Pitchers and catchers have yet to report to spring training, and the Seahawks are still celebrating, but the PECOTA projections are here, earlier than usual.
(We interrupt this announcement to bring you the direct link to this year's PECOTA weighted-means spreadsheet.)
In order to access the PECOTA forecasts, you need to be a Premium (or soon-to-be-available Super-Premium) subscriber to Baseball Prospectus. Monthly subscribers will have access to certain PECOTA features but will not have access to downloads like the PECOTA spreadsheets. The best value we offer is a yearly subscription, which not only gives you access to the full PECOTA product offering, but also unrestricted access to our website's extensive baseball coverage.
In the 11 years since the first "Introducing PECOTA" article in Baseball Prospectus 2003, PECOTA has proven to be one of the best predictive systems in the industry. The current edition was the brainchild of Colin Wyers, who was hired by the Houston Astros as this winter's work took place. Colin's core concept of having the batter-pitcher matchup quantification center around batter TAv and pitcher Fair RA remained untouched, and the annual weights that Colin explained in detail back in 2012 remain intact. Without his work, PECOTA wouldn't have been possible this season, and Baseball Prospectus' loss is the Astros' gain.
Our playing time estimates are based on the input of the Baseball Prospectus staff and coordinated by Tim Collins. Keep in mind that these aren't etched in stone: pitchers and catchers still haven’t reported to camp, and some key free agents still haven’t signed. We will be updating these playing time estimates frequently as teams make moves and we see how rosters shape up in advance of Opening Day.
One other note: players who aren't projected to play in 2014 have hypothetical projections that can be seen atop the player cards, with fictional playing time numbers. This fiction is based on actual playing time from past seasons (with MLB playing time much more highly weighted) or a minimum of 250 PA for hitters or about 33 IP or more for pitchers, depending on role. As an example, here's Matt Harvey, who had 59 and 178 MLB IP the past two seasons, respectively (and a lot more MiLB IP in 2012):
Of course, we don't think Harvey will actually pitch nearly 150 innings in 2014. We know he's hurt, which is why the Mets Depth Chart doesn't include him. This is a "what if" scenario that should give an idea of what he could have accomplished (in even more innings) if he'd been healthy.
Right now, we’re rolling out the following:
Click the "Feature Focus" links below for an explanation/walkthrough of each BP product
- The PECOTA weighted means spreadsheet (available under your digital downloads) (Feature Focus)
- The PECOTA summaries atop the player cards (Feature Focus)
- The Depth Charts (Feature Focus)
- Note that Depth Charts include projected team records and standings.
- Visual Depth Charts (Feature Focus)
- The Player Forecast Manager
- The Scoresheet Draft Aid (Feature Focus)
- And the PECOTA portions of Team Tracker (Feature Focus)
Between now and the start of the season, we will roll out additional parts of the PECOTA product:
- The preseason version of the Playoff Odds Report
- Additional PECOTA player card content: PECOTA percentiles
- Additional PECOTA player card content: 10-year forecasts (not available in 2013)
- Additional PECOTA player card content: UPSIDE ratings (not available in recent years)
- PECOTA projections for the Team Audit pages
Downloading PECOTA:
Quick notes for first-time subscribers: Once you subscribe, the PECOTA spreadsheet will be available to you at any time as a Digital Download. To access your Digital Downloads, click on "manage your profile":
Customer service FAQ on downloading the spreadsheet:
- The downloadable PECOTA Weighted Means spreadsheet has three tabs at the bottom (labeled "Release Notes", "Hitters", "Pitchers"). If you're not able to view the tabs in your spreadsheet, please see help here: http://office.microsoft.com/en-us/excel-help/where-are-my-worksheet-tabs-HA010278741.aspx
- To download the PECOTA Weighted Means spreadsheet, the BP site will require you to enter your CASE-sensitive username and password. If you need password assistance, please contact Customer Service, https://legacy.baseballprospectus.com/contact.php (changing the response our spam question to "no").
- If you change your password, it may take up to 20 minutes for the PECOTA server to recognize your new password and allow the download of the PECOTA Weighted Means spreadsheet.
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
I had not seen an update on him yet, which is OK since this is new info, but the problem is I can't find evidence of *any* updates to FPM since it's release. A running list of updates would be very helpful.
One question: in the spreadsheet, why do the hitters have VORP and WARP, but the pitchers only have WARP? On the depth charts, you show pitchers' VORP, but where can I find that projection for pitchers not shown there?
We're calculating things like mean average error (weighted and unweighted), RMSE, correlation on common projections to compare the systems.
We have some other ideas already, but if folks have things they'd like to see us measure/compare we're happy to hear your suggestions.
http://www.fantasypros.com/about/faq/accuracy-methodology/
http://razzball.com/category/testing-projections-player-raters/
Take an example of rolling 30 fair dice: If a predictive system was to figure out the expected value of each die, it would be an impossible 3.5. So, none could be said to have an expected result of "6". Yet, of those 30 dice, one could expect 5 to come up with a result of "6".
For the projected standings here, what you're seeing is the result of the expected wins based on the expected runs scored and allowed. And we're fairly confident that if the 2014 season was replayed a million times, given the playing time estimates we use here, that the win totals of those million times would average out to about what we have here. But each one of those million times is very likely to have individual team results that are much higher or much lower than the expected values.
Alternatively, re-running and publishing the team results based on 40%, weighted mean, and 90% player projections, as an approximation of expected results?
Doing something based on percentiles is an interesting suggestion, and we'll consider some form of adding variance into the projections-based standings, as you are proposing here. I'm sorry that it's probably not going to happen this offseason, but it definitely is a suggestion we'll consider seriously for some future product offering.
Whew. Deep breaths. It's gonna be okay.
jdm
10. Adam Eaton - 4.2
9. Elvis Andrus - 4.4
T-7. Desmond Jennings and Jacoby Ellsbury - 4.5
T-5. Brett Gardner and Ian Kinsler - 4.7
4. Jonathan Villar - 5
3. Ben Revere - 5.9
2. Michael Bourn - 6.1
1. Billy Hamilton - 10.8
*barring an Effectively Wild listener email-esque change to the rules.
Rest assured that as the season approaches, and playing time becomes more clear, situations like that will be ironed out.
For past-season stats, you can sort by using the Team Tracker capabilities on the sortable reports, though that doesn't help for Forecasts, obviously.
Our league uses Wins - Losses as part of its stats ensemble, and because relievers all are better than average in this category, they float to the top of the projected pitcher performances.
Using the Depth Chart page, Arizona's starters are projected to win 51 games and lose 66. Their relievers Win 28 games and lose 19.
Why is PECOTA conservative for starters but thinks all relievers are net winners?
Is it user error? If not, what product should I be relying on?
Sorry for the inconvenience.
Thanks