|
In order to help our members better use the starting information that The Huddle provides, the intent of this page is to discuss perceived differences between The Start Bench List and Predictions and Projections . The players below were identified by multiple members. The two reports are written by two different people so differences are always possible, but at least this page should help you make up your mind about these players on your roster.
Greg Jennings (at CLE) - Projection: 60 yds, SBL Rating: S2
Tuvey: As the initial writeup indicates, there should be plenty to go around for all the Green Bay receivers against Cleveland. Driver has been the more productive member of the bunch, but I haven't soured on Jennings to the point that I'd knock him down to an S3 with such a favorable matchup. And I don't think Driver has been so dominant to warrant an S1breakout from this group.
Dorey: I gave Driver 100 yards and a score because I do like him much more this week than Jennings. Driver has been far more consistent than Jennings and been banging around 90 or 100 yards in three of the last four games. He has two scores while no other Packers wideout has more than one so car. Jennings has just ran very hot and cold with two efforts over 100 yards (CHI and @STL) and then only 95 yards from the other three games combined including getting blanked in week two. I am just cooler on Jennings who really has relied on just long passes for his two good efforts. Driver is getting more catches and is more reliable than Jennings right now.
Consensus: We both believe Jennings has fantasy value but only marginal according to Dorey and barely moderate by Tuvey. Take Dorey's 60 yards as his likely worst case and consider him with a little upside this week.
This week was a surprise. We had about 25 emails asking about a player (some asked about two or three). And yet of all those emails the only player with two or more people interested was Greg Jennings who literally only had two. So evidently we agree a lot this week but rather than give you just the one name, we thought we could use this just to discuss in a bit more depth what the Start Bench List rankings represent and how the projections are arrived at in Predictions and Projections.
A Brief Start/Bench List Primer by John Tuvey
Don't tell DMD, but I think it's easier to come up with the S1/S2/S3/B rankings than to lock in on an exact projection for each player. It certainly provides for more wiggle room, but of course it also opens things up for interpretation. The code key at the top of each page of the Start/Bench List provides some clarification, but to help you understand where I'm coming from when compiling these rankings, here's a bit more of a peak behind the curtain.
S1: Start 'em Tier One (Stud / Great matchup)—I reserve this for the best plays of the week, the ones I feel fully confident will deliver monster numbers for you. From time to time I take some heat for having so few of these (for example, there were only eight last week), but I want these to be the sure thing plays. And last week, seven of the eight S1s were spot on including the Brady/Moss/Welker carnage and Thomas Jones' record setting performance. The lone whiff was a doozy; thank you, Matt Hasselbeck.
S2: Start 'em Tier Two (Solid matchup)—This is the "good start" category, and unfortunately for you that's a pretty wide umbrella tied to your league's size, scoring system, roster requirements, etc. Essentially what I'm saying with an S2 is that this is a pretty good play; maybe not the monster game an S1 would indicate, but you can be pretty confident with an S2 in your lineup you won't be disappointed. And that touches on the larger theme of the S/BL, which is the risk each start entails. DMD might think, for example, that both Jennings and Driver have a "good chance" of scoring, but if his projections only indicate one TD pass he has to give those points to only one guy—even if he's at 51/49 that it goes to that guy. With an S2 I don't have to pin down that score; an S2 for both Driver and Jennings suggests they both have a "good chance" of scoring in addition to putting up their expected yardage or better.
S3: Start 'em Tier Three (Borderline / Barely)—This category suggests that a player should give you what you expect, but there are risks involved; maybe it's the matchup, maybe it's someone else in the offense horning in on their role, but something is putting their potential production in peril. Often times you'll find "near-stud" players with difficult matchups in this grouping; you wouldn't necessarily sit the guy, but on the other hand you need to dial back your expectations.
U: Upside player (Possible sleeper)—Let's use a couple of this week's U guys as an example. Among running backs, the secondary New England backs (Green-Ellis and Faulk) get Us because the Pats should light up the Bucs and you could easily see BJGE getting enough garbage time work to have fantasy value. Faulk has value in a PPR league, especially given the dwindling competition for carries in the New England backfield; I wouldn't recommend him as a start, but his potential is worth looking at in larger PPR leagues. That's the crux of a U: while not necessarily a starter in most formats, there is something to this guy's matchup and potential production that warrants examining—and in some situations would make for a viable fantasy starter.
X: Unclear situation / Could go either way—I try to limit these because, quite frankly, they don't help you at all. You'll see more Xs on the first release of the S/BL as we wait out injury information; then, on Friday, I try to update as many of the Xs as possible to reflect how much a player practiced, whether they're expected to play, at what percentage of full-go they'll be playing at, and whether what they bring to the table this week is worth a starting designation given the match-up.
B: Bench 'em (Bad Matchup / Too much risk)—Pretty self-evident. If I guy gets a B, I can't see a situation other than the most dire of straits in which I'd want to plug that guy into my lineup.
Maybe that will help clarify where DMD and I are coming from with our lineup recommendations. It doesn't necessarily explain why he hates Steve Slaton so much, but maybe this peek behind the curtain will help you interpret this and future S/BL vs. PP questions.
A Brief Predictions and Projections Primer by David Dorey
Don't tell Tuvey, but I think it's easier to come up with the S1/S2/S3/B rankings than to lock in on an exact projection for each player. While he is throwing codes onto players, I have to come up with hard number expressions of what I feel is the most likely to occur given all the variables that I have to consider. Let me give you a little background on our projections and other sites you may consult.
Once upon a time, there was a major huge sports conglomerate who took a fancy to The Huddle and even investigated acquiring it (more than once actually). Said company flew a couple of executives in to speak with us about potential events which never passed. But as we were discussing what The Huddle is and does, they mentioned that they had tracked the projections of many sites and that ours had been the best. One of the guys asked me - "So what sort of algorithm do you use to come up with your projections?" Whitney just pointed at my head and I picked up my jaw from the floor because it was the first time I realized that many and maybe most sites just calculate projections based on a formula. How easy would that be? I think maybe they just wanted my formula and were disappointed there was none.
All projections here on the site are the product of much research and analysis, reviewing stats and current news and factoring in many variables to come up with those numbers. If you have ever noticed, our projections are not made by a computer nor could they possibly have more yards by receivers than passed by the quarterback or other inconsistencies. From the start of The Huddle, I have lumped in Predictions and Projections together because they belong together. You have to build a game of the parts and what I am trying to do is to make a box score of what is most likely to happen. And come up with a score.
Now I could write an entire book on how detailed projections are made. But in a brief way, what I do is first review the history of the two teams as they played each other. I usually note on the page any previous stats if they have played each other within the last year. I review what the teams have been doing this season in the sort of teams that they have faced and the outcomes. The game score I predict is the product of what the teams have been scoring or allowing while weighing some past games more heavily than others. I'll certainly consider any injuries on the team and the dynamics at play that week. I have a basic idea of the score first and then adjust that as I dig into the individual players and see what I find.
For passing numbers, I will consider what sort of passing totals are most likely based on everything the quarterback has done and what the defenses have been allowing and considering what past opponents have been most like what they will be facing this week and whatever factors come into play for this one game - injuries, weather and the like. Once I am comfortable with the total passing numbers for yards and scores, I then have to determine what the breakdown is most likely. Again - all that has happened with the offense and all that has been allowed by the defense. I do not usually project for more than three or four wideouts even though as we all know there can certainly be more. There could be two tight ends. Three or four running backs. But this is fantasy football so I am going to consider mostly the primary receivers that will have any sense of reliable fantasy value. I break out the yardage and then assign the scores. Those touchdowns are the biggest factors for people wondering why I am ranking so highly one guy over another. The reality is that I may see them almost exactly the same but someone has to get the score.
The rushing stats are the easiest part since that is player specific as opposed to one quarterback distributing passes to at least five to eight players each game. Committee backfields suck because I have to break them down along with rushing scores and potentially receiving scores. I have a white hot hatred for the Patriots backfield for example. I am not fond of game time decisions. I get very into the statistics and have been doing this for 14 years now. I have predicted over 3500 NFL games. I have done over 50,000 player projections. And yet I am still always learning and still love trying to improve my predictions and projections.
Along the years, I have had requests for assigning partial touchdowns (0.5 TD) or going more granular and predicting 72 yards instead of 70 yards. I agreed to let the projections be put onto one page and now as sortable stats using your league's scoring rules. But what I mostly want to do is to consider it first as a game, then break that down into the two teams, and break those down into their notable players and create a predicted box score because it is all inter-related. It blows me away how I think some places do their projections by either pushing a button or projecting by position and not by team and game. I want everyone to read every page but naturally some only look at the sortable projections and wonder why without ever reading the explanations on the game page.
It gets rather complicated to explain because there are so many variables and factors to consider in a game. When I predict a game and project players, I have spent considerable time coming up with what should happen. And of course most are pretty close, some are not and a few probably end up very wrong for various reasons.
Which leads me to end by saying that I am truly an expert at this. Because an expert is not someone who is always right - no one ever is. A expert is just someone who can explain in detail why he was wrong.
And it still kills me when I am.
Start Bench List Ratings:
| S1: Start 'em Tier One (Stud / Great matchup) |
U: Upside player (Possible sleeper) |
| S2: Start 'em Tier Two (Solid matchup) |
X: Unclear situation / Could go either way |
| S3: Start 'em Tier Three (Borderline / Barely) |
B: Bench 'em (Bad Matchup / Too much risk) |
|
Predictions and Projections:
The projections are what is mostly likely to occur given everything that has recently happened added to the current team dynamics. They consider total rushing and passing yardage and scores and then distribute them between the most active starting players based on how the track record of the offense and the opposing defense along with variables like injuries, depth chart changes, venue, weather and so on.
|
|
Facebook 
Twitter 
Mobile
"hiding behind" math???
projections MUST be based on probabilities!!!
rounding to whole TDs is still using probabilities, just less accurate ones...
that .4 TD--while impossible--is more ACCURATE...it means the chances of a score are 40%...if rounded to zero TDs, that player's true projected value is understated...
using partial TDs in the projections would produce truer relative rankings...if two players are 50/50 to score, their true projection would be .5 TDs reflecting the true chance of a score...guessing whole TDs only arbitrarily skews the projections...
most of us use the projections to help decide who to start...we need a projected ranking based on true odds of scoring to keep the relative projected player values in order...
good article guys...nice to see behind the numbers...keep up the good work!
box scores would be viable with fractions DMD [even in perenthesis after the whole number]...if you think only one rushing TD, then a .6 for Maroney, a .2 for Faulk, and a .2 for Green-Ellis works just fine...
Like others, I've been a Huddler for about 10 years. A few times a year I'll do some writing for my league and do my own 'predictions' for the outcomes of the league games. It's all in good fun and I honestly don't know how you guys keep this site going at such a high level. Kudos.
I know TD's are very hard to predict, but take a position, which DMD does. I don't expect incredible accuracy... just another opion based on solid stats, facts, trends and then something that I often rely on myself. It's that gut feel one gets after all the number crunching and "hard data" are done being analyzed, and one is left with a hunch about a particular player. It isn't a guess, I think it's based on all the info one contemplates while doing due diligence when evaluating a player's prospects for the next game. Algorithms don't have hunches. That's why they are only a guidline, another bit of info to base a REAL decision on. Like projecting TD's. That jump makes a projection wrong or right. Projecting 0.4 TD is a safe, and gutless way to project FF. You can't be wrong.
Tuvey's approach is also better than the "partial TD" probabilty projection. Not as risk taking as DMD's projections, but still better than 0.4 TD! Curious... JT, have you ever attempted to measure your S1, S2 etc projections after the games have been played? If you do, does that ever factor into future projections?
Oh, and should I start Rothlisberger or Schaub this week? My hunch calculator is not working for this one... LOL. This is a great feature, thanks for the explanations!
one small suggestion for the start bench ratings.. and this just came to mind while thinking about willie parker in particular. it may be helpful if there was a (d) drop, (t) trade, (ww) waiver wire notation next to some of those relevant S1, S2s, Bs, etc. In parker's case, you have B this week. I agree entirely and would even put a (d) or (t) next to it, indicating that his value may be dried up or he may be bait. As for someone like crabtree, choice, a (ww) next to there name may be helpful to indicate that their stock is rising and this week's S3 may be next week's S2 based on all the research and trends you've seen while formulating the projections/suggestions.