Mr. Chumley
Footballguy
Are there any articles here that compare the projections for the past few years with how these players actually performed?
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!
past projections vs performance (1 Viewer)
- Thread starter Mr. Chumley
- Start date
Dancing Bear
Footballguy
Mr. Chumley I presume ...
I am not aware of a comprehensive article that shows actual vs. projected performance, but interestingly enough I just finished a study for QB, RB, WR, TE, KI and DE in 2005. Here is a table which summarizes things for Footballguys.com.
[Pos] ----- [# Ranked] ----- [# Within 5] ----- [ % ] ----- [Top 10 Correct]
==================================================
QB ----- 60 ----- 16 ----- 26.7 ----- 5
RB ----- 110 ----- 20 ----- 18.2 ----- 5
WR ----- 132 ----- 16 ----- 12.1 ----- 3
TE ----- 57 ----- 17 ----- 29.8 ----- 9
KI ----- 33 ----- 12 ----- 36.4 ----- 5
DE ----- 32 ----- 15 ----- 46.9 ----- 3
TTL ----- 424 ----- 96 ----- 22.6 ----- 30
What's it all mean?
Footballguys.com predicted a little over 22% of all players in their ranking spot within 5 spots! [i.E. If a WR was predicted to finish WR 35, then they finished somewhere between WR 30 and WR 40]
Footballguys.com predicted 50% of all Top 10 finishers and they achieved this in spite of the losses of key guys like Culpepper, McAllister, Holmes, Walker, Moss, Jackson and the like to injury.
Note: Modified for readable format of the table ...
I am not aware of a comprehensive article that shows actual vs. projected performance, but interestingly enough I just finished a study for QB, RB, WR, TE, KI and DE in 2005. Here is a table which summarizes things for Footballguys.com.
[Pos] ----- [# Ranked] ----- [# Within 5] ----- [ % ] ----- [Top 10 Correct]
==================================================
QB ----- 60 ----- 16 ----- 26.7 ----- 5
RB ----- 110 ----- 20 ----- 18.2 ----- 5
WR ----- 132 ----- 16 ----- 12.1 ----- 3
TE ----- 57 ----- 17 ----- 29.8 ----- 9
KI ----- 33 ----- 12 ----- 36.4 ----- 5
DE ----- 32 ----- 15 ----- 46.9 ----- 3
TTL ----- 424 ----- 96 ----- 22.6 ----- 30
What's it all mean?
Footballguys.com predicted a little over 22% of all players in their ranking spot within 5 spots! [i.E. If a WR was predicted to finish WR 35, then they finished somewhere between WR 30 and WR 40]
Footballguys.com predicted 50% of all Top 10 finishers and they achieved this in spite of the losses of key guys like Culpepper, McAllister, Holmes, Walker, Moss, Jackson and the like to injury.
Note: Modified for readable format of the table ...
Last edited by a moderator:
Mr. Chumley
Footballguy
Thanks
Of course I ask because I wante dto see how accurate the projections have been over the past few years. Obviously you can't account for injuries.
Of course I ask because I wante dto see how accurate the projections have been over the past few years. Obviously you can't account for injuries.
Rather than studying by slots (As in how many projections were within 5 slots of actual finish), I want to know by points. I mean, the difference between WR11 and WR 35 is usually no larger than 2ppg or so. When FBG, or anyone, predicts a WR at 14, but he actually finishes at 34, that prediction is far more accurate than when you predict a RBat RB2, and he finishes at RB7.
Dancing Bear
Footballguy
Z-Dog,
You're wish is my command.
For Footballguys.com Rankings
[Pos] ----- [# Ranked] ----- [# Within 5] ----- [ % ] ----- [Top 10 Correct]
==================================================
QB ----- 60 ----- 16 ----- 26.7 ----- 5
RB ----- 110 ----- 20 ----- 18.2 ----- 5
WR ----- 132 ----- 16 ----- 12.1 ----- 3
TE ----- 57 ----- 17 ----- 29.8 ----- 9
KI ----- 33 ----- 12 ----- 36.4 ----- 5
DE ----- 32 ----- 15 ----- 46.9 ----- 3
TTL ----- 424 ----- 96 ----- 22.6 ----- 30
For Footballguys.com projections within +/- 2 ppg
[Pos] ----- [# Ranked] ----- [# Within 5] ----- [ % ] ----- [Top 10 Correct]
==================================================
QB ----- 60 ----- 28 ----- 46.7 ----- 7
RB ----- 110 ----- 74 ----- 67.3 ----- 5
WR ----- 132 ----- 70 ----- 53.0 ----- 3
TE ----- 57 ----- 49 ----- 86.0 ----- 7
KI ----- 33 ----- 17 ----- 51.5 ----- 5
DE ----- 32 ----- 14 ----- 43.8 ----- 6
TTL ----- 424 ----- 252 ----- 59.4 ----- 33
For Footballguys.com projections within +/- 1 ppg
[Pos] ----- [# Ranked] ----- [# Within 5] ----- [ % ] ----- [Top 10 Correct]
==================================================
QB ----- 60 ----- 16 ----- 26.7 ----- 2
RB ----- 110 ----- 51 ----- 46.4 ----- 2
WR ----- 132 ----- 42 ----- 31.8 ----- 2
TE ----- 57 ----- 31 ----- 54.4 ----- 4
KI ----- 33 ----- 10 ----- 30.3 ----- 2
DE ----- 32 ----- 5 ----- 15.6 ----- 2
TTL ----- 424 ----- 155 ----- 36.6 ----- 14
With a +/- 2 ppg bracketing of the Footballguys.com projections, they predict nearly 60% of all projected performances, and their success for Top 10 candidates goes up to 66%! Note that the TE predictions are 86%!
With a +/- 1 ppg bracketing of the same projections, they predict a little over 36% of all players projected, but the Top 10 candidates drops dramatically to only 14%.
You're wish is my command.
For Footballguys.com Rankings
[Pos] ----- [# Ranked] ----- [# Within 5] ----- [ % ] ----- [Top 10 Correct]
==================================================
QB ----- 60 ----- 16 ----- 26.7 ----- 5
RB ----- 110 ----- 20 ----- 18.2 ----- 5
WR ----- 132 ----- 16 ----- 12.1 ----- 3
TE ----- 57 ----- 17 ----- 29.8 ----- 9
KI ----- 33 ----- 12 ----- 36.4 ----- 5
DE ----- 32 ----- 15 ----- 46.9 ----- 3
TTL ----- 424 ----- 96 ----- 22.6 ----- 30
For Footballguys.com projections within +/- 2 ppg
[Pos] ----- [# Ranked] ----- [# Within 5] ----- [ % ] ----- [Top 10 Correct]
==================================================
QB ----- 60 ----- 28 ----- 46.7 ----- 7
RB ----- 110 ----- 74 ----- 67.3 ----- 5
WR ----- 132 ----- 70 ----- 53.0 ----- 3
TE ----- 57 ----- 49 ----- 86.0 ----- 7
KI ----- 33 ----- 17 ----- 51.5 ----- 5
DE ----- 32 ----- 14 ----- 43.8 ----- 6
TTL ----- 424 ----- 252 ----- 59.4 ----- 33
For Footballguys.com projections within +/- 1 ppg
[Pos] ----- [# Ranked] ----- [# Within 5] ----- [ % ] ----- [Top 10 Correct]
==================================================
QB ----- 60 ----- 16 ----- 26.7 ----- 2
RB ----- 110 ----- 51 ----- 46.4 ----- 2
WR ----- 132 ----- 42 ----- 31.8 ----- 2
TE ----- 57 ----- 31 ----- 54.4 ----- 4
KI ----- 33 ----- 10 ----- 30.3 ----- 2
DE ----- 32 ----- 5 ----- 15.6 ----- 2
TTL ----- 424 ----- 155 ----- 36.6 ----- 14
With a +/- 2 ppg bracketing of the Footballguys.com projections, they predict nearly 60% of all projected performances, and their success for Top 10 candidates goes up to 66%! Note that the TE predictions are 86%!
With a +/- 1 ppg bracketing of the same projections, they predict a little over 36% of all players projected, but the Top 10 candidates drops dramatically to only 14%.
Mr. Chumley
Footballguy
Good stuff. 
Thanks
Thanks
DreadedParms
Footballguy
What years did you do this study? I was thinking of conducting a similar study over several years of projections to see where footballguys.com is as far as making projections.
I can't find any projections prior to 2005 on the website, however.
I can't find any projections prior to 2005 on the website, however.
John Mamula
Moderator
Stating accuracy by rank means nothing. You could be 10 ranks off but if you missed by only 15 total points that is really good. Standard deviation by points is the most accurate way to judge projection accuracy IMO.In a way, I really don't care about player rankings when I do my own projections. I just care about the points I do project for them. Obviously rankings naturally fall out from this, but as long as I am fairly close to points scored in my projections I will be doing well.
The reason why FBG doesn't do this (from my understanding) is that they do not adjust projections down to historical values by ranking (aka AVT). Thus all of their projections are overvalued, which does not mean a lot if you are only looking at rankings, but as I just stated you shouldn't only be looking at rankings (especially if you use VBD).
The retort to the above is that you cannot project injuries (which is partially true). You can't project season ending injuries but you can fairly accurately project average player production that falls below average team production. You can also realize that a lot of hyped players will not be a 1,000 yard RB.
All of this is slightly going off on a tangent, but bottom line is you want to compare projected points to actual points.
John Mamula
Moderator
Is 60% good though? You are giving FBG a 64 total point margin of error. Regarding TEs, of course they will be within that range since TEs score so few points it is very difficult to miss by that many points.
Last edited by a moderator:
Dancing Bear
Footballguy
Bagger,
Whether or not 60% is accurate or not is really up to me. I was simply analyzing the data.
Without knowing how other sites or projections match up to compare with Footballguys.com you really cannot make a good judgment. The competition might be at 40% accuracy which makes Footballguys.com substantially more successful.
For me projections are only a stake in the sand from which to make correlative judgments. Trying to believe that they will be accurate for every player is pretty unrealistic ...
Whether or not 60% is accurate or not is really up to me. I was simply analyzing the data.
Without knowing how other sites or projections match up to compare with Footballguys.com you really cannot make a good judgment. The competition might be at 40% accuracy which makes Footballguys.com substantially more successful.
For me projections are only a stake in the sand from which to make correlative judgments. Trying to believe that they will be accurate for every player is pretty unrealistic ...
John Mamula
Moderator
The question of accuracy was not so much posed specifically to you, but rather a general question. It seemed to me that you were lauding FBG for doing so well based on a %, but what does that % tell us?Obviously nobody's projections will be 100% accurate. But what is an acceptable range?
