SHUFELT Computer Poll; Week 4
by JShufelt on Oct 1, 2010 10:21 AM PDT
Email
PrintI'm a nerd. So, I got that out of the way.
But I've been working on a spreadsheet that is essentially a resume based computer poll. Getting into the formula is somewhat confusing, but it makes sense to me. I can make the numbers simple for you to understand. Negative is bad, positive is good. The further the number is from 0, the more extreme it is. -5 is bad, +5 is good.
I wanted to have a system that was more objective than a human poll, but still based on a resume. I probably won't make a Fanshot every week, but I will follow up the AtQ BlogPoll ballot with a comment on it - unless the masses demand a Fanshot.
Here are things about the poll:
- It has a memory. It will remember what happened in Week 1. Team A may look bad Week 1, but in Week 6, they don't look so bad.
-
- That means the values are dynamic. To take an example from this year, Boise State squeaked by a highly ranked Virginia Tech team. The next week, Virginia Tech lost to James Madison. Boise State's close call to Virginia Tech does not look NEARLY as impressive now. Week 1 scores will actually change dynamically based upon results, not just Week 1, but every week of college football.
-
- Style points matter. This includes the margin of victory. Nobody is impressed when you pull of a 1 point squeaker to Saint Mary’s school of the blind. But everyone would be impressed when Alabama is beaten 35-0.
- There is no pride in playing an FCS team.
- There are three main values. (There are several formulas that make up these values, but I am going to try and simplify it.
- Team Avg (Team Average); This is simply your team’s performance in a game based upon a formula I’ve made up.
- Opp Avg (Opponent Average); This is an aggregated value of your opponent's Team Average.
- T+O Avg (Team Average plus Opponent Average.); This is the resume score. Good team vs. Good team will = Great ; Bad team vs. bad team = Awful
- I’ll also be putting up a number next to each extra curricular awards. This is T+O Game, which is the "points" allotted for their win, using a very similar formula as T+O Avg (T+O Avg is an average against all opponents; T+O Game is against that week's opponent). I know, it’s confusing. But it’s a computer poll, it’s suppose to be. And this isn’t used for the BCS.
- You get points for playing well against good teams, win or lose. You'll get more for beating good teams, but there is no shame for losing to a good team - as long as it is actually a good team. If you're 4-0, but it's all against FCS teams and the New Mexicos, you'll probably have a decent Team Avg, but your Opp Avg is going to make an embarrassment out of your team.
- Prior to week 1, ALL FBS teams are equal. Previous years have no influence on the poll.
- There are no bonus or penalty points for being in or absent in a BCS conference.
- A team's opponent's opponent's opponent's opponent's (potentiality infinite) score has influence on a team's perception, but the half life of that value is weekly. Meaning: Texas Tech's (Who played New Mexico, who played Oregon) score will influence Oregon's score, but only half as much as it will influence New Mexico's score.
- You now already know more about my computer poll than the ones that actually influence the BCS.
So, without further ado; here is the SHUFELT Computer Top 25; all scores are based on games played through
Week 4. The numbers are the team's T+O Avg
- Oregon (5.929)
- Ohio State (4.576)
- Alabama (4.210)
- Stanford (3.911)
- Florida (3.864)
- Oklahoma State (3.667)
- Utah (2.991)
- Miami (FL) (2.848)
- Arizona (2.795)
- Iowa (2.781)
- TCU (2.667)
- Missouri (2.649)
- Boise State (2.649)
- San Diego State (2.545)
- LSU (2.332)
- Nebraska (2.327)
- USC (2.271)
- Nevada (2.205)
- California (2.170)
- Auburn (2.070)
- Wisconsin (1.964)
- NC State (1.719)
- Arkansas (1.570)
- Florida State (1.576)
- South Carolina (1.570)
Now on to the extra curricular awards:
Flavor of the Week
This is awarded to the team that has the best win of the week. This is generally given to a team that has a solid win against a good team.
- Iowa over Ball State; 6.893
- Florida over Kentucky; 6.464
- Florida State over Wake Forest; 5.393
- Michigan over Bowling Green; 4.714
- Missouri over Miami (OH); 4.286
Here’s the down side of this system. It’s very resume. It’s SO resume that a win over Ball State in week 4 is impressive. Ball State isn’t exactly an impressive team (T+O Avg -1.536), but a shutout is impressive, and so is piling on 45 points to someone that isn’t an FCS team. Now, as the season goes on, we can look back, and I would assume UCLA’s win over Texas will be the top result for week 4.
Face Palm Award
This is awarded to the team that really should be embarrassed for their loss. This is generally given to the team that played poorly against a bad team
- New Mexico; -6.286
- Bowling Green; -3.679
- New Mexico State; -3.321
- Idaho; -3.321
- Texas; -3.286
New Mexico is bad. They are actually rated the worst team in this poll. UNLV is also pretty bad. To lose 35 points to UNLV is definitely a face palm score. Now, you might be wondering how Michigan gets a mention on the previous award, while Bowling Green gets mention for a Face Palm. You may also be wondering why I'm wearing this over sized hat. Keep your wonders to yourself! The real reason why, is because Bowling Green got throttled in the end. Losing by 44 just sucks, and Bowling Green isn't a New Mexico level of a bad of a team. But the difference between New Mexico bad and Bowling Green bad is pretty significant - nearly twice.
No Shame in Losing Award
This is awarded to the team that had the most respectable loss of the week. This is generally given to the team that played well against a good team.
- Arizona State; 6.107
- Arkansas; 3.678
- California; 3.178
- Notre Dame; 2.607
- BYU; 1.750
What’s funny in the No Shame in Losing Award, their resume actually improves for that week despite losing, granted, not as much as if they would have WON, but if they play against good competition, a loss is expected. Keep it close, and the computer applauds them. Arizona played Oregon closer than what many people expected. Considering how awful Arizona State’s schedule has been (Two FCS teams), this isn’t too shocking. Arkansas makes an appearance with a good showing against Alabama. They weren’t far off from pulling off a major upset.
Bitter Sweet Award
This is awarded to the team that the most disastrous win of the week. This is generally given to the team that played close against a bad team. Teams will find themselves here if they fail to win very convincingly against an FCS team.
- University Louisiana-Monroe; -9.857
- Nebraska; -8
- Tulsa; -6.143
- Michigan State; -4.471
- Virginia; -4.143
ULM, I know you’re in the Sun Belt, but pulling off a 1 point win against Southeastern Louisiana? Ugh.
Nebraska is up on here because they failed to shut out, and score high. Tulsa gets a nod because they allowed two TDs verses University of Central Arkansas.
This is a FanPost and does not necessarily reflect the views of SB Nation or the Addicted To Quack Moderators. FanPost opinions are valued expressions of opinion by passionate and knowledgeable Oregon fans.
47 comments
|
6 recs |
Do you like this story?
Comments
Any system that has fBSU just one place ahead of San Diego State has my blessing.
""Just remember," she said, "I'm holding you responsible for all of this.""
by benzduck on Oct 1, 2010 11:29 AM PDT reply actions 1 recs
So how many different sets of computer formulas did you go through before they ranks Oregon number 1?
But in all seriousness, this looks awesome. Great work on a crazy computer poll that sounds like it takes into account what a college football poll should look at.
I hate trees. Chop them down.
Set them on fire. Cause I hate Stanford!
In all honesty, I didn’t try to do that. The reason why Oregon is ranked number 1 is almost entirely because of their smack downs against their weakest opponents, and that Arizona State is actually a pretty good win against a pretty good opponent.
I predict that the winner of the Oregon/Stanford game is going to establish themselves as a No. 1 for a little while.
If Oregon loses, and then they don’t just obliterate WSU, they’ll take a big dive in the ranks.
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
For more fun things, here are the conference rankings:
- SEC
- Pac-10
- Big Ten
- Big 12
- Independent
- ACC
- WAC
- MWC
- Big East
- C-USA
- Sun Belt
- Mid-American
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
how much higher is the WAC than the MWC?
I only ask because you refuse to send me your spreadsheet to be proofed.
Hey Daisy, git me anudder one!
Addicted to Quack
Interesting. I’ve been looking at doing a formula of my own – where did you get good, clean, downloadable data for each team? And how much time did it take to put all of that data into a spreadsheet? Or did you use something else?
Defending maligned chants since 2009
I’ll probably make a copy of the excel sheet available next week.
In terms of data – the schedules, I found a spreadsheet that had full schedules for each conference. I… unlocked it… and then meshed them all on to a single sheet. I then went through and added several columns for the remainder data and the formulas.
I went through all the games, and punched in the scores for each team myself. The formulas then take care of the rest of the data. (That means yards, 3rd downs, all that other stuff is NOT factored. It is purely based on end results.) Because it isn’t statistics based, it is not predictive, outside of poll ranking shifts.
The amount of time to set up 120 teams, on 14 week schedules was approximately 4-5 hours, with 3-4 hours done on ensuring they were correct. It took about another 60-90 minutes to punch in all the scores. To figure out the formulas to do what I wanted, lay it out, and find ways to get the information I wanted, it was probably about 5 hours. (I went through about 3 iterations of formulas, trying to avoid circular references and proper VLOOKUP formulas)
I now have all the formulas set entirely in variables, so I can easily adjust values in the future, in case I realize X should be greater than it’s current value. This will be done purely off of results that I see, however I believe the parody between top 25 and bottom 25 is a good indication that the poll has some reflection within reality.
The worst teams in the country are, or order of worst to “best”: New Mexico, Washington State, Ball State, San Jose State, and Akron.
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
When you say “punch in scores” do you mean scores of the games? Or the scores that you calculated/assigned?
Defending maligned chants since 2009
I think
You just made this all up and there really is no computer model… Or maybe you really hired a Furd Nerd to put a computer model together for you.
Maybe I am just projecting that upon myself ’cause I can barely spell excel and would have no chance at putting something like that together…
Seriously, this is phantastic and I would love to get a look at the SS.
Both gold and muck come out of the same shaft...
I'm looking at it
and I can tell you it’s awesome! Great work by Shufelt and I’m sure once he tweaks it a little eventually Jared will be using JShufelt’s Computer Poll for his blog poll the rest of the year.
Hey Daisy, git me anudder one!
Addicted to Quack
So I let Matt Daddy do a review of it, and I tweaked the formula a bit based upon his review. I believe the current poll is sufficient (Though Oregon is now No. 2; Ohio State is No. 1), and I’ll post the results and write up a summery after next week.
So props to Matt Daddy for helping me out, and I’ll post the file next week for people to review.
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
I love this
Although I’m not in love with the idea of using MOV as one of the inputs for the quality of a win.
For instance, take the New Mexico game. We demolished them…almost everything went perfectly for us during the 1st half, and we’re up 59-0. Should we really be striving to get to 100? Does Chip keep the starters in that extra 2-3 drives (or the entire game?) to rack up as many points on a hapless team as possible, or kick that last field goal with 3 seconds left in the game when we could take a knee?
I do agree that some sort of “style points” should be awarded though, because you’re right, if the Ducks beat Wazzu by 2 points, Oregon’s stock should drop considerably. Perhaps there would be ways to tweak this so that it doesn’t encourage running up the score as much. Maybe the benefit of the MOV could decay exponentially (i.e., a 56 point win isn’t worth twice as much as a 28 point win… although I’m not sure what your calculations actually do) so that it doesn’t really matter too much if you beat a team by 42 or 72.
Again, I think what you’ve set up is awesome, and my comment is intended more to see what other people think than it is criticism. Great idea, and very nicely done.
I had a decay value originally, but I realized, no team is going to care about my poll, and they aren’t going to run up the score just to game my formula. If this was an honest computer poll that factored in the BCS championship – well first of all, I would object, and second of all, I would definitely add a decaying value.
HOWEVER, what I am thinking about doing is going to adding a bit of a bonus for a win, and/or a bit of a penalty for a loss, while making the MOV not meaningless, it will decrease it’s importance.
You do make a good point, and maybe I shouldn’t shelf the decay prospect entirely.
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
The other point about MOV, and why I think they told the computers to leave it out, is that the “human” voters do take MOV into account. Having the computers include it also in the BCS formula just magnifies the influence of MOV on the system.
""Just remember," she said, "I'm holding you responsible for all of this.""
How about an “adaptive” MOV score that would give less weight to points scored after a huge halftime lead, but give credit for pulling away and winning big in a one-score game after halftime. Also, MOV could count for more against quality opponents than UNM.
Example: USC last year. We led 24-17 and blew them out in the second half. You should get more MOV credit for a 30 point win in that situation than for a 30-point win where you led big early and the other team scored a bunch of mop-up points late.
""Just remember," she said, "I'm holding you responsible for all of this.""
That makes sense, but I’m not sure it’s feasible to manually input halftime scores for all those games. Correct me if I’m wrong, Shufelt. If it could be done in a reasonable amount of time, though, it might be used to add a little weight.
I do like weighting the MOV component based on the opponent.
Defending maligned chants since 2009
It has to be manually input
and since Benz is offering, he can spend all day Sunday researching scores quarter by quarter and inputing them into the spreadsheet
Hey Daisy, git me anudder one!
Addicted to Quack
He does have that unquestionably productive writing motor.
But I’m not sure if we’ll really get the same sort of poetry out of his work in excel.
JonathanPDX's #5 Benzduck fan.
by Bill Musgrave on Oct 1, 2010 7:11 PM PDT up reply actions
The word is "polymath."
I’m a fucking spreadsheet Shakespeare.
""Just remember," she said, "I'm holding you responsible for all of this.""
How many pages does your resume take up?
Advanced Photography Skills
Unmatched historical sports knowledge
eBay Treasuer Hunter
Eloquent Writer..
and now you want to stick Polymath on there too? I suppose you couldn’t be accused of being well rounded.
In all seriousness I am a spreadsheet nerd too. Some of my proudest accomplishments have been those for work (calculate the vesting expense of stock option grants in a single formula, estimate asset recovery timelines) and for personal use (draft my fantasy football team based upon my set criteria, score an NFL Draft prediction competition)
Ah Football, how I've missed you!
by AcadianTraverse on Oct 2, 2010 12:09 AM PDT up reply actions
You left out a few:
Party Vegetable
Perforated Septum
Family history of terminal flatulence
But anyone who’s met me in person will tell you I’m absolutely well-rounded.
As for Excel, I used to write spreadsheets so complex they would bring my computer to its knees, for things as mundane as operating expenses of a Volvo. Nowadays I hire people to write the formulae, and busy myself with things like conditional formatting and XML conversions.
""Just remember," she said, "I'm holding you responsible for all of this.""
Terminal Flatuence?
Is that Flatuence that manages to always be the final event of an exchange?
Or flatuence that is bad enough that family members have died from it?
Or does it strike up only while waiting for an airplane?
Inquiring minds want to know.
Ah Football, how I've missed you!
by AcadianTraverse on Oct 2, 2010 12:54 AM PDT up reply actions
TF. It's nothing to be sniffed at.
""Just remember," she said, "I'm holding you responsible for all of this.""
MOV is important when you take into account
that a) it shows relative strength versus common opponents and b) it is no longer factored into “real” polls so coaches aren’t incentivised to do it for ranking purposes (although I believe it would give a better comparison between teams if all teams “tried” to run it up.)
As for this poll, it is going to be resume based so yes, beating a team by 72 will matter if when a common opponent only beats them by 24. I don’t think it means team 1 is 3x better than team 2 but I think that it needs to be represented in some way.
Hey Daisy, git me anudder one!
Addicted to Quack
S-H-U-F-E-L-T-L-I-K-E-F-I-L-L-I-N-G-I-N-A-H-U-G-E-S-P-R-E-A-D-S-H-E-E-T
Awesome work Shu, I am nerding out over this.
I annoint thee king of the nerds.
JonathanPDX's #5 Benzduck fan.
Why not
eliminate games against FCS teams, unless the FBS team loses (or even wins by less than 20 points). Beating PSU by 70 tells you nothing about a team, while losing to PSU would have told you all you need to know.
true
But Oregon beating PSU 69 and ASU beating them by 45 should tell you something. Not a lot but still something. Especially if you are building this off of a resume.
Hey Daisy, git me anudder one!
Addicted to Quack
Playing poorly against FCS teams is quite telling. The way it is now, beating a FCS team, even by around 70 points, is a negative value. So you actually get penalized for scheduling an FCS teams nearly every time. You would need a shut out plus about 90 point Margin of Victory in order to get in the realm of positive points.
Teams that have scheduled multiple FCS teams have a significant mountain to climb to be a proven team, and in fact many teams will always have a better resume simply by NOT scheduling any FCS teams.
So essentially, beating PSU by 70 doesn’t tell me anything about a team, as reflected in the polls. However, beating PSU by 70, and another team beating them by 45 DOES tell something – albeit not a lot.
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
beating PSU by 70 doesn’t tell me anything about a team, as reflected in the polls. However, beating PSU by 70, and another team beating them by 45 DOES tell something – albeit not a lot.
I submit that once you get to around the 40-point level, a FCS blowout is a FCS blowout, and the point differential tells you nothing.
See? See what I did there? I’m throwin’ down.
""Just remember," she said, "I'm holding you responsible for all of this.""
Incredible effort for sure
Naturally if it’s going to be the “Shufelt Computer Poll” it should reflect your own biases towards MOV, quality of teams and whatnot. I’m sure you’ll refine the formula over the season.
After that it’s a matter of simply inputting the data in as detailed a format as you desire, but you can always farm that kind of thing out to Malaysia.
It would be interesting to see how you ranked the teams outside of the formula each week and then compare that to your formula to see how they matched up.
Ah Football, how I've missed you!
by AcadianTraverse on Oct 2, 2010 12:13 AM PDT reply actions
Well, a score of 0 would be about dead middle of the pack. Teams ranked around 58-62 are all near a score of 0. The actual median will be slightly higher or slightly lower year-to-year based on a few factors (Number of FCS teams, performance against FCS teams), but it will generally be near 0.
So if you have a score of 0, you neither were good, nor bad.
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
Cal #19? We CGBers always knew you were a closet Bear fan JShufelt!
The #1 greatest threat to America: BEARS
They looked to be rank this week too. That’s coming soon.
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
They looked to be rank this week too.
Cmon, that’s not nice! It was their bye week! So things got a little musky…
Defending maligned chants since 2009
I don’t intentionally raise or lower any team. I just punch in numbers and review to make sure there aren’t any obvious discrepancies.
It’s spelled "S-H-U-F-E-L-T-L-I-K-E-P-I-C-K-I-N-G-Y-O-U-R-N-O-S-E"
"I’ll give it my best shot, but you know I’m a signature ho." - daisyduck
and woosh! went the joke, sailing effervescently over Shufelt’s head…
Addicted to Quack, the home of Tako Tuesdays
Welcome to the club, Cody Ross. You can stay at my house.
Sailing effervescently?
There were bubbles trailing in its wake?
My name is Cliff Harris, and I’m here to lock shit down.
by HoodRiverDuck on Oct 5, 2010 1:23 PM PDT up reply actions
I’m entering the literary phase of my commenting career. I’m positively enkindled about it.
Addicted to Quack, the home of Tako Tuesdays
Welcome to the club, Cody Ross. You can stay at my house.
Very nice work Joe
I look forward to seeing your “awards” each week.
Where do each of the Pac-12 teams rank, at this juncture?
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Related FanPosts on Addicted To Quack
FanPosts
Community blog posts and discussion.
Recent FanPosts
FanShots
Quick hits of video, photos, quotes, chats, links and lists that you find around the web.
Recent FanShots
