Yet another new PECOTA spreadsheet was released today. Special thanks again goes out to Colin Wyers and Rob McQuown for continuing to enhance and update the product. Fantasy and Premium subscribers can download the spreadsheet by clicking here.
​Highlights from the latest update:
- Total Bases (TB) has been fixed.
- Sacrifice Hits (SF), Sacrifice Flies (SF), and Double Plays (DP) have all been added to the batters' spreadsheet, and Quality Starts (QS) are now included on the pitchers' spreadsheet.
- Both batter and pitcher spreadsheets now contain hyperlinks to player cards and team depth chart pages.
We will continue to make tweaks and adjustments as necessary and will soon be publishing a FAQ article that addresses many of the comments and questions surrounding PECOTA.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now
not able to find him in the pitcher's tab
I understand that an immense amount of work must have gone into converting PECOTA from Nate Silver's initial Excel spreadsheet-palooza and in maintaining/updating the process since - and I respect the hell out of everyone at BP who's gone through this every year - but simply put, PECOTA hasn't been "right" since it left Nate's hands. There are a few tangible issues - the one raised here and another I repeatedly raised in a Beta testing page (and over e-mail) a couple of years ago that was ignored - and many more intangible "that just ain't right" issues (e.g. Rockies' position players last year).
At this point I use PECOTA as just another data point to provide a forecast (among the swath of other projection systems), and when PECOTA provides an outlier I assume it's the result of an error rather than some prescient bit of information (e.g. the famous Wily Mo Pena forecast or Brandon Webb's insane comparable list the year he got called up).
I've been pretty patient with the process - at least I think I have been - but at this point I'm just sad. I've given up hope.
As always - BP's willingness to make the process transparent is appreciated, as is the hard work to try to get this right.
Regarding the comparables issue - leaving other issues aside for now - the sticking point for me is less that some comparables are clearly wacky, but that the *projections* are influenced by the errors in those comparables. The comparables obviously drive some portion of the projections, and because this one visible element of the system is clearly wrong, our confidence in the projections themselves is reduced. Given the presence of other issues over the last 2 years, for me personally my confidence in the projections is REALLY low.
Again, I almost feel guilty discussing this openly. I've just reached the point where I've lost all confidence that Nate's old PECOTA system - a system which I trusted to tease some meaningful relationships out of the data that weren't being picked up elsewhere - will return.
What we do first is come up with a “baseline projection†that takes into account a player’s past performance, accounting for some regression to the mean. (We break it down based on components, so that a player with a high strikeout rate is treated differently than a player with a low strikeout rate, even if things like batting average are the same.)
Once those baselines are generated, we find comparable players to those baseline forecasts (as well as to some attributes not reflected in the baselines, such as age, position, height and weight). We look at the performance of the comps going from their baselines to the following season to come up with composites-based age adjustments (internally to PECOTA this is referred to as a “cage†forecast). We assess the reliability of the composite-based aging curve, and use that to regress it towards a generic age adjustment for players at that age (we refer to this as a “genage†adjustment). The less reliable a composite-based age adjustment is, the further it’s regressed.
I'm sure there were valid reasons for doing so, but still -
I'd rather know a guy is comparable to some minor-league shlub I never heard of than have the implication out there that Justin Christian is Manny Mota, or Everth Cabrera is Derek Jeter, or Koby Clemens is Mark Teixeira, or every 19 year old in the minor leagues is Robin Yount (or perhaps Ed Kranepool).
I mean, if a guy walks like a Stubby Clapp and talks like a Stubby Clapp, is it really helpful to have someone tell me the dude's comp is Robin Ventura?
As a keeper-league drafter, I want to know when I'm in danger of getting the Clapp.
New$$ = Original$$ x (DollarsRemaining/ValueRemaining)?
I set centric inflation to '0' and inflation 'YES', however, of the remaining players, Closers keep creeping up the dollar value rungs once all my keepers have been added. Originally only one closer was worth more than a certain player, however now 4 closers are. With centric inflation set to '0', shouldn't the ranking of all remaining players in the auction stay constant?
If you still have concerns, please contact Customer Service and provide a description with the link that has the cid at the end, like cid=12345. You can get this by saving the league settings, which will allow us to pull it up and see exactly what you see, so we can get to the heart of the issue. Thanks!
And, there seem to be many player comments that are like the one for Ubaldo Jimenez - where the gist of the comments is positive ("Tribe will be glad they locked him up with affordable options!") but the 2012 projections have him regressing from what was a down year in 2011 (yet the author calls him an "ace" with a projection of 10-8, 3.86, WAR 1.5 . . .and comps of Jared Burton, Aaron Heilmann, and Ray Narleski. Jared Burton??? He pitched 164 innings in his entire career . . . Ubaldo is at 916 and counting.
Last part first, we don't mind occasionally deviating from the projection in the player comment. The writer might have a different expectation and that's fine with us. Also, player comments written about the projected lines are boring.
A projected player's PECOTA comparables are the players that are the most similar to him in power on contact, contact rate, walks and strikeouts, height, weight, handedness, and position (for hitters) or start/relief split (for pitchers). This can result in some pretty dissimilar-quality players. We are planning on offering some more detail about that part of the process.
Thanks in advance for the info.
You can always get to the latest Fantasy updates in the right sidebar of
http://www.baseballprospectus.com/fantasy/
the spreadsheet was updated on Feb 29, and here's the link to download it directly
http://www.baseballprospectus.com/download.php?resource=PECOTA2012
Feb 14 i asked about SP Yu Darvish TEX
still waiting on him and "other japanese and cuban" projections
(hoping)