Week 10 BTBS Picks!
by Bill C. on Nov 4, 2009 9:15 AM CST
in National Football 2009
13 comments
On we roll...
Last Week: We crept back over .500 last week, which is good I guess. Sometime last week, I started reaching a conclusion that Bill Simmons reinforced in his weekend NFL picks column. The fact is, the spreads are a whole lot tougher in late-October and November than in September and early-October. I really want to be doing better than 50%, and I think 55%-60% is still possible, for the rest of this year (until I can update my methods) I'll take anything over 50% as victory. And just looking at how tight a lot of my projections are compared to the spread, some good or bad breaks could make a pretty humongous difference this week.
(Meanwhile, knock on wood, but my overall struggles haven't translated to my ten-per-week FO picks, in which I've gone 14-6 over the last two weeks. Again, KNOCK ON WOOD. In fact, why did I even bring this up with my amazing jinx power?? I suck.)
| Category | Last Week | Season |
| All Games | 27-26 (50.9%) | 245-195-6 (55.6%) |
| Big 12 | 2-4 (33.3%) | 34-28-2 (54.7%) |
| "LOCKS" | 6-4 (60.0%) | 26-22-1 (54.1%) |
Yep, the "LOCKS" are still anything but.
| Date | Time | Game | Projection | Spread | ATS Verdict |
| Sat., 11/7 | 11:00am | Central Florida at Texas | UT by 32.2 | UT -34.5 | UCF (WIN) |
| 11:30am | Kansas at Kansas State | KU by 4 | KU -3 | ||
| 12:30pm | Texas A&M at Colorado | ATM by 3.5 | ATM -5 | CU (WIN) |
|
| 1:00pm | Baylor at Missouri | Mizzou by 16.5 | Mizzou -17 | Baylor (WIN) |
|
| 2:30pm | Oklahoma State at Iowa State | ISU by 3.0 (?) | ISU +7.5 | ||
| 7:00pm | Oklahoma at Nebraska | NU by 6.1 | NU +6 | NU (WIN) |
- My professional recommendation: lay off of all of these games, particularly the first four. It makes me feel pretty confident in the quality of my picks with how close they are to the spread in most cases, but...those first four are TOO close to the spread.
- Meanwhile, the last two picks are interesting to say the least.
All games after the jump.
| Date | Time | Game | Projection | Spread | ATS Verdict |
| Thurs., 11/5 | 6:30pm | Virginia Tech at East Carolina | VT by 24.0 | VT -11.5 | VT (WIN) |
| Eastern Michigan at Northern Illinois | NIU by 17.1 | NIU -21 | |||
| Miami-OH at Temple | Temple by 7.0 | Temple -14 | M-OH (WIN) |
||
| Fri., 11/6 | 7:00pm | Boise State at Louisiana Tech | Boise by 11.7 | Boise -21 | La Tech (WIN) |
| Sat., 11/7 | 11:00am | Central Florida at Texas | UT by 32.2 | UT -34.5 | UCF (WIN) |
| Illinois at Minnesota | Minny by 17.8 | Minny -6 | |||
| Louisville at West Virginia | WVU by 18.2 | WVU -19.5 | L'ville (WIN) |
||
| Northwestern at Iowa | Iowa by 31.8 | Iowa -17.5 | |||
| Purdue at Michigan | UM by 2.5 | UM -4 | Purdue (WIN) |
||
| Syracuse at Pittsburgh | Pitt by 13.4 | Pitt -20.5 | |||
| Virginia at Miami-FL | Miami by 21.8 | Miami -13.5 | Miami (WIN) |
||
| W'ern Michigan at Michigan State | MSU by 25.2 | MSU -18 | MSU (WIN) |
||
| Wisconsin at Indiana | Wiscy by 15.8 | Wiscy -11 | |||
| 11:21am | South Carolina at Arkansas | Hogs by 8.8 | Hogs -5.5 | Hogs (WIN) |
|
| 11:30am | Kansas at Kansas State | KU by 4.0 | KU -3 | ||
| 12:00pm | Eastern Kentucky at Kentucky | UK by 23.8 | N/A | ||
| Tennessee Tech at Georgia | UGa by 46.2 | N/A | |||
| Maryland at N.C. State | NC St by 2.2 | NC St -7 | Maryland (TIE) |
||
| 12:30pm | Furman at Auburn | Auburn by 29.4 | N/A | ||
| 12:30pm | Texas A&M at Colorado | ATM by 3.5 | ATM -5 | CU (WIN) |
|
| 1:00pm | Baylor at Missouri | Mizzou by 16.5 | Mizzou -17 | Baylor (WIN) |
|
| BYU at Wyoming | BYU by 7.6 | BYU -13 | |||
| Florida Atlantic at UAB | FAU by 2.2 | FAU +5 | |||
| 1:30pm | Navy at Notre Dame | ND by 8.8 | ND -11 |
Navy (WIN) |
|
| 2:00pm | Rice at SMU | SMU by 20.8 | SMU -18 | ||
| 2:30pm | Army at Air Force | AFA by 28.1 | AFA -17 | AFA (WIN) |
|
| Duke at North Carolina | UNC by 13.2 | UNC -8.5 | UNC (WIN) |
||
| Kent State at Akron | Akron by 4.9 | Akron +3 | Akron (WIN) |
||
| LSU at Alabama | 'Bama by 23.4 | 'Bama -9 | 'BAMA (TIE) |
||
| Ohio State at Penn State | PSU by 15.9 | PSU -4 | |||
| Oklahoma State at Iowa State | ISU by 3.0 | ISU +7.5 | |||
| Oregon at Stanford | Ducks by 17.7 | Ducks -5 | |||
| UL-Lafayette at Arkansas State | ASU by 17.9 | ASU -12 | |||
| UTEP at Tulane | Tulane by 1.7 | Tulane +7 | Tulane (WIN) |
||
| Wake Forest at Georgia Tech | GT by 11.1 | GT -15 | Wake (WIN) |
||
| Washington at UCLA | UCLA by 9.1 | UCLA -4 | |||
| Washington State at Arizona | Arizona by 31.7 | Arizona -31 | Arizona (WIN) |
||
| 3:00pm | TCU at San Diego State | TCU by 20.2 | TCU -24.5 | ||
| 3:30pm | Florida Int'l at Middle Tennessee | MTSU by 20.8 | MTSU -12.5 | MTSU (WIN) |
|
| 4:00pm | Troy at Western Kentucky | Troy by 15.9 | Troy -24 | WKU (WIN) |
|
| 5:00pm | New Mexico at Utah | Utah by 30.9 | Utah -27 | Utah (WIN) |
|
| 6:00pm | Memphis at Tennessee | Vols by 37.3 | Vols -26 | Vols (WIN) |
|
| Oregon State at California | Cal by 4.6 | Cal -7 | OSU (WIN) |
||
| 6:15pm | Vanderbilt at Florida | UF by 50.4 | UF -32.5 | ||
| 6:30pm | Northern Arizona at Ole Miss | Ole Miss by 32.1 | N/A | ||
| Houston at Tulsa | Tulsa by 3.9 | Tulsa +1.5 | Tulsa (WIN) |
||
| 6:45pm | Florida State at Clemson | Clemson by 14.7 | Clemson -8.5 | Clemson (WIN) |
|
| 7:00pm | Connecticut at Cincinnati | Cincy by 22.4 | Cincy -16.5 |
||
| Oklahoma at Nebraska | NU by 6.1 | NU +6 | NU (WIN) |
||
| USC at Arizona State | USC by 7.9 | USC -11 | ASU (WIN) |
||
| 9:00pm | Colorado State at UNLV | CSU by 7.1 | CSU -1.5 | ||
| 9:05pm | Utah State at Hawaii | Hawaii by 11.7 | Hawaii -3 | Hawaii (WIN) |
|
| 9:30pm | Fresno State at Idaho | Fresno by 4.7 | Fresno -7 | ||
| Sun., 11/8 | 7:30pm | Nevada at San Jose State | SJSU by 2.7 | SJSU +13 | SJSU |
- Not a ton of "LOCKS" this week, though I'm sure I've missed a couple. Loved the "Florida by 50" projection. I know Florida's catching some crap for not living up to expectations, but again...they and Alabama are so far ahead of everybody else in the rankings. The only reason they've seemed disappointing is that they were supposed to be the most recent G.O.A.T., and they're only really good instead.
- 28-18-2 for the week.
0 recs |
13 comments
|
Email
Print
Comments
Teams the projections are most accurate for
I want to start by saying that this is not meant as a criticism of Bill’s projections in any way, I have been following them (and profiting from them)for the majority of the season.
I was reading some of the comments last week about teams being over-valued so I decided to try and see if that was the case. I found this a good way to avoid studying for my contracts exam, so armed with a solid B in stat 3200 3 years ago I used the absolute values of difference from Bill’s projections for the season. Now this doesn’t tell which teams are over-valued or under-valued, but it does show for which teams the numbers have been the most accurate.
For each team, I averaged the absolute values for each week and then took an average for all of the teams and conferences. I then took the average of each team’s average absolute value. Then I took the standard deviation to find out which teams were outside that standard deviation from the mean.
The extremely accurate teams(outside 2 std. dev) were: San Jose St.(off by an average of 3.844 points) and Utah (3.85).
Other teams outside 1 standard deviation: FAU, Utah St., kansas, Wyoming, Syracuse, Alabama, Michigan St., FIU, West Virginia, Arizona, Oregon St., Wisconsin, Northwestern, Minnesota, UConn, Wake, Notre Dame, Buffalo, Ball St, and North Texas.
The extremely inaccurate teams were: Rice, ATM, and Cal.
The others outside of 1 std deviation were: Texas Tech, Duke, Virginia, Kansas St., UTEP, New Mexico, Tulsa, Air Force, NC St, Va tech, Nebraska, Miss St, Iowa, UNLV, Tulane and E. Michigan.
The inaccurate teams were not terribly surprising in that most those teams are extremely schizophrenic, but if anyone wants to take a run at what is common about the accurate teams go for it because I did not notice anything.
Hmmm...
The inaccurate teams were not terribly surprising in that most those teams are extremely schizophrenic, but if anyone wants to take a run at what is common about the accurate teams go for it because I did not notice anything.
Sounds like Bill might have a basis for another WinCorr analysis and see if there are any per play measures that might tip this off.
by RPT on Nov 4, 2009 7:08 PM CST up reply actions
Locks
I was also curious about why the locks were hitting at the same rate as the regular picks. My assumption was that when two teams that the numbers have been accurate for play each other the record against the spread would be better. Back throgh week one I added th two teams averages together, then I looked at all the match-ups that were outside one standard deviation from the average.
On the season when the sum of the teams playing is less than the average-standard deviation Bill’s picks are 85-35.
When the sum of the two teams playing is greater than the average+standard deviation Bill’s picks are 40-54-1.
Using the standards for this weeks games, the “locks” would be:
Louisville-Pitt; Illinois-Minnesota; Syracuse-Pitt; LSU-Alabama; Miami Ohio-Temple; Nevada-San Jose St; W. Mich-Mich St; Wake Forest-Ga Tech; Ohio St-Penn St; BGSU-Buffalo; Wisconsin-Indiana; and TCU-San Diego St.
The “stay-aways” would be:
SMU-Rice; UTEP-Tulane; Houston-Tulsa; Virginia-Miami; Maryland-NC St; ATM-Colorado; Oklahoma-Nebraska; Colorado St-UNLV; Army-Air Force; Duke-UNC; Oregon St.-Cal.
So pretty much the bet the Big 10 and stay away from C-USA and the ACC. Now having said all this Id assume every thing will end up .500.
This is awesome, awesome stuff.
I do plan on diving into what did and didn’t work this offseason, but I really didn’t have much of an idea where to begin…this helps…thanks…
Rock M Nation
Thrust nunchuk upward!
No problem
Also, Do you have your projections from some of these goofy Tuesday night games. I know Im missing Buffalo and Bowling Green from yesterday and East Carolina and Marshall last week, I cant find a projection for Arkansas and SMS from week 1 either. Thanks
To make sure I am following you
For each team you compared the projected outcome against the actual outcome for each game?
Using Tennessee as an example. In week 1 TN was projected over WKU by 35.4. TN won the game by 56. So:
Week 1: ABS == 21.6
Week 2: ABS == 12
Week 3: ABS == 6
Week 4: ABS (29.4,11) == 18.4
Week 5: ABS (9.2,-4) == 13.2
Week 6: ABS (8.6,26) == 17.4
Week 8: ABS (9.6,-2) == 11.6
Week 9: ABS (17.4,18) == .6
Then you averaged these to give you MEAN (21.6, 12, 6, 18.4, 13.2, 17.4, 11.6, 0.6) == 8
So the the projections for Tennessee were within 8 points of the final outcome, on average. So Tennessee’s (I’m about to make up a phrase) “average projection discrepancy” (APD) is 8.
Then you did this for every team, and averaged those numbers together to get the overall average APD Let’s assume, for the sake of argument, that the overall average APD is 10.
Then you figured out the first and second standard deviation from this overall number. Let’s assume that each standard deviation is 3 points. You could then break all the the teams into five categories, namely:
1) the “impossible to project” category (APD > 16)
2) the really tough to project category (13 < APD < 16)
3) The “about average to project” category (7 < APD < 13)
4) The “fairly easy to project” category (4 < APD < 7)
5) The “nail on the head” category (APD < 4)
Actually, maybe I don’t even need those categories. Correct me if I am wrong here, because I want to make sure I am following. But in this example, We got team X playing team Y. We take the average APD of those two teams, and if that average is less than 7, then Bill is probably going to get it right.
On the other hand, if team Q is playing team P. We average the APD of team Q and P and it comes out to be greater than 13, then Bill probably going to get that one wrong.
Did I get that right? Am I following you?
________________________________
Eric Berry is better at football than you.
yup
Yeah, thats basically how I did it. The actual numbers were:
Average: 12.34
Std dev: 3.82
So adding opponents would be 24.68-3.82=20.86, and 24.86+3.82=28.5
So for the season if the total is 20.86 or less Bill is 85-35; if the total is greater than 28.5 Bill is 40-54-1
Awesome
And my guesses weren’t even that far offf.
Thanks.
Also, I’d be happy to trade you some of my vast contract law knowledge (the VA bar exam was my beeotch) for your listing of team APD’s. My email is kidbourbon at gmail dot com
UCC, check
Parol evidence rule, check
Statute of frauds, check
assignment, check
third party beneficiaries, check
offer and acceptance, check
consideration, check
I could go on for days
________________________________
Eric Berry is better at football than you.
Email sent
Thanks for the offer, Contracts is done with, I have torts on Friday before getting on a plane to St. Louis for the fierce Baylor game
FWIW
Notre Dame vs. Navy Is just a smidgen over lock status.
________________________________
Eric Berry is better at football than you.
Interesting stuff ewmvv6.
If you really have some time and energy one analysis that might seem fun to run on your data set would be to correlate each team’s APD (for lack of a better term) with its opponents absolute difference for that game. In other words, if you play a team that is highly unpredictable, does that throw off your predicted performance?
Right now you are just adding the APD’s together for a team, so the implicit assumption is that two unpredictable teams playing would be more chaotic than one predictable and one unpredictable team. This is trying to test if that is a reasonable assumption. Perhaps a predictable and an unpredictable team is almost as bad as two unpredictable ones…
If so, there are other rules you could consider for combing scores for particular games. You could just flag any games where one of the two teams is highly unpredictable, or you could even try multiplying the two APD scores together instead of adding them (which would suggest an interaction such that two moderately wacky teams will be harder to predict than one very wacky team and one very safe team.
You could try a few different ways of combining them, and see what does better in terms of flagging the games that have high miss rates. Your best case scenario might be if you can identify a set where Bill is waaaay below chance, then you have a reliable indicator that you should actually bet AGAINST Bill on those teams to make money :)
and the Mustache of Truculence (formerly Canada4Mizzou)
And when I say “flag any games where one of the two teams is highly unpredictable”, I mean, you could just have an algorithm which picks whichever APD is higher between two teams, and uses only that one – so the most chaotic team is the rate determining step for how crazy the outcome is. This is the polar opposite of a multiplication combining rule, which would punish games extra hard for including two teams with high uncertainty.
and the Mustache of Truculence (formerly Canada4Mizzou)
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Fighting mob mentality since Oct. 2007
Have you purchased your copy of the 2010 Missouri Football Preview yet?
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2010 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
