keyboard_arrow_uptop

Again, we won't bury the lead. Many of you have been asking about the Player Forecast Manager. Right now we're planning on having that available a week after the initial release of the PECOTA spreadsheets—not next Monday but the Monday following. The PFM is one of our most popular features, and last year the demand nearly swamped our server. We want to be absolutely certain that doesn't happen again, so we've improved the plumbing behind the PFM and we're moving it to its own separate web server. That should ensure snappier performance for PFM users and keep the deluge of PFM traffic from interrupting the rest of the site.

One of the reasons the PFM is so popular is because of its flexibility, but that also makes it intimidating to use. We want to fix that, without taking away any of the flexibility that users have come to expect. So fantasy guru Marc Normandin is hard at work preparing updated instructions for PFM users, which we'll be building into the PFM interface. We think this will go a long way towards alleviating the problems some users have reported in the past.
 
And we know that for the avid fantasy player, the PFM is just one of many tools you'll find yourself using. So in order to make it easier to link up PFM output (or any source for PECOTA forecasts, for that matter) with other tools, we'll be providing uniform ID codes with all of our projection materials.
 
(And if there's any way we can get it out to you before then, we will. We're just trying to give you the best idea we can of what we're doing and when, without having to worry about missed deadlines or broken promises.)
 
But let's talk about something else for a minute now. Recently, I've been reading "The Bullpen Gospels," Dirk Hayhurst's account of his time in the minor leagues. I know, I'm behind the times—if you're as behind as I am, I highly advise that you go out and read it now.
 
Well, Hayhurst has been something of a free agent sensation in Tampa Bay, after having landed there from Toronto. And so I was curious to see what PECOTA had to say about him:
 
DIRK HAYHURST Born: 3/24/1981   Age: 30 Bats: L   Throws: R   Height: 6' 3"   Weight: 200  
Breakout: 0%   Improve: 2%   Collapse: 3%   Attrition: 5%   MLB: 5%   Comparables: Kazuhito Tadano, Chris Smith, Troy Cate (84)
YEAR TEAM LVL AGE W L SV G GS IP H HR BB SO BB9 SO9 GB% BABIP WHIP ERA FRA WARP
2011 TBA MLB 30 4 3 0 51 6 99 109 12 36 74 3.3 6.8 44% .321 1.49 4.89 5.31 -0.4
 
Frankly, that's not what I was hoping to see.
 
It was pointed out in the comments yesterday that I was anthromorphizing PECOTA, treating it as though it has a mind of its own. And that's true. Now, on one hand, that's preposterous: PECOTA is nothing more or less than a series of instructions I give to a computer. (And by the way, I checked and there weren't any subroutines in there specifically designed to be pessimistic about the White Sox.) So it's interesting to watch PECOTA follow my directions to the letter, and end up with a conclusion I'm unhappy with.
 
Another thing I mentioned is that I feel a sort of fatherly affection for PECOTA. Sorry, can't help it at this point. But like any parent and their child (after that child has grown up, and is starting to stand on their own), we don't always see eye to eye. I don't bring this up say that I have a methodological reason to disagree, or some intuiton that PECOTA is missing something here. What I want to point out is something I think that gets missed—that a forecast is what we expect to happen, or at least what is likely to happen. There's little point in forecasting what we want to happen, for its own sake. For one thing, I like to think I have too much pride for that. For another thing, what good would it do me? I'm not quite so hubristic to think I can change the future by sheer force of will; my models can only hope to mirror reality, not change it.
 
So if PECOTA seems particularly down on a favored team or player, isn't not because it hates you, and it's not because I hate you. Sometimes, you're gonna hope PECOTA is wrong. Trust me, sometimes I do too.
 
(Another thing you'll note, if you're wondering about the Breakout/Improve/Collapse/Attrition numbers:  those are designed to describe the MLB performance of players. This year, we're encouraging PECOTA to rely more heavily on minor league comps for minor league players. That means some players have fewer comps playing in the majors the following season than they would have before. For players with few or no comps making it to the majors, PECOTA shows little chance of their doing anything in the majors.)
You need to be logged in to comment. Login or Subscribe
markpadden
2/04
Can you post, or point me to, a summary of what has been changed in the PECOTA algorithm for this season?
rawagman
2/04
I know I used to know this, but the comps listed are age comps, right? In other words, closest comparables in their age-30 season?
hessshaun
2/04
Maybe I didn't have enough coffee yet today..... "we'll be providing uniform ID codes with all of our projection materials" What does that mean?
tbsmkdn
2/04
ID codes are unique alphanumeric strings that are attached to each player. Let's say you had a pitch f/x database and a PECOTA database and you wanted to combine the data, or otherwise cross-reference it in some way. You would need some variable in common for each player you wanted to link up. That's what ID codes allow you to do.
hessshaun
2/04
Sweet and thanks!
o2bnited
2/04
We want you to be able to take the information that you download from projections or the PFM and seamlessly integrate it into other materials, so we are going to give you the unique ID for each player in the output. In the past you would have had to build a system for name matching or something similar. Hope that helps.
o2bnited
2/04
Ah, yeah, what Tommy said. Sorry, should have reloaded first. :-)
rosborne
2/04
Is this ID unique to BP or will it be the same as some other systems (i.e. BDB, etc.)? Thanks.
cwyers
2/04
We'll be using what's typically called the MLBAM ID - the same ID used at MLB player pages, provided in the Gameday XML files, etc. (ID schemes like BDB/Retro/etc. are simply not comprehensive enough to cover all the players PECOTA projects, as they only cover players who have played in the majors.) Others have gone to the effort to link MLBAM IDs to other common ID schemes, so they're a very good "lingua franca" for us to use so that you can link our data up with everyone else's.
rosborne
2/04
Thanks Colin. Do you know if BP statistics will also use this ID system?
deepblue64
2/19
That's great news - thanks!
BStephen
2/04
Ok Newbie Here! I get in 2 weeks the pecota Numbers will be in The Forecast manager. BUT where will the be next week?
brownsugar
2/04
In a downloadable spreadsheet with a weighted mean forecast for each player.
Michael
2/04
If you want to see what the format will look like approximately, you can see last season's spreadsheet by Click on Fantasy in the menu at the top of this page Find a box with a blue header in the right column Click on PECOTA Weighted Means Spreadsheet
Guancous
2/04
Thank you very much. Excel's VLOOKUP function will make organizing all this data much easier.
BStephen
2/04
Where would VLOOKUP come in ?
mgolovcsenko
2/04
No updates to the PFM engine and algorithms itself then? (Not that any are needed)