game season, the Mariners almost certainly wouldn’t be in playoff contention, but it’s possible they would be and likely that some other Cinderella team would be.

For other examples, consider individual COVID-19 models and hurricane forecast models. Most of the raw predictions are wrong for any given run but if you run enough simulations and average the results, they do a pretty good job.

I think it’s interesting to watch the B-R simulated season despite its flaws.

[FourGoldGloves]

Looked to me like the simulation started off with the Sox way too hot so someone realized it was out of line and turned a dial or two to settle the Sox down to where everyone expected them to be, and they dropped like a rock. That creates a dilemma. Does it destroy the objective credibility of the simulation if there was human intervention, or did/could it ever have anything like objective credibility?

Early on I was thinking they should start the season for real with the records established by the simulation. Now not so much.

Abreu hitting .244? Really? Where could that come from?

[A. Cavatica]

I don’t know how much they have tweaked the simulation, although they have obviously done some tweaking (e.g. bringing Robert up). I think the results feel like normal variation.

According to Baseball Reference, the most similar batter (statistically) to Abreu through age 32 was Josh Hamilton. Hamilton at 32 batted .250. Hamilton at 33 batted .263. And Abreu batted .253 in the first half of 2019.

[Krs1]

The Sox lineups that I just perused made me not care at all about this. Could have been cool.

[Dumpjerry]

Do they take into account injuries? That is a huge variable if a stud player "is injured."

[A. Cavatica]

Not sure about injuries.

At least the simulated players and simulated owners are getting along.

[vegandork]

They're using OOTP baseball to simulate the games. They tend to undervalue younger guys or guys that just got good. Starting a season as the White Sox is kind of a nightmare. Giolito usually ends up being back end for his career, Robert never develops the ability to hit for average, and Madrigal is never worth a roster spot.

[A. Cavatica]

I checked in on the simulated season to see how it compares to reality. The SimSox are 59-62, about where our winning percentage was two games ago.

SimYasmani is leading the team with a .931 OPS, while the real one is at .656. Advantage: SimSox.

SimNomar has 34 HR and an .866 OPS. Advantage: SimSox.

SimLeury somehow has a .906 OPS, with a small sample size of 137 AB. I thought the real one was overachieving at .758.

SimEloy is doing a little better than the real deal, .803 vs .732.

SimYoan is at .872 and flesh ‘n’ blood Yoan is at .800.

Players who are currently close to their simulated stats include Abreu, Encarnacion and Robert. Abreu’s batting, on-base and slugging are each within .014 of the simulation! EE is sporting a simulated .593 OPS, only slight below the real .632.

The players who in real life are crushing their sim counterparts are Anderson, McCann, and Engel — projected for only .682, .535 and .452 respectively.