CSS Button No Image Css3Menu.com

Baseball Prospectus home
Click here to log in Click here to subscribe

Chat: Harry Pavlidis

Chat Home

Welcome to Baseball Prospectus' Friday February 08, 2019 11:00 AM ET chat session with Harry Pavlidis.


Join members for the BP Stats team for a PECOTA release week chat. Ask us about anything stat related, including DRC+1.1. Moderated by Harry Pavlidis, the Director of Research and Development at Baseball Prospectus.

Harry Pavlidis: Good morning, welcome to the first of our 2019 Stats chats. This week the featured topic is PECOTA, but everything is fair game. This is Harry, but I'm joined by a whole bunch of our staff and friends. So you'll get useful answers!

Vic (Baltimore): So who are some traditionally big name OFers who will be outperformed by Ramon Laureano this year?

Harry Pavlidis: Depends on how you define big name, but PECOTA does have him pegged to beat quite a few guys. Matt Trueblood wrote about this ... we'll find the link for you and update this ... found it https://www.baseballprospectus.com/news/article/46994/the-lineup-card-13-noteworthy-pecota-projections-2019/ (item 3)

Dusty (Colorado): Thoughts on Twins prospect Wander Javier? What's his upside?

Harry Pavlidis: Jeffrey Paternostroís next chat is Thursday, February 14th - submit your prospect questions here! https://legacy.baseballprospectus.com/chat/chat.php?chatId=1527 (Rob Mains says hi, Dusty)

Guancous (Clark & Addison): Does PECOTA foresee any major macro shifts in the general environment next year due to injury and attrition?

Harry Pavlidis: PECOTA relies on depth charts (humans) for playing time estimates, so not directly, no. We have a somewhat related response to another question...stay tuned

Duhbear (Vermont): As a Rays fan, I generally understood the projections for the team. However, one that stood out was Austin Meadows, who is projected to produce -0.5 WARP with a 77 DRC+ over 450 ABs. Brandon Lowe, after struggling in 2017 and having a similarly productive season in 2018, is projected for 1.1 WARP. I get Meadows didn't produce a ton in 2017, but that projection is still a pretty steep drop in performance for a guy that put up a 150ish DRC+ at two AAA stops and topped it off with a 95 DRC+ in the majors as a 23 year old. I know there's it's not really a subjective matter, but what do you think is dragging his projection down?

Harry Pavlidis: Good question. I think the issue is how much weight you put into each stop, which is going to be based both on recency and, naturally, amount of plate appearances. Also, MLEs are hard.

Bearry (VT): Does Vidal Brujan have the most exciting comps in the minors outside of Vlad Jr? Because DAMN, there are four current MVP candidates there and another former all-star for good measure.

Harry Pavlidis: PECOTA has Betts, Ramirez, and Schoop. I think Schoop is there to keep you grounded. There's a funny thing about prospects and their comps--they can start really nice but they can start to shift away to normalcy after some exposure. Aaron Gleeman had a funny note about how Byron Buxton and Miguel Sano have gone from gaudy to pedestrian comps. And by funny note I mean deeply saddening.

mitch (Philadelphia): where do you think bryce and manny go and how will it impact the proejctions?

Harry Pavlidis: Going to depend who they replace in the depth charts. Harper projects to 4.5 WARP, Machado to 3.7 -- let's call it four and make it easy. If they are replacing a decent everyday player, they may "only" add a couple wins to the bottom line. That's enough to change the standings. But if you have an already strong team with a glaring weakness that these guys can correct, well, in that case it's quite possible a club could jump from pretender to contender (does anyone still use that phrase?)

VB (Tacoma): Do you think the idea of "player types" still holds up, as we have better data and do updates to DRC, etc? As in, should we still be grouping players into "Quad-A Slugger" categories, or are there now better ways to create more accurate sub groups?

Harry Pavlidis: This is one of my favorite subjects, because I think there is something to be learned about how player comps are developed. That thing about the attrition in quality of the comps, I think that's what projection systems should learn to avoid! Get it righter faster. Since comps are mostly stat based, there are a couple fronts where we can make progress both for PECOTA and for general understanding of player development paths.

One is to do a better job of handling useful things about the players' paths to date--when were they drafted, where did they play and how old were they--things like that which will help us with groupings earlier.

Related is doing a better job of selecting the stats and understanding 'which ones matter'--I'd say our methodology there is dated, and we want to improve that.

Lastly, we want to talk to more experts about what types exist, and not just fish the numbers for them.

Coach (USA): What is the one piece of currently unavailable data that, if you had it, would improve projections the most?

Harry Pavlidis: We could really use final 2019 pitching, hitting, fielding, and running stats. Short of that, playing time down to the exact number of plate appearances. Failing that, truly detailed health and injury data (we'll sign the HIPAA forms, we promise). And if we have to settle, a massive collection of scouting scores, with redundancy and overlap, covering several decades, at least.

jg (Chicago): Do you guys have any plans to share the distribution of win totals from teams' simulated seasons?

Harry Pavlidis: it's been discussed, it's feasible, but planned is too robust a term. But it would be neat to do so. Once we start up the playoff odds engine we'll possibly see if we can make that happen sooner than later.

Common (Detroit): What were the most surprising projections, good or bad?

Harry Pavlidis: Our writers had a lot of fun with this one: https://www.baseballprospectus.com/news/article/46994/the-lineup-card-13-noteworthy-pecota-projections-2019/

Derek (Minneapolis): Interesting, Buxton #1 project is Austin Jackson right now. Good or bad?

Harry Pavlidis: It depends on how you look at it. This is one we alluded to earlier--Austin Jackson is not a bad Major League player. But when you start out looking like your comps are HOF quality, it can feel really really bad. Which probably isn't fair to anyone.

boatman44 (Liverpool): Cleveland with 96 wins from PECOTA,please explain.My own projection system (gut feeling), has them barely scraping 88, Is PECOTA's love of my Indians purely to do with the putrid division they're in?

Harry Pavlidis: Yea, we think you have a weak division and a projected front three of Kluber/Bauer/Carrasco ... and PECOTA likes their rotation even as you go further into it ... and you have a good strike preserving catcher .... you get a big win total. It does feel high, but it seems justifiable upon inspection.

Harry Pavlidis: Thanks for coming to our chat! We got some ideas out of it, and we hope you enjoyed it and learned something, too. We'll see you next time...

Baseball Prospectus Home  |  Terms of Service  |  Privacy Policy  |  Customer Service  |  Newsletter  |  Masthead  |  Contact Us