Is it just me or the projections look high for most players? If I apply my league scoring, projections for week #1 are higher than the best players' averages from 2004 across the board. For example, J Peppers and C Grant averaged 9.2 and 8.3 pts per week last year in my league. Based on the week #1 projections, 20 players will score over 9.2 and 26 players will score over 8.3. There is something strange with tackles + assists I think. For LBs, there is a very large number of players projected to go over 10. It seems like 8 is generally a pretty good game for most players. I would say projections are at least 20% overstated for total fantasy points. I understand that relative rankings are more important than the absolute level of projections. But that seemed odd to me.
Fantasy Football - Footballguys Forums
This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!
Defensive Projections (1 Viewer)
- Thread starter Bouddha
- Start date
quiksilver
Footballguy
I think that tackles is really total tackles. So to get solo tackles you'd take total tackles less assists. If you do that their solo tackles for LBs is 98% of the 2004 weekly average, their solo tackles for DBs is 98% of the 2004 weekly average and their solo tackles for DLs is 97% of the 2004 weekly average.In a given week in 2004 all LBs averaged 462 solo tackles a week.In a given week in 2004 all DBs averaged 592 solo tackles a week.In a given week in 2004 all DLs averaged 344 solo tackles a week.If that's not the case and FBGs tackles = solo tackles, their projections are about 25% too high.Might help to clear up the headings on those projections.
That would make sense, but 1. Then the notation would not match how the season projections are made (a tackle is a tackle and an assist is an assist)
2. In the week #1 projections, some players have more assists than tackles, so I don't think your assumption is correct, although based on overall numbers it seemed reasonable
quiksilver
Footballguy
I don't see a single player that has more assists that tackles.
After looking at some of the comments here I'm not real sure what some of you are looking at or maybe where you are getting your numbers for comparison. The average NFL game will have between 45-50 solo tackles about 13.5 assists awarded. In week two there were 1445 credited tackles (a tad over 45 per team on average) and 435 credited assists (13.6 on average). The week three projections average just under 50 solo tackles per team and about 11.5 assists. While the projections are slightly higher (somewhere around 5%), its hardly anything to make a stink about. There were 3 players projected with 10 or more solo tackles this week (all linebackers). Last week there were 6 players with 10 or more solo tackles (5 LB and 1 DB)I don't know where you got your totals and averages for last season but I use stat's inc which is the service that drives most of the management programs out there. Stats inc doesn't always match up with NFL.com but since these are the numbers that the software programs use, they are what I go byTo answer the question about combined tackles, I never use combined tackles for anything. Most leagues only give half (or less) the value of a full tackle for an assist so lumping them all together would seriously tilt the numbers/player values for the vast majority. If by chance some of you are using last years overall numbers and comparing them to week 2 projections, don't forget to consider bye weeks. Is "any given Sunday" between weeks 3 and 10? If so there were 4 more teams projected in week 2. Hope that answers a few questions. Reb
John,Thanks for your response! When comparing to last year, I was using pts per game so bye weeks would not be considered.
If you think the sum of projections for all players is pretty much in line with total actual results, then I wonder if stats are not tilted too much in favor of the top players. I am saying this because I compared approximately the top 20 DLs, LBs and DBs to their projections (there are about 60 defensive players drafted in one of my league).
Last week for example. Using my scoring system:
Tackle = 1
Assist = 1
Sack = 2.5
INT = 4
Forced fumble = 3
Fumble rec = 1
Pass defended = 1.5
Here are the projections for a few "top 20" players and their actual score.
C Grant 13.0, actual 3.0
O Umenyiora 7.0, actual 8.0
D Edwards 17.5, actual 8.0
J Vilma 17.5, actual 11.0
L Fletcher 14.0, actual 5.0
M Peterson 13.5, actual 18.5
R Harrison 9.0, actual 9.5
M Boulware 12.0, actual 5.5
Etc.
You get the idea.
For players who are about the top 20 at each position, for week #2 I get this:
Projected 592.0
Actual 452.0
I got a similar thing for week #1. So what I do is I actually shave 25% off of the projections to obtain projected numbers in line with actual numbers.
As I said, maybe the NFL best defensive players' projections are too high and the less important players have projections too low? I don't know. But I think something is off a little. I agree that 5% off would not be something to complain about. But 25% is on the high side.
I am sure for most people what you have currently is more than adequate because it provides a cheat sheet. But in my league it is important to compare the actual projections on offense vs defense for various reasons, so that's why I picked the 25% reduction to your projections.
I will report on this with week #3 results and we could dig a little deeper regarding where the difference comes from.
Thanks!
Here is an update with week #3 scores. The player base I am using is smaller this week because of the bye weeks and more players missing games due to injury. But again I am finding the projections about 20% too high for the NFL's top players.Players included in this calculation:C Grant, O Umenyiora, D Edwards, J Vilma, M Peterson, L Fletcher, R Harrison, M Boulware, A Schobel, S Rogers, K Bulluck, R Lewis, A Davis, R Williams, T Polamalu, M Lewis, M Strahan, J Kearse, J Taylor, D Brooks, Z Thomas, M Washington, M Trufant, M Green, R Griffith, J Peppers, S Ellis, KGB, Spikes, J Farrior, E Barton, D Robinson, N Clements, B Scott, P Kerney, S Rice, D Howard, R Hayward, N Barnett, A Wilson, J Peterson, R Barber, D Sharper, L Milloy, T Suggs, J Smith, K Gregg, J Sharper, M Vrabel, W Witherspoon, E Reed, G Baxter, K Hamlin, K Williams, D Freeney, B Urlacher, K Brooking, D Nguyen, A Wilson, M Williams.Projected week #3 = 484Actual week #3 = 393John: any feedback?Thanks
We could try to do this prospectively as well. There are currently 5 LBs which have a projection of 14 tackles+assists or more for week #4. 14 T+A is a very high number which is reached by very few players. I will be shocked if 3/5 reach their projected (14, 14, 15, 15, 16). My guess is only 1/5 will reach projected. 0/5 would not surprise me either. 2/5 is possible. So in other words, I think these numbers are not "expected value". Of course I won't list these 5 LBs because it is subscriber material. That does not mean that there won't be 5 LBs in the NFL reaching 14. But I think for each player, the projections should equal their expected value for the week. For these LBs, I feel it's not the case.I hope I am not interpreted as being too critical. I am simply trying to be helpful - the goal should be to have the most accurate projections possible. I certainly would not be able to do these projections myself and it's easier for someone like me to point out the possible weaknesses than to do this from scratch!
quiksilver
Footballguy
Adding up all the NFL.com gamebooks:Week 1:
Gamebooks - 1356 defensive tackles
Footballguys - 1758 tackles
Week 2:
Gamebooks - 1385 defensive tackles
Footballguys - 1649 tackles
Week 3:
Gamebooks - 1232 defensive tackles
Footballguys - 1374 tackles
So the Footballguys projections for 30% higher than actual (per Gamebooks) in Week 1, 19% higher than actual in Week 2, and 12% higher than actual in Week 3. So it's gotten a lot more accurate since this thread was originally started. That's why comparing the Week 2 actuals to the Week 3 projections makes it look pretty good. But if you compare Week 1 projections to Week 1 actuals Week 2 to Week 2, etc., you can see why these questions orginally came up.
Highway Robbery
Footballguy
I agree that the tackles seemed high. Further since Sacks and Ints are so big, and tough to predicts, they should be predicted like touchdowns on offense. A player should be projected for 0.8 Sacks (ie an 80% chance for one sack). This is much more meaningful than the discreteet polarized 0,1,2 numbers.
After reading the comments I took a look at the first two weeks and found that they were indeed a little too high. As you see with week three I adjusted them and am tuning in. I think week 4 should be pretty close. Thanks for pointing these issues out. As bouddha correctly stated, it's all about getting it as close to correct as possible. I'll check the thread again next week and see how I scored :X Reb
Update with week 4 scores for the 60 players I am tracking .Projected 451.5Actual 363.0So actual was 20% lower than projected. For the 5 players who had projected T+A of 14-16, there were 4 that were below their projection and one that tied the projection so 0-4-1. Quiksilver do you have the week 4 total T+A vs projections for all players handy? I am curious.I think EB's suggestion to use decimal projected sacks, INTs, FFs, etc. is good. In addition to the T+A being a little overstated, I think the top players in the NFL have projected points from INTs, sacks, etc. a little too high as well because they are assigned full numbers. For example, the top DBs will be assigned a full INT more often than 0 INT. Regardless how good a DB is, I doubt 1 INT is the expected value. I could go along with 0.6, but 1 is high. Same for DLs who are projected to get 2 sacks. Maybe 1.1-1.3 would be more reasonable for example. So always using integers probably contributes to the high projections for "top" players. I am looking forward to the week #5 projections coming out tomorrow. I will come back to provide some feedback.Thanks for the input from everybody.
Week #5 update.For the 60 players I am tracking:Actual 349.5 ptsProjected 490.5 ptsSo projected was 40% above actual. For the players projected to get a very high number of T+A:N Barnett projected 16, actual 8D Edwards projected 14, actual 15M Peterson projected 15, actual 3J Vilma projected 16, actual 15J Trotter projected 13, actual 9A Archuleta projected 13, actual 3So 1 over, 5 under. The total of these 6 players: projected 87, actual 53.I have not looked at INTs, sacks, etc. yet. I believe most of the difference between actual and projected comes from T+A. Without comparing actual vs projected, I think the INT and sack projections were greatly improved in week #4, especially with half sacks being used in the projections for several players, and one DB getting 0.5 INT projected. This is more precise than always projecting 0, 1, or 2.
Last edited by a moderator:
BloodyPigSkin
Footballguy
Maybe we should just have TACKLE projections for each IDP and rank accordingly. Then supplement that with a list of projected INT/FUMBLE/SACK for the TEAM, not an individual player. I think when we guess that a guy with 2 tackles is getting the sack or pick, it scews his ranking up the board which in turn can send better players down the board. In my opinion, tackles are far easier to predict for player to player. We could look at the tackle projections, compare it to the supplementary list of team takeaway/giveaway projections, and then decide accordingly.
