RB Projections Experiment: Week 11 attempt

[Challie]

I'm trying my hand at a RB Projections system. Here's a post of week 10 results: http://www.fantasyfootballcafe.com/foru ... 768#678768

Hell, it can't be any worse than Hector & Victor.

DISCLAIMER: This is still a work in progress, so really take it with a grain of salt.

Basically this system attempts to average the RBs past performance, the teams performance at RBs weighted to the back's likely role in the offense that week, and the opponent DEFs average performance against RBs.

Since the RB's role in the offense is sometimes not very predictable til closer to Sunday some things could change drastically in an update (for instance, these stats assume Chris Brown is getting 90% of the Titans RB stats, if it was rerun for Antowain Smith, there would be a different prediction). So a late week update could change things, especially for cloudy backfields like CAR, MIN, TEN, PIT, OAK, etc.

Here goes nothing:

Rank Running Back Opponent Predicted
1 Shaun Alexander MIA 13.2
2 Curtis Martin CLE 12.5
3 LaDainian Tomlinson OAK 12.2
4 Tiki Barber ATL 12.0
5 Jamal Lewis DAL 11.8
6 Emmitt Smith CAR 11.7
7 Corey Dillon KC 11.4
8 Reuben Droughns NO 11.4
9 Derrick Blaylock NE 11.3
10 Jerome Bettis CIN 11.2
11 Chris Brown JAX 11.1
12 Clinton Portis PHI 11.1
13 Edgerrin James CHI 11.0
14 Michael Pittman SF 10.8
15 Deuce McAllister DEN 10.8
16 Ahman Green HOU 10.7
17 Domanick Davis GB 10.7
18 Duce Staley CIN 10.6
19 Willis McGahee STL 10.3
20 Kevan Barlow TB 10.3
21 Rudi Johnson PIT 10.1
22 Fred Taylor TEN 9.6
23 Warrick Dunn NYG 9.6
24 Thomas Jones IND 9.5
25 Marshall Faulk BUF 9.2
26 Eddie George BAL 8.9
27 Brian Westbrook WAS 8.4
28 Anthony Thomas IND 8.3
29 Kevin Jones MIN 8.3
30 T.J. Duckett NYG 8.1
31 Lee Suggs NYJ 7.9
32 William Green NYJ 7.7
33 Troy Hambrick CAR 7.7
34 Tyrone Wheatley SD 7.6
35 Onterrio Smith DET 7.5
36 Sammy Morris SEA 7.2
37 Shawn Bryson MIN 7.1
38 Travis Minor SEA 6.8
39 Jesse Chatman OAK 6.4
40 Steven Jackson BUF 6.4

[matmat]

you're running into a similar problem that I decided I was having ranking defences. Notice how small the spread in your projections is -- that's unreasonable. There has got to be a better way of doing this that takes into account probabilities of scoring touchdowns and not "average" production.

[Challie]

Surprises in these numbers:

* Blaylock about as high as Dillon on Monday. We'll see.
* Droughns vs. NO isn't bigger - but looking back at Shanny and those 1-yard TE throws, I guess it's not surprising.
* McGahee being so low vs Rams - but til he scores TDs, the numbers can't project him very high
* None of the MIN backs making an impact vs. DET, but until it becomes clear who's getting carries, this algorithm is going to reflect that.

[Challie]

you're running into a similar problem that I decided I was having ranking defences. Notice how small the spread in your projections is -- that's unreasonable. There has got to be a better way of doing this that takes into account probabilities of scoring touchdowns and not "average" production.

I totally agree, mat. But I don't see any way around that. It's hard to statistically predict something like 3 rec TDs for Muhammed or 2 fumbles for Brown in a given game. I have no idea of how to predict the likelihood of specific things like that. That's way beyond what I'm capable of dreaming up.

[matmat]

you're running into a similar problem that I decided I was having ranking defences. Notice how small the spread in your projections is -- that's unreasonable. There has got to be a better way of doing this that takes into account probabilities of scoring touchdowns and not "average" production.

I totally agree, mat. But I don't see any way around that. It's hard to statistically predict something like 3 rec TDs for Muhammed or 2 fumbles for Brown in a given game. I have no idea of how to predict the likelihood of specific things like that. That's way beyond what I'm capable of dreaming up.

I toyed with this a little bit earlier this season, but had many other things to work on. I looked at this for passing TDs for QBs. I plotted the #tds thrown a game vs # games that number was thrown in (a histogram) and found that the mode was 1 with a bell-curve like distribution of std dev. of about 0.8 (I'm doing this from memory now). This was for the entire league, not individual defences or QBs. Anyhow... I couldn't really discern much from that graph at the time.

Looking at it from another direction, it might be interesting to make a plot of passyds/G vs. pass TDs/G and
passyds/G vs. rushyds/G etc. for various teams and things and see if we can get any info from that...

lots of math/statistics and such.


and

[Sixxgunn]

My head hurts.I'll just let my girlfriend pick her favorite color uniforms.

[chinch sacs]

My head hurts.I'll just let my girlfriend pick her favorite color uniforms.

thats funny stuff

[Mercer Boy]

I'm not liking Ahman Green at #16...

I guess Houston does have a fairly stingy run defense on TD's...but I hope a surging Packers squad can lay down the smack on a fading Texans team.

[Warpigs]

I'm not liking Ahman Green at #16...

I guess Houston does have a fairly stingy run defense on TD's...but I hope a surging Packers squad can lay down the smack on a fading Texans team.

I'm with you Mercer, the biggest surprise to me was Ahman so low versus a defense that he should fare well against. Also wasn't expecting Shaun Alexander to the the best option of all RB considering his matchup with Miami

[BrutallyHuge]

Miami's run D stinks.

With Miami's O on the field for so little time, opponents are just wearing down the clock. Seau is out and they cannot stop the run anymore (weren't they near the top of the NFL last year?)

I predict 400 yards and 5 TDs for SA and 6 of 25 for 74 yards with 4 INTs and 2 fumbles for Feeley.