***Official 2009 projections*** by the Shark Pool, For the Shark Pool (1 Viewer)

[moleculo]

I’ve done my own projections for several years now, to varying success. I’ve found that my projections really break down on particular teams – I simply don’t have the interest or the time to really understand what is going on for all 32 teams. It’s simply too much to handle. However, I can handle one team. I know the Broncos very well – I subscribe to several blogs, I read both the local news papers on a fairly regular basis, and I monitor a couple of forums. I feel that I can project for the Broncos much more accurately than, say, the Seahawks.

I would like to leverage the expertise of the Shark Pool towards making better projections. Almost everybody here is an expert on at least one team. I would like to set up a system where at least 32 Shark Pool contributors make projections for their team only, and then compile those projections to come up with composite rankings.

For this to happen, we need some ground rules. To facilitate compilation, all rankings will need to be done using a standard template and format. All projections will need to include the same parameters (rushing yards, TD, etc), we will need a common set of assumptions (project injuries?). We will need committed contributors – once you start, you’ve gotta finish. All projections have to be reasonable and historically consistent – projecting LT2 to get 32 TDs next year isn’t going to help anyone. Ideally, all projections would be done with the same method and formulas, but ultimately this isn’t really that important. I’m sure there are other facets that merit consideration; it will all come out as this gets developed.

I envision a system where all pre-registered contributors send projections (e-mail? ftp?) to the compiler on a regular basis (timing and schedule TBD). The compiler will then post rankings in this thread, and e-mail out complete projections to all compilers. The benefit to the participants is that they will have complete projections from which they can derive custom rankings – in addition to leveraging expert knowledge from the pool. I think that, if executed nicely, this could be pretty powerful (and free).

 

[moleculo]

I have finished the first draft of a spreadsheet that can be used to generate balanced team projections and format nicely for export. Right now, I'm looking for participants and beta users to test it out and let me know what needs fixed. If anyone wants a copy, you can download at the following:

link

or, if you e-mail me @ fbgmoleculo@gmail.com, I'll send you a version with your team loaded up.

==========================

The spreadsheet operates under the following assumption: production = opportunity * talent.

Opportunity is defined as attempts - carries for a RB, catches for WR, pass attempts for a QB. To define the number of attempts, we look at the total number of plays, the run/pass ratio, and the attempt distribution. For example, the Vikings ran 970 plays lat year. Run/pass ratio was 53.4% runs. Of the rushing attempts, APeterson had 71% of the carries - that all comes to 366 attempts.

Talent is what one does with the opportunity. For RB's, its YPA (4.83 for APeterson). Of course, there is a team YPA as well, which is really quite consistent year to year. So, I have some stuff set up where you can adjust the team YPA and all players individual YPA's will balance out - consider a new OL blocking scheme is installed - you might project a higher team YPA than last year, which means you can adjust this and all of the backs YPA will increase accordingly.

Additionally this spreadsheet is balanced. Team passing yards = team receiving yards, team completions = team receptions, etc. I have some functionality coded in to keep everything consistent.

eta file location