clock menu more-arrow no yes mobile

Filed under:

Community Projections Best Practices

So with almost 75% of the teams now having leaders for our Community Projections Project, I thought it would be smart to let everyone share their experiences so far.  This thread is for sharing ideas, solving problems, and increasing motivation.

Feel free to link to anything helpful you or others have written.  Feel free to suggest third-party tools that will make things easier.  Feel free to share your frustrations.  And feel free to help others solve their frustrations.

I'll start with some things I've noticed going on:

  • jihad over at BrewCrewBall wrote a very good introduction.  And the first two comments in Doghouse's Nationals' intro post are a great question and answer combo.
  • Some blogs (DRaysBay, for example) have successfully gone player by player asking for OBP/SLG/PA projections.  Some have grouped positions together (infielders, starters, etc.)  Some have tried to post one all-inclusive thread and asked for comments on everything.  That last one seems to have broken down every time's it's tried, but I've also heard bloggers say reader interest can break down if you go player by player.
  • Some folks asked readers to provide projections from scratch (maybe posting the last few years' data as a jumping off point) and some started with CHONE projections and asked readers to argue for bumping specific players up or down.
  • devil_fingers brought up percentiles in his (not short) Royals post.  He entered CHONE's 50th percentile projection into the spreadsheet, and asked readers to think about who would under-perform that level in addition to players they thought would over-perform it.  Not a bad way to tell people to mix the optimism with some realism.  Update: Slyde provided 30th through 70th percentile OBP/SLG projections to help his readers.
  • You've probably pick up on it, but my favorite algorithmic projection system is CHONE.  PECOTA is also excellent, but isn't free and isn't out yet for 2009.  ZiPS and THT aren't bad, either.  I tend to steer clear of fantasy baseball proejctions, Bill James' system, and Marcel.  The last two aren't bad, but BJ is overly optimistic and advanced systems start with Marcel, but evolve into something better (at a minimum, CHONE is Marcel but with minor league data.)
  • The algorithmic projection systems tend to do much better with rate stats than playing time.  They don't know about injuries, managerial tendencies, and roster construction, but fans do.  Predicting playing time should be the main strength of our community projections.
  • Linking to the EditGrid workbook (no, I don't know how to link to a specific sheet) is obviously one way to share what you've put into it, but getting readers to follow links can be tough.  As an alternative, you can take screen shots and post jpgs of relevant data.  Or you can imbed a "portal" of selected cells in a post (select the range of cells, go to "Publish", select "Cell/range...", and copy the code.)
  • Colin Wyers created a cool form for readers to input their projections. To copy it for your own team, go to "More Actions" and then "Copy Application".  Slyde from Red Reporter has imbedded the form into his projection post, which look awesome.  (This won't work with FanPosts, though -- maybe talk to the head blogger and ask for front page privileges for this project?)  Warning: you'll to do some Excel-type work after collecting all the projections, although there will be help if you need it.
  • If you're looking for tips on actually using the spreadsheet, you can check out the original spreadsheet post (you can download your own copy, too), email me, or ask in the comments.  There are also 140+ comments in the original Community Projections post.
  • I'm tagging all these posts with "2009 community projections".  Feel free to do the same to help readers click through to others' discussions/projections.

Comment away!

Eight teams still need volunteer leaders: PIT, SDN, NYA, CHA, CLE, DET, LAA, and SEA.