Mr. Tremblay (1 Viewer)

[No Way Jose]

I think most of the discussion in here has focused on the first issue, and I could not disagree more with not taking a stance on what will happen with the individual situations (e.g., Favre) and using that stance as the basis of projections.

I don't understand why you want MT to have one method for generating Favre projections and another method for generating everyone else's projections.Take C.J. Spiller, for instance. Maybe he'll be in the three-way timeshare on a horrendously bad offense and post 550/3 and 200/0. Or maybe he really is Chris Johnson Lite and posts 1200/8 and 550/3. And there is an entire continuum of possibilities in between. MT does not pick one of those infinitely many possibilities and roll with it. He uses a weighted average of all of them. Or at least he tries to do that to the extent that it's possible.

Although Favre is an extreme case, MT is "not taking a stance on what will happen" for any player. Why should Favre be different?

[Apologies to MT if I'm misrepresenting him]

A weighted average is still a guess....I would rather take in all information given to me and use my expertise of scouting and information given and project. Making up averages and percentage of games just seems kinda silly to me. I would rather have a gut feeling from someone that follows FF very closely like MT, than percentages honestly. I see that Favre has been raised to 18 in MT's ranking and I think that is a little bit closer to reality. I'm not trying to single anyone out and I appreciate the responses. I find it interesting and helpful to understand how you guys rank and project.

 

[Doug Drinen]

FWIW, while I am good with his method and philosophy, I think Tremblay's Favre projection was awful from the start.

Favre's chance of retiring is no greater than Brees's, IMO.

 

[Chase Stuart]

FWIW, while I am good with his method and philosophy, I think Tremblay's Favre projection was awful from the start. Favre's chance of retiring is no greater than Brees's, IMO.

You have now ensured that Favre will get hurt in week 10 this season. Vikings fans know whom to blame.

 

[Maurile Tremblay]

Favre's chance of retiring is no greater than Brees's, IMO.

My Favre projection probably was awful from the start. But you can't really mean the above. Even if Favre's chance of retiring is only a few tenths of one percent, it has to be at least 100 times greater than Brees's chance of retiring (between now and the start of the season).

 

[Doug Drinen]

Favre's chance of retiring is no greater than Brees's, IMO.

My Favre projection probably was awful from the start. But you can't really mean the above. Even if Favre's chance of retiring is only a few tenths of one percent, it has to be at least 100 times greater than Brees's chance of retiring (between now and the start of the season).

Yeah, that was just my smart-alecky way of saying that I've never given any meaningful consideration to the possibility that Favre might retire.

 

[Tau837]

I think most of the discussion in here has focused on the first issue, and I could not disagree more with not taking a stance on what will happen with the individual situations (e.g., Favre) and using that stance as the basis of projections.

I don't understand why you want MT to have one method for generating Favre projections and another method for generating everyone else's projections.Take C.J. Spiller, for instance. Maybe he'll be in the three-way timeshare on a horrendously bad offense and post 550/3 and 200/0. Or maybe he really is Chris Johnson Lite and posts 1200/8 and 550/3. And there is an entire continuum of possibilities in between. MT does not pick one of those infinitely many possibilities and roll with it. He uses a weighted average of all of them. Or at least he tries to do that to the extent that it's possible.

Although Favre is an extreme case, MT is "not taking a stance on what will happen" for any player. Why should Favre be different?

[Apologies to MT if I'm misrepresenting him]

I agree that the same method should be used for Favre as for everyone else. I just used him as an example because (a) he is part of the premise of the thread and (b) he is a good example of someone who will almost certainly differ quite substantially from what MT is projecting.I think the appropriate way to value Spiller is to do the same thing I advocated earlier in the thread. Take the best information available and base a projection on how you think he and the other RBs will be used. That is, choose whether you think he will be Chris Johnson Lite, will be in a horrendous offense and full fledged timeshare or something in between. In Spiller's case, a projection can look like someone just projected "somewhere in between." And that is likely true of most players. That's exactly why Favre is a great example to discuss in this conversation.

 

[Doug Drinen]

I think the appropriate way to value Spiller is to do the same thing I advocated earlier in the thread. Take the best information available and base a projection on how you think he and the other RBs will be used. That is, choose whether you think he will be Chris Johnson Lite, will be in a horrendous offense and full fledged timeshare or something in between. In Spiller's case, a projection can look like someone just projected "somewhere in between." And that is likely true of most players. That's exactly why Favre is a great example to discuss in this conversation.

So you're saying you would prefer to see us project the mode instead of the mean (I do agree that in most cases, they're probably indistinguishable).I hate to keep harping on this, but do you want us to do games the same way?Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

 

[Dragon1952]

Seriously? He pointed out that the ranking is based on the liklihood of Favre coming back (56% according to Trembley). Trembley was ranking him based on where he would take him in a redraft draft. While most would concur that he should (and likely will) go higher, I completely understand why someone would be skeptical of taking him as their QB. I don't understand what's so confusing.

I just think it's more helpful to project what he's likely to accomplish if he does come back, and then put an asterisk next to him or something. Drafters aren't so stupid that they don't know Favre's situation. If they want to gamble that he's coming back it would be nice to know what his projections are. I mean, giving him bogus projections because there's a chance he won't play isn't very helpful. It doesn't tell you the rationale in the projections, so if you didn't read this thread you might be scratching your head. I suppose if you were smart enough, and knew the rationale, you could project his 9 game stats out to 16 in which case he'd project out to the #3 QB with 4120/30/14 and around 336 FP's (if my math is correct).

 

[Ramblin Wreck]

Seriously? He pointed out that the ranking is based on the liklihood of Favre coming back (56% according to Trembley). Trembley was ranking him based on where he would take him in a redraft draft. While most would concur that he should (and likely will) go higher, I completely understand why someone would be skeptical of taking him as their QB. I don't understand what's so confusing.

I just think it's more helpful to project what he's likely to accomplish if he does come back, and then put an asterisk next to him or something. Drafters aren't so stupid that they don't know Favre's situation. If they want to gamble that he's coming back it would be nice to know what his projections are. I mean, giving him bogus projections because there's a chance he won't play isn't very helpful. It doesn't tell you the rationale in the projections, so if you didn't read this thread you might be scratching your head. I suppose if you were smart enough, and knew the rationale, you could project his 9 game stats out to 16 in which case he'd project out to the #3 QB with 4120/30/14 and around 336 FP's (if my math is correct).

Yes, rankings should be done on a points per game basis. Favre with 200 points in 10 games is worth a lot more than Favre with 200 points in 16 games, obviously.

 

[GregR_2]

I think part of the problem is in the methodology of going from projections to rankings and why just multiplying the projections by these particular odds and then sorting them doesn't give a good result... in this case because how you value a guy who is expected to play like a backup compared to a guy who is a coin flip to not play or to be a top 5 QB isn't the same, and that's being entirely lost here.

If you're going to give me a set of rankings, I should be able to believe that if we have a league of 12 MT clones made the day he did his rankings, that Favre would be about the 30th QB taken in a redraft.

I have a really hard time believing that if MT was in such a draft the day that day he thought the odds were 56%, that he would have taken 17 other backup QBs before Favre.

As we all know, the value of backups vs starters is different. Backup value only comes in weeks they start, and only in an amount over what the next best option would have been. For most backups that is very few games. We're talking about what, 2-3 games worth on average for bye and injury, plus at best a small fraction of their weekly output from another 2-3 weeks on average they played the backup over the starter due to matchup.

Favre on the other hand, if he plays, is probably an every week starter. Especially for a team who drafted QBs in the 9-12 range. We still have the same 2-3 games expectation that our original QB can't play for bye or injury where Favre gets his full points. Then add in the 2-3 points per game the rest of the season that MT has him ahead of the lesser fantasy starting QBs on average. Then after you sum all that up, now you apply the odds so he only gets 56% of those points when we compare his expected contribution to the other backups.

Just eyeballing the numbers, I think they agree with my gut instinct that no one would actually let Favre slide to QB 30 if they had drafted the day MT did and believed as he did that the odds were 56% that Favre would play. Once you factor in things like the cost of replacement, even the factor that if Favre plays you can trade him for a good upgrade at another position, his value would have far exceeded where he ranked.

Now when you give me the projections themselves, I can see what MT thought Favre's chances are of playing and I can account for it in coming up with a ranked list of QBs myself. If I saw that in his projections I would bump Favre up from where the projections sorted him out to. But when you're going to publish rankings, you need to have taken that into account yourself since the person seeing that list isn't seeing those projections.

 

[GregR_2]

I think the appropriate way to value Spiller is to do the same thing I advocated earlier in the thread. Take the best information available and base a projection on how you think he and the other RBs will be used. That is, choose whether you think he will be Chris Johnson Lite, will be in a horrendous offense and full fledged timeshare or something in between. In Spiller's case, a projection can look like someone just projected "somewhere in between." And that is likely true of most players. That's exactly why Favre is a great example to discuss in this conversation.

So you're saying you would prefer to see us project the mode instead of the mean (I do agree that in most cases, they're probably indistinguishable).I hate to keep harping on this, but do you want us to do games the same way?Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

Definitely do not want you to do weekly projections like that.

 

[Ramblin Wreck]

Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

I think that's probably better than saying Johnson gets 0.5 TD, Burleson 0.4, and Scheffler 0.1. That's just a way to hedge your bets so you're never really wrong but you're never really correct either, in my opinion.

 

[No Way Jose]

Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

I think that's probably better than saying Johnson gets 0.5 TD, Burleson 0.4, and Scheffler 0.1. That's just a way to hedge your bets so you're never really wrong but you're never really correct either, in my opinion.

To add to that....if you feel like Burleson will have a better game, I would expect your projections to reflect that.

 

[ookook]

Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

I think that's probably better than saying Johnson gets 0.5 TD, Burleson 0.4, and Scheffler 0.1. That's just a way to hedge your bets so you're never really wrong but you're never really correct either, in my opinion.

I say exactly the opposite. Use the decimals and generate projected fantasy points based on them.

 

[GregR_2]

Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

I think that's probably better than saying Johnson gets 0.5 TD, Burleson 0.4, and Scheffler 0.1. That's just a way to hedge your bets so you're never really wrong but you're never really correct either, in my opinion.

Calvin Johnson 1 TD, Burleson 0 TD, Scheffler 0 TD Andre Johnson 1 TD, Walter 0 TD, Daniels 1 TD. That's giving you the projections in the format you're asking for. So how do you tell from what you're asking for.which player between Burleson and Walter they think is more likely to get a TD, and thus is the better start if yards are otherwise equal?Wouldn't you prefer to know that they think Burleson is more likely to get a TD than Walter, and by how much? It has nothing to do with hedging bets, it has to do with giving you the best information they can that expresses their feelings about the player.

 

[JKL]

Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

I think that's probably better than saying Johnson gets 0.5 TD, Burleson 0.4, and Scheffler 0.1. That's just a way to hedge your bets so you're never really wrong but you're never really correct either, in my opinion.

I would direct you to the cbs sports website to get your projections. One week last year, they projected my starting kicker, Matt Prater, to go 0 for 1 with no extra points. They took a stand. They certainly weren't hedging their bets. That team was going to get shutout but miss their only field goal. Didn't happen by the way. I don't go by their projections at all, but I find it annoying when they do something similar to a point spread for each fantasy matchup and it causes crazy results where one team is favored by about 40 over another because they happened to predict a whole value touchdown for some and not others. I wish I could glance at those spreads and actually think they had some basis in logic and fact.

 

[bostonfred]

Why would he only play 9 games unless injury is predicted?

Are you a 58 year old man who goes by the name of Dragon?

Still want some closure on this

 

[Dragon1952]

Why would he only play 9 games unless injury is predicted?

Are you a 58 year old man who goes by the name of Dragon?

Still want some closure on this

Gee, my own personal troll. You can see post #31 for your closure......hopefully.What other creature would you prefer? How about 'Dolphin', or is that too cornyof a name for a grown man too? I don't think that one is on the Chinese calendar though.BTW, is my maturity level actually being questioned here by a guy with Tinky Winkythe Teletubbie as his avatar

 

[Ramblin Wreck]

Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

I think that's probably better than saying Johnson gets 0.5 TD, Burleson 0.4, and Scheffler 0.1. That's just a way to hedge your bets so you're never really wrong but you're never really correct either, in my opinion.

Calvin Johnson 1 TD, Burleson 0 TD, Scheffler 0 TD Andre Johnson 1 TD, Walter 0 TD, Daniels 1 TD. That's giving you the projections in the format you're asking for. So how do you tell from what you're asking for.which player between Burleson and Walter they think is more likely to get a TD, and thus is the better start if yards are otherwise equal?Wouldn't you prefer to know that they think Burleson is more likely to get a TD than Walter, and by how much? It has nothing to do with hedging bets, it has to do with giving you the best information they can that expresses their feelings about the player.

Look at the yards. And Receptions too if you use PPR.

 

[Ramblin Wreck]

I would direct you to the cbs sports website to get your projections. One week last year, they projected my starting kicker, Matt Prater, to go 0 for 1 with no extra points. They took a stand. They certainly weren't hedging their bets. That team was going to get shutout but miss their only field goal. Didn't happen by the way.

 

[Stinkin Ref]

not to but just curious.....what factored in to him going from 30 to 18.....is 18 where you think you should have had him to begin with.....?

and i guess bottom line MT, if:

1. I thought you were the mutha of all that is fantasy football

2. and I chose to only use your projections/rankings......

3. and I had a draft tonight

would you really reccomend I take 17 other QB's before Favre.....?

 

[GregR_2]

Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

I think that's probably better than saying Johnson gets 0.5 TD, Burleson 0.4, and Scheffler 0.1. That's just a way to hedge your bets so you're never really wrong but you're never really correct either, in my opinion.

Calvin Johnson 1 TD, Burleson 0 TD, Scheffler 0 TD Andre Johnson 1 TD, Walter 0 TD, Daniels 1 TD.

That's giving you the projections in the format you're asking for. So how do you tell from what you're asking for.which player between Burleson and Walter they think is more likely to get a TD, and thus is the better start if yards are otherwise equal?

Wouldn't you prefer to know that they think Burleson is more likely to get a TD than Walter, and by how much? It has nothing to do with hedging bets, it has to do with giving you the best information they can that expresses their feelings about the player.

Look at the yards. And Receptions too if you use PPR.

I said if yards are equal. But they believe one is more likely to score touchdowns than the other. How do you tell that from the type of projection you're asking them to provide?The reason for not doing it as you're asking should be obvious, they can't give you their true assessment of the players without indicating how likely they think someone is to score a touchdown. And being forced to put a "0" or a "1" there doesn't let them give you the information you went to the projections looking for.

 

[Ramblin Wreck]

Lions vs. Jets: the most likely outcome is the Lions score 10 points. Revis will be on Calvin Johnson, so he probably won't get the TD. So I'll project one TD for Burleson and no TDs for anyone else.

I think that's probably better than saying Johnson gets 0.5 TD, Burleson 0.4, and Scheffler 0.1. That's just a way to hedge your bets so you're never really wrong but you're never really correct either, in my opinion.

Calvin Johnson 1 TD, Burleson 0 TD, Scheffler 0 TD Andre Johnson 1 TD, Walter 0 TD, Daniels 1 TD.

That's giving you the projections in the format you're asking for. So how do you tell from what you're asking for.which player between Burleson and Walter they think is more likely to get a TD, and thus is the better start if yards are otherwise equal?

Wouldn't you prefer to know that they think Burleson is more likely to get a TD than Walter, and by how much? It has nothing to do with hedging bets, it has to do with giving you the best information they can that expresses their feelings about the player.

Look at the yards. And Receptions too if you use PPR.

I said if yards are equal. But they believe one is more likely to score touchdowns than the other. How do you tell that from the type of projection you're asking them to provide?The reason for not doing it as you're asking should be obvious, they can't give you their true assessment of the players without indicating how likely they think someone is to score a touchdown. And being forced to put a "0" or a "1" there doesn't let them give you the information you went to the projections looking for.

I see your point but on the other hand, the way they do weekly projections now.... they can never claim they were very wrong. They just project out to a basic average amount. They rarely have a QB over two TDs a game, rarely have a RB/WR over one TD a game. How often does a QB throw 3+ touchdowns in a given week or a RB/WR score 2+ touchdowns? Go out on a limb.If they say Burleson and Walter are both 40 yards and 0 TDs, they are telling you they are an equal start. If they tell you Burleson is 40 yards and 0.3 TD and Walter is 40 yards and 0.2 TDs, are they really telling Burleson is the better starter? That's going to be 0.6 points difference in the two players for that given week.

 

[Doug Drinen]

If they say Burleson and Walter are both 40 yards and 0 TDs, they are telling you they are an equal start. If they tell you Burleson is 40 yards and 0.3 TD and Walter is 40 yards and 0.2 TDs, are they really telling Burleson is the better starter? That's going to be 0.6 points difference in the two players for that given week.

That's exactly what we want. In that hypo, we think the two are just about even and we want to tell you that they're just about even, but that Walter looks just a smidge better. If you force integer TD numbers, you lose those distinctions.But much worse, you create large distinctions where there are none.Burleson: 50 yards, .5 TDsHines Ward: 80 yards, .4 TDs.With fractional TDs, we're able to tell you that we think Ward is the better starter, which we do. If we tell you that we think Burleson will catch a TD and Ward won't, Burleson ends up looking like the better start (by a not-insignifcant margin) and our projections are telling you the opposite of what we think. We care about you way too much to do that

 

[Ramblin Wreck]

If they say Burleson and Walter are both 40 yards and 0 TDs, they are telling you they are an equal start. If they tell you Burleson is 40 yards and 0.3 TD and Walter is 40 yards and 0.2 TDs, are they really telling Burleson is the better starter? That's going to be 0.6 points difference in the two players for that given week.

That's exactly what we want. In that hypo, we think the two are just about even and we want to tell you that they're just about even, but that Walter looks just a smidge better. If you force integer TD numbers, you lose those distinctions.But much worse, you create large distinctions where there are none.Burleson: 50 yards, .5 TDsHines Ward: 80 yards, .4 TDs.With fractional TDs, we're able to tell you that we think Ward is the better starter, which we do. If we tell you that we think Burleson will catch a TD and Ward won't, Burleson ends up looking like the better start (by a not-insignifcant margin) and our projections are telling you the opposite of what we think. We care about you way too much to do that

You could project Ward for basically 65 yards and 0.5 TD each week. He's most likely going to end up with 50-80 yards most weeks and catch a TD in almost half the weeks. The projection is never really wrong and not really useful to me, IMO. If he puts up 50 yards and 0 TD, that's a dud of a week but you were close on the yards and only missed by half a TD. That's just as close as the weeks he puts up 80 yards and a score.What is useful is projections with analysis. Like you said earlier, Revis is on Calvin so I'll project Burleson to score the TD. But putting both at 0.5 TD doesn't tell us that at all. To each his own I guess. Maybe there is a way to do both.... the fractional stuff with gut feeling stuff too.

 

[GregR_2]

If they say Burleson and Walter are both 40 yards and 0 TDs, they are telling you they are an equal start. If they tell you Burleson is 40 yards and 0.3 TD and Walter is 40 yards and 0.2 TDs, are they really telling Burleson is the better starter? That's going to be 0.6 points difference in the two players for that given week.

That's exactly what we want. In that hypo, we think the two are just about even and we want to tell you that they're just about even, but that Walter looks just a smidge better. If you force integer TD numbers, you lose those distinctions.But much worse, you create large distinctions where there are none.

Burleson: 50 yards, .5 TDs

Hines Ward: 80 yards, .4 TDs.

With fractional TDs, we're able to tell you that we think Ward is the better starter, which we do. If we tell you that we think Burleson will catch a TD and Ward won't, Burleson ends up looking like the better start (by a not-insignifcant margin) and our projections are telling you the opposite of what we think. We care about you way too much to do that

You could project Ward for basically 65 yards and 0.5 TD each week. He's most likely going to end up with 50-80 yards most weeks and catch a TD in almost half the weeks. The projection is never really wrong and not really useful to me, IMO. If he puts up 50 yards and 0 TD, that's a dud of a week but you were close on the yards and only missed by half a TD. That's just as close as the weeks he puts up 80 yards and a score.What is useful is projections with analysis. Like you said earlier, Revis is on Calvin so I'll project Burleson to score the TD. But putting both at 0.5 TD doesn't tell us that at all. To each his own I guess. Maybe there is a way to do both.... the fractional stuff with gut feeling stuff too.

In regard to the bolded, yes, it does tell you that. Assuming some semblance of reality, I'm sure Dodds will have Calvin as likely to score more touchdowns on the season than Burleson. So if in the Jets game specifically they show Calvin and Burleson as equally likely to get a TD, that's telling you that Calvin dropped and Burleson went up for this game relative to each other compared to other weeks. As importantly it also tells you that at least as far as TDs goes, they think they are equally good starts. Something you won't see if they make one of them a 0 and one a 1 when they are equal in their eyes.

 

[Raiderfan32904]

Projections are a time tested, proven method of getting the decision-making tools all on the table to field your best team. I don't trust the baselines presnted in July for a yearly projection. Too many unknowns. The topic begins to stray over to a very important but unrelated topic of in-season management. But for this discussion, a lot of people are turning to the FBG individual and consolidated rankings to adjust their personal draft ranking list.

I respect Tremblay's opinion, and always consult his rankings prior to going into a draft. Coincidentally, post-draft, I get the highest ratings from Tremblay on the automated "rate-my-team" program. However, a lot of posters on here, myself included, feel that he needs to go out on a limb and project what a gimpy old media-hore will do in 16 games. If healthy, he should be ranked in the top 10 in all formats, based on the evidence that he hasn't retired (anything else is speculation). To worry about if Favre retires causes an adjustment in rankings, then go ahead and adjust for Mike Vick on the likelihood he commits another felony.

I am sure that MT will give us more usable projections in the future. Really no big deal for me as I don't do serious drafting till late August. But MT's projections are most in line with my personal rankings list, and I look forward to his revised list.

 

[Ramblin Wreck]

If they say Burleson and Walter are both 40 yards and 0 TDs, they are telling you they are an equal start. If they tell you Burleson is 40 yards and 0.3 TD and Walter is 40 yards and 0.2 TDs, are they really telling Burleson is the better starter? That's going to be 0.6 points difference in the two players for that given week.

That's exactly what we want. In that hypo, we think the two are just about even and we want to tell you that they're just about even, but that Walter looks just a smidge better. If you force integer TD numbers, you lose those distinctions.But much worse, you create large distinctions where there are none.

Burleson: 50 yards, .5 TDs

Hines Ward: 80 yards, .4 TDs.

With fractional TDs, we're able to tell you that we think Ward is the better starter, which we do. If we tell you that we think Burleson will catch a TD and Ward won't, Burleson ends up looking like the better start (by a not-insignifcant margin) and our projections are telling you the opposite of what we think. We care about you way too much to do that

You could project Ward for basically 65 yards and 0.5 TD each week. He's most likely going to end up with 50-80 yards most weeks and catch a TD in almost half the weeks. The projection is never really wrong and not really useful to me, IMO. If he puts up 50 yards and 0 TD, that's a dud of a week but you were close on the yards and only missed by half a TD. That's just as close as the weeks he puts up 80 yards and a score.What is useful is projections with analysis. Like you said earlier, Revis is on Calvin so I'll project Burleson to score the TD. But putting both at 0.5 TD doesn't tell us that at all. To each his own I guess. Maybe there is a way to do both.... the fractional stuff with gut feeling stuff too.

In regard to the bolded, yes, it does tell you that. Assuming some semblance of reality, I'm sure Dodds will have Calvin as likely to score more touchdowns on the season than Burleson. So if in the Jets game specifically they show Calvin and Burleson as equally likely to get a TD, that's telling you that Calvin dropped and Burleson went up for this game relative to each other compared to other weeks. As importantly it also tells you that at least as far as TDs goes, they think they are equally good starts. Something you won't see if they make one of them a 0 and one a 1 when they are equal in their eyes.

But if one believes Revis will shut down Calvin, then why give him half a touchdown at all? It skews the numbers. Also skews the numbers for Burleson. Maybe not when comparing him to Calvin per se, but when comparing him against Ward, Walter, Driver, or Evans that particular week.In week 14 last year, 37 different WR were projected to catch 0.4 to 0.6 touchdowns. In week 13, it was 38 different WR. In week 12, it was 36 different WR. That's absolutely hedging your bets and not really telling your readers anything. A cynic could claim every player is wrong every single week because no one ever scores half a TD. FBG will claim they are the most accurate but the majority of players every week will score 0 or 1 TD and they project right in the middle for so many of them. In a standard scoring system, WR14 and WR29 are separated by 1.2 points, which is basically nothing.

 

[GregR_2]

...

But if one believes Revis will shut down Calvin, then why give him half a touchdown at all? It skews the numbers. Also skews the numbers for Burleson. Maybe not when comparing him to Calvin per se, but when comparing him against Ward, Walter, Driver, or Evans that particular week.

Ramblin, you're kind of setting up situations to argue against that aren't what they are doing... or if they are doing then they would definitely be wrong.

If they think Calvin gets totally shut down by Revis, then he should have 0 TDs, yes. And if they have him at 0.5 TDs that means they don't think he's totally shut down. It would mean they think he would normally score TDs more often but that the Revis factor is going to drop him down to 0.5 TDs that week when against another opponent it might be higher.

So saying "he has 0.5 TDs" and saying "they believe Revis will shut him down completely" (and I'm adding the completely for clarity) is something FBGs wouldn't put out there. Or if they are, then I'd be on your side. But I haven't seen it. You're creating a situation then saying they would give a projection that they wouldn't give if that was truly their feelings on it.

In week 14 last year, 37 different WR were projected to catch 0.4 to 0.6 touchdowns. In week 13, it was 38 different WR. In week 12, it was 36 different WR. That's absolutely hedging your bets and not really telling your readers anything. A cynic could claim every player is wrong every single week because no one ever scores half a TD. FBG will claim they are the most accurate but the majority of players every week will score 0 or 1 TD and they project right in the middle for so many of them. In a standard scoring system, WR14 and WR29 are separated by 1.2 points, which is basically nothing.

This is the difference between projecting individual specific players versus projecting a nameless set of results. If I were to come up with a projection for how many passing yards the top QB in week 1 will have, I'd probably say something in the neighborhood of 350.

However if Peyton Manning is my top QB, and I were to project Peyton Manning's results, I should not project him at 350. The odds are very good that he won't be the top QB this week even though I think he's got the best chance overall. If I think he has a good matchup I should probably project him in the 300-320 range... which would be 40-60 yards better than his career average. If I think he has an average matchup I'd put him down around his 260 yard career average.

I think your objection is due to a common misbelief that projections should look the same as final results. But the reality is that they shouldn't. If my goal is to project what Peyton is actually most likely to do, I need to focus on him and not on what the QB who will lead the week in passing will do just because I think he has the best shot to be the top QB that week. I don't have a 100% certainty he's going to be that top guy so if I want to give the best projection I can, it needs to be lower than I would project the top QB at.

Various dice examples have been given several times already I believe. If I have a number of different sided dice... a 4 sided, a 6 sided, an 8 sided, a 10 sided, a 12 side, etc, on up to 20... it's probably very good odds that a 15 or higher will be rolled. If I'm projecting what the highest result will be, I should probably say around 15. But if I'm trying to project what the 20 sided die will roll, the most accurate prediction for it would be to say 10.5.

Now is that useful to you? You bet it is. If you're deciding between the 20 sided dice and the 18 sided dice, you know that your expectation for them is that the 20sided will average 10.5 and the 18 sided will average 9.5. That tells you just how close they actually are to each other... even if the 20 might toss up a 19 while the other dice could pop up a 2.

While football does have a little more that lends itself to predicting than dice rolls do, it is still very much something that is difficult to predict. It's much easier to predict a season's worth of games than it is a single game... and predicting a season accurately is very difficult.

What you see as a problem is actually an indication that FBGs understands the truth about how hard it is to predict these things. Rather than give you what they know is the wrong answer, they are giving you the best one that doesn't try to pretend it knows more than it does.

When they don't have a huge spread across those players when predicting a single game for them, what you should be walking away from them with is the understanding that they are telling you there often isn't that huge of a difference in what you should expect from one player to the next in a given week.

 

[radballs]

Good points Greg. Some introductory statistics classes would help a few people in this thread.

 

[LittlePhatty]

I would be fine with a no rank....but 30th?

Makes sense to me.He's saying that if Trent Edwards and Jake Delhomme are all that's left on the board, he'd rather take this other dude who might not even play this season.

 

[dgreen]

("If he plays, he's just ahead of Jay Cutler; if he doesn't play, he's just ahead of JaMarcus Russell").

 

[No Way Jose]

I would be fine with a no rank....but 30th?

Makes sense to me.He's saying that if Trent Edwards and Jake Delhomme are all that's left on the board, he'd rather take this HOFmer that will most likely play this season and be a top 10 QB.

Fify

 

[Tau837]

Good points Greg. Some introductory statistics classes would help a few people in this thread.

I'm not sure if you are lumping me in with this, since I have posted a few times that I disagree with MT's method for projections, best illustrated by the Favre example.As I said earlier, however, I view weekly projections and season projections differently. The thing I am disagreeing with in the season projections has to do with number of games played and the opportunity expected (i.e., passing attempts, rushing attempts, targets) in playing those games, which is much less of an issue with weekly projections. I'm not taking issue with the method used to project how a player will play once his opportunity has already been estimated.And since my stance is being viewed as inconsistent, I have to ask if the staff is being consistent. For example, some staff members project all QBs to play fewer than 16 games, based upon some average that is created because some number of QBs get hurt or are ineffective and thus miss some games. Well, couldn't the exact same thing be said about weekly projections? For example, some number of QBs who start each week will be hurt or pulled due to ineffectiveness and miss part of a game. So shouldn't the average of games played by QBs in a given week be less than 1.0? Does that mean those who use this factor in season-based projections should reduce the passing attempts and resulting numbers they project for Peyton Manning or Drew Brees each week, since the average QB won't play a full game? I think this is a silly notion... for the same reason I think discounting their games in full season projections is off base.I don't think I need to take a statistics course to understand the rationale being used, nor to disagree with it. But maybe I just don't know enough to know that I need it.

 

[Ramblin Wreck]

Good points Greg. Some introductory statistics classes would help a few people in this thread.

I'm not sure if you are lumping me in with this, since I have posted a few times that I disagree with MT's method for projections, best illustrated by the Favre example.As I said earlier, however, I view weekly projections and season projections differently. The thing I am disagreeing with in the season projections has to do with number of games played and the opportunity expected (i.e., passing attempts, rushing attempts, targets) in playing those games, which is much less of an issue with weekly projections. I'm not taking issue with the method used to project how a player will play once his opportunity has already been estimated.And since my stance is being viewed as inconsistent, I have to ask if the staff is being consistent. For example, some staff members project all QBs to play fewer than 16 games, based upon some average that is created because some number of QBs get hurt or are ineffective and thus miss some games. Well, couldn't the exact same thing be said about weekly projections? For example, some number of QBs who start each week will be hurt or pulled due to ineffectiveness and miss part of a game. So shouldn't the average of games played by QBs in a given week be less than 1.0? Does that mean those who use this factor in season-based projections should reduce the passing attempts and resulting numbers they project for Peyton Manning or Drew Brees each week, since the average QB won't play a full game? I think this is a silly notion... for the same reason I think discounting their games in full season projections is off base.I don't think I need to take a statistics course to understand the rationale being used, nor to disagree with it. But maybe I just don't know enough to know that I need it.

Pretty sure the snide comment about stats classes was directed at me because I disagree with the weekly projections and would like to see the writers on this site make a few more calls on players and then explain their reasoning as opposed to the way they do it now. I get what GregR explained but just disagree with which method is better. Radballs is just being a tool with his stats class comment.

 

[Ramblin Wreck]

Maurile, not to hijack but when did Beanie Wells become the next Marshall Faulk. Guy averaged one target a game last year and you have him catching almost 3 1/2 passes a game? Why?

Bump

 

[ookook]

Maurile, not to hijack but when did Beanie Wells become the next Marshall Faulk. Guy averaged one target a game last year and you have him catching almost 3 1/2 passes a game? Why?

Bump

He assumes Beanie will take some of Hightower's receptions. And he well may. 3.5 seems a bit high, but 2 seems bottom, so okay.

 

[Winky the tunnel ferret]

Favre's chance of retiring is no greater than Brees's, IMO.

My Favre projection probably was awful from the start. But you can't really mean the above. Even if Favre's chance of retiring is only a few tenths of one percent, it has to be at least 100 times greater than Brees's chance of retiring (between now and the start of the season).

Yeah, that was just my smart-alecky way of saying that I've never given any meaningful consideration to the possibility that Favre might retire.

 

[by_the_sea_wannabe]

Favre's chance of retiring is no greater than Brees's, IMO.

My Favre projection probably was awful from the start. But you can't really mean the above. Even if Favre's chance of retiring is only a few tenths of one percent, it has to be at least 100 times greater than Brees's chance of retiring (between now and the start of the season).

Yeah, that was just my smart-alecky way of saying that I've never given any meaningful consideration to the possibility that Favre might retire.

plenty of crow to eat in here

 

[No Way Jose]

I had a feeling this would get bumped....lololol

 

[LittlePhatty]

I would be fine with a no rank....but 30th?

Makes sense to me.He's saying that if Trent Edwards and Jake Delhomme are all that's left on the board, he'd rather take this HOFmer that will most likely play this season and be a top 10 QB.

Fify

Favre = retiredI'd still take him over Edwards or Delhomme, on the off chance that he comes back mid-year.

 

[Tau837]

Favre's chance of retiring is no greater than Brees's, IMO.

My Favre projection probably was awful from the start. But you can't really mean the above. Even if Favre's chance of retiring is only a few tenths of one percent, it has to be at least 100 times greater than Brees's chance of retiring (between now and the start of the season).

Yeah, that was just my smart-alecky way of saying that I've never given any meaningful consideration to the possibility that Favre might retire.

plenty of crow to eat in here

Why? Does the fact that nothing has happened yet mean it was appropriate to project Favre to play 56% of the games this year?

 

[Doug Drinen]

Are you guys serious?

Favre is playing week one.

 

[Maurile Tremblay]

Even if he plays week one, if the ankle is serious enough to make him retire a few more times between now and then, I think it significantly reduces the probability that he'll play week 17.