IBC

Source?

Pat @ DM.

So what do we do? Zips? Pecota? Some combination? To me the key is finding something that will do LH/RH splits. Meantime, spread it on the board over there and start bitching, those of us who never bought the software should basically say that we will not buy it without the disk, and those of us who did buy it should say they won't upgrade without one. This got one made in time for the season last year, if we start now there may still be time.

ZiPS will have splits this year.

Our other option is trying to create our own database. Either with PECOTA or with some sort of weighted projections using Pecota, Zips, ML and MILB stats and such. Which would be a huge undertaking.

There isn't a projection system out there that does L/R splits. <edit> i do have a couple messages out to some people at BP about using PECOTA as a projection system though.
Frankly though, I think DMB is making a huge mistake and this is a project that is SCREAMING to be open-sourced.

As I said, ZiPs will have splits this year when he releases the disc.

Pirates wrote:As I said, ZiPs will have splits this year when he releases the disc.

I saw that after my post, tried to insert an edit and managed to do something weird to it. not a fan of Zips personally though, I'll take splitless PECOTA over Zips with splits.

The other alternative would be to switch to using Season disks. I'm not a huge fan of the concept, but it's something that should probably at least be discussed.

I disagree. Splits are a MAJOR factor. Even if the projections aren't as sound, just having the splits is necessary I think.

RedSox wrote:The other alternative would be to switch to using Season disks. I'm not a huge fan of the concept, but it's something that should probably at least be discussed.

I don't think we could do that. Teams are built based on projection. Even if the projections aren't as good, it's not like DMB has been great the past few years. Season Disc is something I would be vehemently opposed to.

I definately think PECOTA projection are typically better than the rest, however splits are a big part of it and without them I'm not sure we could/should go that route. ZIPS is definately lower quality but if they have the splits, I imagine it would be the route I'd have to recommend.

Voiced my two cents over at the DMB forum. Can't hurt to send them the message.

Splits with a poor projection system v. no Splits with a good one. I'll take the splitless ones.

Forget the forum, email DMB!

RedSox wrote:Forget the forum, email DMB!

I'll do that when I get a couple more free minutes.

Shawn Lape, is what Walko and those guys do incredibly intensive?

Shawn's league uses a formula that, if it's at all realistic time-wise, seems superior to just using either ZiPS or PECOTA. It's a weighted combination of major league, minor league, Japanese league stats and PECOTA projections. I just don't know how they incorporate splits. They do use the other program though, OOTP. Probably too complicated for this year anyway.

My other concern as we've talked about this for the last few weeks is defensive ratings. ZiPS really sucks and I think all that he gives are overall rankings, which in DMB would give everyone the same error rating. I think that a few of us could generate a pretty solid defensive ratings set for major leaguers based on some of the more advanced stats, but minor leaguers seem like a bit of a problem.

Tigers wrote:My other concern as we've talked about this for the last few weeks is defensive ratings. ZiPS really sucks and I think all that he gives are overall rankings, which in DMB would give everyone the same error rating. I think that a few of us could generate a pretty solid defensive ratings set for major leaguers based on some of the more advanced stats, but minor leaguers seem like a bit of a problem.

If this is possible, we could use DM season disk defensive ratings which are the exact ones usually forwarded to the projection disk. With minor leaguers we could agree to basic range and error ratings that match similar players that had a cup of coffee in the show (made the season disk).

I think it's worth reaching out to other projection leagues as well, and working together to split some of this work up (as well as encouraging people not to buy v10 without a projection disc). As someone can't remember if it was here or on the DMB boards said this is a project that is begging to be open-sourced.

Athletics wrote:I think it's worth reaching out to other projection leagues as well, and working together to split some of this work up (as well as encouraging people not to buy v10 without a projection disc). As someone can't remember if it was here or on the DMB boards said this is a project that is begging to be open-sourced.

I agree with all of that, but I think it will probably wind up applying more to 2010. Unless something surprising happens, we're going to wind up choosing from a pretty finite list of options this year. It's almost March.

I understand that, but I'm thinking the thing that's going to take the longest (assuming we use a fairly simple combination of already existing stats and projection systems) is going to be data entry, which can be done fairly quickly if there are a lot of people involved.

I think that's something that's better served undergoing a long and serious discussion with a few different methods being looked at and not something we need to rush to in order to make our own projections, which may or may not even be better than what ZiPS has to offer. That's a route I'm willing to go down, but just not for 2009. That would require a lot of discussion and a LOT of time and we'd be very well served starting that around November or December rather than February 28.

I would argue we might be best served starting that in April and May rather than November and December. That was we prevent biased opinions based on end of season statistics.

So what's next? Do we list all the options and vote? This is a HUGE change for the league and I don't think its one that should be made quickly...but time is a factor here. We're a month away from the season. If we use ZiPS because it has splits, and import defensive ratings from the season disc...does that mean someone has to go through and do that manually? I'm not completely opposed to switching to the season disc. At least all of us would be equally screwed. Not saying it's my favorite option or anything, I'm just not sure we should throw it off the table so quickly.

I'm also not in favor and trying to force DMB to make a projection disc. I think we can all agree that the quality has really been lacking these last two seasons and trying to force them to throw together a projection disc in a month is just asking for even shittier product. I think its time to write off the DMB projection disc and look for a new method going forward.

The 2008 ZIPS projection disk had error ratings for fielders. There's no reason not think they won't have it for 2009.

Also, a major problem I have with the season disc is it's basically everything we're not. It's fun for goofy winter leagues or whatever or to pass the time, but, we have all built our rosters based on projections. DMB projections have slipped in quality the last few years, so it's not like it's the end of the world using ZiPS.

Also, since rosters have been built with projections in mind, teams that are HIGH in rebuilding mode (TB etc) have a large number of prospects. on a season disc, he'd be totally fucked and would have to restructure his roster in order to even field a team to lose 120 games because he'd be getting the scrap heap to fill out the roster. It would suck for those rebuilding teams.