JotCast | Dan Szymborski FanGraphs Chat - 3/25/21

Dan Szymborski FanGraphs Chat - 3/25/21

powered by

Prev Page View Full Chat Next Page

economist

12:34

OK, I sent the last response before I saw your full reply (sorry, I'm a jerk!). Anyway, the other thing I think you'd get here is some deeper insight into the performance of a front office. The good front office will manage future holes to line up with markets where the price per win is likely to be low.

Dan Szymborski

12:34

Wait, where are you being a jerk?

The problem is that there are relatively few free agent decisions made

We're hampered by the sparseness of our reality here.

Appa Yip Yip

12:35

If ZiPS got Weird Science'd what would its human form look like?

Dan Szymborski

12:36

cool

1

laughing

1

BringBackDeSpanielHair

12:36

Matt Harvey makes Orioles Roster: Discuss

Dan Szymborski

12:37

I don't expect much, but his velocity was up and he was more effective after that first appearance

Bob Loblaw

12:37

Blink once for "yes," twice for "no:" have you ever used ZiPS for betting purposes?

Dan Szymborski

12:37

Of course! I've explicitly done so!

12:38

https://www.espn.co.uk/mlb/preview13/insider/story/_/id/9111343/best-p...

TooDamnTall

12:39

At this point in you and Zips' career, do you enjoy changes in the game to see how responsive Zips can be, or would you prefer lengthy stability?

Dan Szymborski

12:39

I like stability!

I'm less focused on mean projections than the other stuff

Joe

12:40

On Correa, we know he and Mike Elias are somewhat close. Any chance he comes to Baltimore?

Dan Szymborski

12:40

I guess it's possible, but I don't think it's likely at this point

Alby

12:40

You said last week that cats shouldn't have human names. Does Joey Bag of Donuts count?

Dan Szymborski

12:40

It's OK to have a human name as PART of the name

Tim Tebow's Thunder Thighs

12:40

So a buddy of mine decided to make "pizza burritos". Is that not just a Chipotle sized pizza roll? Or does the fact that he used a high fiber tortilla in fact qualify it as something else? If I made the same with pizza dough would it be a true, overgrown pizza roll or just a monstrosity?

Dan Szymborski

12:40

I think a pizza burrito is just a mediocre calzone

Tim Tebow's Thunder Thighs

12:40

Now I am considering making a chili filled, burrito sized pizza roll with proper pizza dough.

B

12:40

Do your projections give greater weight to 2020 for metrics that stabilize more quickly? Would you give greater weight to K% that, say, ISO because K% stabilized more quickly (I assume) even though the two are correlated?

Dan Szymborski

12:40

Yes

Though they already do so

Guest

12:41

If you discovered some ultra secret version of ZiPS that was 100% accurate in predicting MLB win totals, would you keep it to yourself and just bet those teams and make a fortune? Because once you release it wouldn't the public/casinos try to backward engineer it and then negate your advantage? Would you still release the "lesser" version of ZiPS for all us plebes?

Dan Szymborski

12:41

If I had that, I'd just bet myself, make a fortune, and retire from baseball writing

If I had this kind of magical accuracy, baseball would be utterly boring for me to watch

Tim Tebow's Thunder Thighs

12:41

On a more serious note, how often do you/have you updated ZiPS with newfangled statcast/trackman/etc data such as EV, Barrel%, etc? Do you wait a few years to verify that these measurements are accurate and useful or do you trust some right from the get-go?

Dan Szymborski

12:41

Oops, sorry I keep answering so many of your Qs

12:42

(to others)

I try to be random, but there were a lot

(don't want anyone to feel left out)

There's always ZiPS versions that are a few years out. I don't like including things until I have a good fix on the nature o their predictive value

economist

12:42

I agree sparse reality is hard. That's why I'm suggesting just counting number of fits (could probably scrape off of MLBTR's free agent overview of something similar) and the count of these teams trying to win and the coiunt not trying to win. So $/win = #fits + constant + error or perhaps dividing #fits by trying to win. It's not "right" but it will help you understand if it's important. And FYI the academic literature definitely does fit auction models with relatively few observations to identify things like underlying valuations and whether collusion is trying to take place. Thanks for your patience!

Dan Szymborski

12:42

I'm quite behind on the academic literature!

Morbo

12:43

Two questions: when using zips to bet do you make your bets before you make zips data public? (I'd assume yes) Do you make a bunch of smaller bets to try and take advantage of let's call it diversifying your risk, since the law of large numbers would require too many bets to be practical for an individual, or a few larger ones?

Dan Szymborski

12:43

I don't really have a pattern

Generally, I bet when I think about it

Kimi

12:44

Curious if you had to rank these 4 recovering aces for 2021, how would you rank them, Sale Clevinger, Syndergaard, Severino? Should we consider them likely to be fully recovered by then?

Dan Szymborski

12:45

Generally, Syndergaard >>>>>>>>>>>>>>>>>>>>>>>>>>>>> Sale/Severino >>>>>>>>>>>>>>>>>> Clevinger

Connecting…

Register

Message History

On Off

On Off

On Off

Your current subscriptions

None

Register now to continue managing your subscriptions

E-mail Address	Chat Owner	Status	Remove

Enter your email address to be notified when Dan Szymborski hosts a chat