Welcome to Our Forums. Once you've registered and logged in, you're primed to talk football, among other topics, with the sharpest and most experienced fantasy players on the internet.
Sorry if this was covered before, but I've never viewed projections as an exercise in accuracy. To me, it's always been an exercise in becoming familiar with the situations and personnel of 32 teams.
![[icon]](/data/avatars/m/5/5167.jpg?1660313464){kind=avatar}
Regarding the division of labor point, I don't think that would work very well. I can't really do Peyton Manning projections without also doing Demaryius Thomas, Eric Decker, and Willis McGahee projections. And vice versa. So I don't know that it would be efficient for one person to do QBs and another person to do WRs.Even dividing by teams I wouldn't be completely comfortable with. What if I have the AFC and Jason has the NFC, and I tend to underestimate overall production while he overestimates it? League-wide, the numbers will look right — but all the NFC guys will be ranked a bit too high and all the AFC guys will be ranked a bit too low.If I'm doing the whole NFL and I'm slightly underestimating overall production, I'll catch my error a lot more quickly than if I'm doing only a subsection of it.I think having each projector do the entire NFL works pretty well. We might be able to add to the total number of projectors in future years — I'm not sure. But if so, I'd still recommend that each person do the entire league instead of dividing things up.
His point was to do the projections assuming 16 games, and let people filter those based on their own feelings on how they evaluate injury prone players, which makes sense to me.As it is, people are coming across a guy like Darren McFadden and deciding "nah, once I factor in that he could miss a few games he's not worth it here". But FBG is already factoring that in, so McFadden is getting dinged twice for the injury risk.People's opinions on "injury prone" vary so much that I think the OP's point makes sense.
I'd like to see you guys do it divided up by team. There are ways to normalize to make sure league-wide trends are captured One way would be to have overlapping teams - A does AFCW and NFCN. B does NFCN and AFCE. C does AFCE and NFCS. D does NFCS and AFCC and NFCC, etc. You, you've got multiple people doing multiple teams, but the same two people are not doing more than a few teams. If you have enough people doing this with limited repetition, it will all smooth out. Ask Drinen about setting up a proper DOE for this, there are rigorous mathematical models for how to set this up.IMO, it's impossible for anyone to be an expert on 32 teams. There is too much detail. I'd rather see folks focused on 4 - 8 teams, in excruciating detail.
it's not that hard if you have a standard format for reporting projections and a method to normalize for historic norms and/or league-wide trends.one thing I've done in the past is put limits on projections - player X's ypc cannot exceed his career best YPC + some small amount. Team Y's plays cannot exceed some maximum number, etc. It's pretty easy to look at historic trends and apply bounds.
+1 on linking articles by author
butt sniffer
butt snifferNice tone to this post.... but Dodds HAS been the most accurate Fantasy Football Projector in the industry over the last 2 years....2011: #5 overall (out of 84)
2010: #2 overall (out of 34)
Independent auditor Fantasypros.com actually had this to say about Dodds:
"Dodds, meanwhile, has now finished in the top 5 in four out of the five accuracy competitions we’ve held. That’s a claim no other expert can make."
OOPS, FACTS.
butt snifferI'm not knockin' Dodds. He provides a great service... many ways they/he can improve the site, butNice tone to this post.... but Dodds HAS been the most accurate Fantasy Football Projector in the industry over the last 2 years....
2010: #2 overall (out of 34)
Independent auditor Fantasypros.com actually had this to say about Dodds:
"Dodds, meanwhile, has now finished in the top 5 in four out of the five accuracy competitions we’ve held. That’s a claim no other expert can make."
OOPS, FACTS.
butt sniffer
Based on Dodds historical success against the industry, I want him doing all 32 teams. I'll look at the other's for a second opinion on a player, but I want to know what Dodds is thining on all the players.
oh...and...butt sniffer
This makes sense for some players...but NOT all players.Say player A is KNOWN in the preseason to be out for the first 3 or 4 games....thus player B is KNOWN to be more prevalent in the first few weeks...more starts. PPG for these two players is than, when based on a 16 game projection....USELESS.
Lol
meh...waste of time1) backups and huys you don't bother to project may get some numbers2) the goal is to get a baseline on a players value, not to predict the exact stats of every nfl player
bAss snifferBass N Brew doing some good schtick in here.
The thing is this is how you make sure you come up with reasonable number for situations like Lloyd/Hernadez/Gronk/Welker. Woodhead/Fells/Branch/Edelman/Others account for 600-2 which is significant numers.
That is your point of view - you read the articles you want to and everyone else does the same. FBG is trying to have articles of interest to many different subscribers/groups and like others I don't want this reduced.
It's true that it's impossible for most people, including me, to be an expert on all 32 teams.I'm going to do projections for all 32 teams anyway, though, because that's how I power my rankings.So if we divvied the teams up, let's say I'd take the AFC West and the other divisions would go to other staffers. I'd still do projections for all 32 teams, but I guess the idea is that I'd pay particularly close attention to the AFC West, and only my projections for that division would be posted? And other staffers would do the same thing with other divisions (for example)?Is that better than publishing everybody's projections for every team? Maybe it is. I don't know.
Soulfy3, how many times have you been banned using other screen names?lol.
Yes that's My Take as well.Don't have projections try an factor injuries, give the projections based on a complete year.Then we as Fantasy Football team owners and drafters can factor what we think may happen andadjust the guy's accordingly, otherwise not knowing this was already being done we may knocka guy even lower then he deserves, and predicting injuries is tougher to get right then stats.-----------Known injuries are different, yes those should be factored in, backup guys that play more ifor when a guy gets hurt are little more tricky, I base those guys on what I feel has a betterchance of happening with them so would rate them accordingly..
as many times as you've secretly begged to be on staff.plentyps: Claratin is good for all that sniffing you're doing.Soulfy3, how many times have you been banned using other screen names?lol.
No, I don't read all the articles. I choose to read the ones that interest me. And I'm sure you choose the ones that interest you. Probably not the same ones depending on the league size, scoring system, rules, etc. So let's say FBGs cuts out the ones I read and just keeps the ones you read. I will no longer subscribe to FBGs. Cutting the number of articles also cuts the number of subscribers.
Sure, but suppose the bottom 10% of the articles get only 1-2% of the page views on the site. They could be cut with virtually no impact.
Actually Fanatic doesn't have time to play ff much less be on staff with his rapidly growing bbq media ventures.as many times as you've secretly begged to be on staff.plentyps: Claratin is good for all that sniffing you're doing.Soulfy3, how many times have you been banned using other screen names?lol.
If you can find me in very many of the player spotlights with projections, or anywhere with a print of projections I'll listen. Good job of digging thru the archives there.
no, I'd want multiple people doing multiple teams, and then some fancy math to level-set the data.Here's an example of how I think it could work - 8 people, projecting 12 teams each. Now, each team has 3 independent projections, which can all be averaged. There are statistical tools that can be used to see if A generally over-projects, etc., and then corrected for. for every player on every team, you now have an average projection generated by someone with a laser-like focus. The projection can be up-to-date because you, as a projector, have a smaller range of teams you are responsible for tracking. But one other piece of data is you can calculate a standard deviation for each projected parameter and use that as a measure of uncertainty.The fancy math/statistics is in the ANOVA or D.O.E family. I'm too far removed from my grad school courses on statistics, but I think there is something interesting there.Pragmatically, it may be difficult for FBG to move towards a model like this. I'm not sure a general user would find additional value here, and in general, would a better quality projection really be more helpful? I'm not so sure. IMO what might be more useful is if you guys gave us a feel as to which teams you feel better about - maybe a return of the "follows closely" bit. I could use that to take the Chargers for Maurile, the Jets for Chase, the Bills for Aaron, etc.I've been thinking about this stuff for years. Several years ago I was trying to do projections myself and ran into trouble with a few teams - Seattle, I think, was one of the key teams I had a hard time wrapping my head around. I am completely indifferent to Seattle. They rarely play the Broncos or any of my local teams. They are almost never on TV in the south. Geographically, they are the furthest team from me. In general, I don't follow NFC very closely. I couldn't care one bit about the Seahawks. How am I supposed to come up with a reasonable breakdown for the Seahawk WR landscape?'Maurile Tremblay said:
two posts in the last week or so you are talking about "your projections".... I didn't exactly have to "Dig" very far there fella.
I agree that it's sort of a waste of time if you project backups to find out THEIR stats. Meaning, nobody really cares how many receptions and yards Devin Aromashodu is going to have. But it's important to get a sense of how many catches he'll have, because those are catches that Percy Harvin and Kyle Rudolph won't have.Speaking only for myself, I'll project a team's total number of offensive plays, break that down by how many rush/pass attempts I expect in the coming year (based in part on changes, individual player improvement or decline, offensive and defensive improvements, etc), and then divvy up passing targets/receptions accordingly. After accounting for the bottom of the barrel guys, I then have a solid number to work with for the top guys. Otherwise if I just project the draftable players like Harvin, Rudolph, etc., they are getting too large a piece of the pie.
Great stuff in your letter to Joe. Well done.On this point- you're right that they can't predict health, but they can project tendencies or likelihood. For example nobody would consider it equally likely that Vick and Eli Manning will play 16 games in 2012. One thing I'd like to see some sort of projection that includes replacement-level performance for guys who are more injury-prone.For example, let's say you project Vick to produce 300 fantasy points in your league in 13 games and E. Manning to produce 340 points in 16 games (just as an example based on 2011 numbers). Shouldn't the projections somehow be able to compare the two by incorporating replacement-level performance at QB for the other three games? Total points overvalues Manning because it ignores the replacement points, and points per game overvalues Vick because it assumes that you're replacing Vick with a player of equal caliber.Is there a way to do this? Or maybe the FBG tools already do it and I just don't know about it? If this sort of calculation were somehow incorporated into the projections I think it would help correct the problem ComfortablyNumb identifies, where the projections that predict missed time for injuries skew the data to the point that it's less useful.
