JotCast | Dan Szymborski FanGraphs Chat 7/23/26

Return to

Dan Szymborski FanGraphs Chat 7/23/26

Prev Page View Full Chat Next Page

This chat is available to members only. Please log in to participate.

Dan Szymborski

12:25

he actually did 60.1

12:26

ZiPS projected, combining preseason, as a 169 wRC+ from 2013-2020

he was 1756

175

benranieri

12:26

What would you do at the deadline if you were the Mariners?

Dan Szymborski

12:27

I'd seriously try to go after the best DH/cOF bat that I could. They tend to do a decent job cobbling together positions, but I want a plug and play guy

BuccoFan

12:28

Hi Dan - how does Zips see Esmerlyn Valdez going forward? Obviously he won't sustain this, but as a 22 year old putting up this level of production, does he seem like a star? Or a guy who caught fire for ~125 PA?

Dan Szymborski

12:30

ZiPS is actually pretty sold on him as a hitter

i mean, not where he is NOW

but he's arlready projecting as an average cOF/DH/1B type going forward

12:31

(I know he has only been used a little at 1B, just illustrating where his offense is)

JustinPBG

12:31

Paul Blackburn is good?

Dan Szymborski

12:31

I think I'm at decent as a reliever, not at GOOD yet

though I coudl be wrong

and I am plenty!

SteveMcAnderson

12:31

You (ZIPs) were the most optimistic aboutt he brewers pre-season hypothesizing something to the effect of-- ZIPs likes a lot of their players and while other orgs wouldn't play them, Milwaukee will, so I'm factoring that into the playing time model and thus they get a bump in the projection.

100+ games in, how would you assess that decision and hypothesis?

Dan Szymborski

12:32

Still missing low, lol

though I don't know if that's the why SPECIFICALLY in 2026 so far

might be worth looking at post-deadline

mikeladd123

12:32

How much influence do 2026 numbers play into the ZiPS Update values? Same question but concerning ZiPS ROS numbers? ZiPS & Steamer seem to be the only two sites the truly reflect current year results.

Dan Szymborski

12:32

Quite a bit 2026 is the most important data for a player, though fans tend to make it TOO important, because recency bias is a thing

12:33

ZiPS mixes in the new 2026 performance into the data every morning. Simply in the FanGraphs model (since it has to be run every morning), more robustly in the so-called full-fat model

This is important. It takes more than a day to update all the proejctions

Yojimbo

12:33

Is Random Forest modeling useful for finding the best variables to include in a predictive models?

Dan Szymborski

12:33

It can be, but it's not really the approach I take, and it would require a fundamental recasting of ZiPS

12:35

I guess ensemble-like nearest-neighbor would be the fairest technical way to describe zips

thephien

12:35

When ZiPS draws comparables from across baseball history, how do you weight it translate old careers for a modern player?

Dan Szymborski

12:36

Less so. ZiPS gives greater precedence to more recent comps

And deadball era players just cloud everything. Baseball was just TOO different prior to 1920

Sandy Kazmir

12:37

Why are so many folks who cover the game saying that Luis Arraez would be a good fit for Tampa Bay?

Dan Szymborski

12:38

There's been this weird vein that the Rays are beating modern analytics by not being obsessed with home runs that doesn't make sense. And Arraez has long been the "analytics is bad as they typically don't love Arraez in 2024-2025" guy

It's weird

opifijikl

12:39

What is your favorite traditional stat? For example mine is RBI.

Dan Szymborski

12:39

Triples