Alright, football fans. The 2017 season is finally here and it's time for some wild speculation and overanalysis of useless information and exposure of the universal certainty that preseason predictions mean diddly.
Last year, I kept up with ESPN's FPI projections of Virgnia Tech's schedule on a week by week basis to see how the model would change as the perceptions of the Hokies and their opponents changed. While there is no perfect formula for predicting outcomes (cause who wants to let the players decide, right?) and a single source of information is looking at the situation narrowly, it's fun to see the trends anyway. This season I've decided to expand my data collecting expedition as follows. There will be three season-long, week-by-week trends coming from ESPN's FPI predictions, Massey Ratings Matchup simulator, and CFB Analytics predicitons. FPI and Massey are nice enough to give a percent chance of victory for each team, but CFBA will only give an expected spread.
I'll do the update format a little bit differently, too. Rather than a comment with the updated charts, I'll rehost the pictures and maybe add a line of commentary to the body of the forum. The Week on the x-axis will be the data after that week of games. Week 1 was plotted as the same as Week 0 so that it will show up instead of a bunch of dots.
ESPN FPI
Week 12: 8.8 wins from the FPI, steady after the GT letdown and the near Pitt-fall. Interesting that FPI has the UVA game at a better Tech chance than preseason.
Massey Predicitons
Week 12: Massey is at 8.9 expected wins and, as was seen throughout the season, the UVA game has had a relatively stable curve from week to week.
CFB Analytics
Week 12: A cool 7.8 point spread in favor of the Hokies this weekend. What I found interesting about the GT spread was that it was locked to 26.8-26.8 pregame. Tech lost 28-22 while leaving 5 easy points on the board. Pretty eerie there.
Bill Connely - S&P+
Week 12: 8.6 wins from Bill, and he's had the UVA game trending down all season long. I'm a little scared here.
Discuss amongst yourselves. Or don't.
Comments
Love following the FPI throughout the year. It's just fun to see the national implications of what the Hokies are doing.
P.S. Did you know Massey is a Hokie himself who came up with the FPI?
I didn't know, but now I do.
Thanks for the fun fact!
I'd say between Kenneth Massey and KenPom, hokies are dominating the power ratings game.
I had him as a professor for a math class while I was at Tech. Nice guy but ZERO social skills at all. I mean he was painfully awkward talking to about anything after class.
I recall Massey was a student at VT when he first came up with that rating system.
Or am I mistaken?
You're correct! Well, partially - he came up with the algorithm in undergrad at a different school, and then developed it and was invited to submit it to be included in the BCS calculations when he was doing graduate work at VT. He was actually featured during a game back in the early '00s I think, that's where I learned he was from VT.
Any way you could add symbols to the lines/legend to help show the difference between like colors?
Okay good, I'm not the only one. I could distinguish 3 colors there blue, black, and orange... just assumed which line was which and looks like we've got a 60% chance to win the ACC
I'll chage it up for next week. I linked each school with their official RGB codes online, but a lot of the blues and golds get lost across the many varieties.
Is there a way to customize the Excel legend symbols to be a pic of the mascot? That'd be fun.
I'll figure something out.
It would be faster just to take a snippy of the graph and drop it into photoshop. I can do it for you if you want
Is that a little more pleasing to the eyes?
Certainly easier to tell what is what.
Wait, am I reading this right and the predicted ODU spread is around 5 points?
huh?
ever played darts blindfolded? Apparently the guys who put together these spreads have.
It's gonna get crowded again really soon when all the projections are crammed between 95 and 99 percent win!
Its going to be fun watching that Miami line correct itself over the course of the season. The fact that ESPN gives them roughly the same odds to beat us as Clemson is comical.
Also, what this is telling me is that our realistic darkest timeline is a 6-6 season. Our most realistic best timeline is a 11-1 season. Most likely, we're split the difference on those middle games and end up with 9 wins, with a chance for 10 in the bowl.
Miami gets FSU really early this season too, so they'll be plenty derailed after the annual loss by the time we get to them.
I wonder how October 7th will change the derailing
74%, 86%, and 9 point favorite versus loluva?! I only partially agree with one of these.
Man, ESPN is all-in on Miami this year
2017 bold predictions: Look out for Urban Meyer's revenge in 2017
Everything you need to know about the Power 5 before they play a game
And for the record, the Hokies are completely ignored in both articles. For those who relish the Hokies playing the underdog role every year, you have your wish this year.
Jesus christ on a biscuit. I have never seen them so outright call a championship like that when they were singing our praises prior to our Juniors leaving for the draft.
This is more than just being the underdog. The way I look at it it's a slap in the face to all those guys who worked so hard CJF on down to not even be in the conversation this year. I hope they use that as fuel to crush the Coastal, and beat Da U 35-0.
nah #38-0bro2.0 would be better
For those that don't know, 31-7 is the traditional score for the VT victory over Miami.
These guys don't know anything. Miami has been in the ACC championship game in August evey year since they joined the league.
Obviously what's happening is that some 'canes fan keeps stepping on sidewalk cracks, you know the old saying
"Step on a crack, break Miami's bak"
This must go green...
Duke and Wake Forest have played in more ACCCG since conference realignment...but Miami is baaack!
FTFY
I had heard that they're bak so this checks out /s.
Cool stuff, always nice to see how the ratings zig and zag against each other, even if FPI is absurdly high on Miami (and I'm usually the one defending the canes' high ranking). One suggestion, you're setting 50% as the win-loss line then projecting our record from that. I think it would be much more useful to count each game as a percentage of a win. So a 90% win expectancy against ECU translates to .9 wins and a 25% expectancy against Clemson is .25 wins and a game against UVA is 1 win.
Also, S&P+ is a good resource as well with both win expediencies and spreads available.
The weighted average gives an FPI projection of 7.9 wins and Massey 7.1 due to the FPI being slightly more polarized and Massey having a grouping toward the middle. 8-4 is possible, with 7-5 probably the floor for this season. 9-3 is certainly doable, but I would be really surprised at 10-2. Looks like the weighted average puts it at a fairly conservative real life scenario.
Does the S&P do specific matchups? All I can find is the list of all 130 team rankings.
Here's the S&P+ VT preview page. It has win probabilities for each game. The page that will be updated throughout the year isn't up yet. We're projected at 8.1 wins.
The good thing is week 1 tells us a lot. We win that game, all the models should have us at 9 or 10 wins easy.
We're currently slotted a little conservatively just because Jackson hasn't proven himself, yet. If he's as good as advertised, you should see our predictions jump a little. I very much expect us to be favored in all but the Clemson game this year, with Miami being a tossup.
Galloping gargoyles, that's thorough. Might as well add a fourth graph to the basket.
this is cool, nice graphical representation...gives me a better idea of where some of our less familiar opponents are at(or projected to be)
To determine Ws & Ls, wouldn't it be more accurate to add up the win percentages to take into account the likelihood of the win?
So, for ESPN FPI, VT is predicted to have a 7.9-4.1 record. For Massey, 8.0-4.0
And you can use this chart to determine win probabilities for spreads
And this is different from most years?
smart dog
dart smog
Week 1 data added. No spread updates yet from cfbanalytics.com yet, though.
That GT fall in the ESPN FPI... Wanna say it's a crock of sh!t, but Marshall is a Dude. GT is going to be better than we projected.
ECU odds go down? Only thing I can think of is that they figure that'd be some shit Tech would pull. Anything can happen...but man I hope those days are behind us.
These models are obviously wrong, only 75% chance of beating loluva?!
Week 2 data updated. Apologies for the delay, would've had it up Monday but I just got power restored at 3 PM Tuesday.
So is Duke really that good? I thought they were fading back into obscurity, especially after Thomas Sirk left them.
Combined score of 101-24 across 2 games.
They've got the 2nd best QB in the conference and he's only a sophomore
Yeah he's the real deal.
Funny you asked. I was thinking of doing a little write-up on Dook, but I don't want to muddle the ECew hate and Dook is six games away. Dook looks like they've already arrived where the Hokies offense wants to be.
They're successfully rotating RBs and WRs so that at least 5 people carried the ball in each of the first two games, and 10 people caught passes (8 runners and 13 receivers in their FCS game). They've had over 230 yards rushing and 280 yards receiving in both of their games. And they're keeping the tempo fast.
Their 2nd year QB is playing well and getting the ball to a lot of different skill guys, although he does have an interception.
When the Hokies put everything together they'll look a lot like Dook.
I...have no words.
I know, I know, if you weren't sure if we were living in the
evilinverted timeline or not, now you know for sureSo the GT game looms very large! Most years, this one seems to be a toss up! We beat Miami, well, because they are Miami! Clemson looks to have an elite defense. Not the best matchup for us, but maybe they pull a klemson and turn it over a few times to our DB's! Go Hokies, looking like another fun year.
All of a sudden I am fearing the Duke game. We can't lose to Duke! Good thing I will be there screaming until it hurts.
Honestly, yes losing to Duke would be a bummer, but any loss is. I would much rather lose to a team with a head coach I respect the hell out of and a staff that does things right than lose to someone like Nardouche or UNCheat.
With that being said I hope we laugh them off the field, but I have come to learn to be nervous when we play Duke.
I agree with everything you said. I'd much rather lose to Duke in a good year than Pitt in any year, but to me, Duke is still a basketball school and a football team with no reputation.
Coastal Chaos beat us, not Duke.
First off, these are great illustrations!
The VT Advanced Statistical Profile at https://www.footballstudyhall.com/pages/2017-virginia-tech-advanced-stat... updates each week including the win probability against each opponent. Alternatively, the S&P+ ratings on Football Outsiders are standardized to points, so you can get a neutral field prediction by just taking the ratings difference between any two teams. Home field advantage is then worth about 3 points.
The Virginia Tech UVA game hangs in the balance tonight actually
Week 3 is up. With more data this time!
Is there a week missing for the spreads or am I just ignant?
There's a week missing.
After the WVU game, there was a long delay for CFBAnalytics to update their data. It was probably done some time over the weekend of the Delaware game, but I was in Blacksburg sans-computer. By the time I got home Sunday, everything through Week 2 had been updated. So the data is a little wonky and Excel is giving me some lip about leaving out Week 1 on the x-axis.
I call shenanigans. No way the LOLUVa game is closer than the ODU game.
Here is a plug for "Podcast Ain't Played Nobody" Bill Connelly and Steven Godfrey's weekly podcast. They explain a bit better the S&P+ data and actually talk about Miami's skewed data. Basically because Miami has only played 1 game (which they dominated) 90% of their data is based on preseason projections.
Data good through Week 4.
The computers are all aboard the Hokie Hype Train. Some general consensuses (that's really a word?):
One thing I noticed in the rankings of teams is that FSU is still ranked above us and a few other undefeated teams even though their expected win percentage is around 6.5.
Week 5 post-Clemson data updated.
Here's some TKPC special analysis: all of the computers believe that the Hokies have major tests remaining against Georgia Tech and Miami, with the rest of the ACC games being likely Hokie victories. You heard it here first.
But really, the computers are still pretty high on Tech between 8.5 and 9 wins. That seems reasonable for Fuente's second year ACC Championship berth.
I'm so torn. On the one hand, I want Miami to get rolled by FSU so people realize that they aren't really bak. On the other hand, there's no way FSU stays home this bowl season unless they lose to Miami tomorrow.
.....and an undefeated Miami vs a 1-loss VT woukd be a lot of fun in November
that would stress me out to no end. I kind of like going into games like that knowing they need to beat us AND have us drop another ACC game.
We all know the u is not bak. GT game looking big.
Data through the BYE week is up. Sorry I missed last week's update. It's been a remarkably busy fall and finally got around to posting. The data for Week 6 should be accurate, there was just no blurb about it last weekend.
Overall trends remain the same as so far. GT and Miami will be tough tests, UVA is a surprising matchup for Black Friday. The rest of the Coastal sucks. Tech is seen as a 9.0 win team with an outside shot at getting back to Charlotte.
Quick side note on the Techmo Bowl. If my memories are correct, the last time the home team won that game was in 2012 in Lane on Labor Day??
2013 Thurs VicTory? @lanta??
2014 GT surprise win (M. Brewer fumble TD) and that one pass that the Jackets threw
2015 Thurs VicTory @lanta.
2016 won't talk about that one
2017 @lanta, we'll see what happens
Week 8 data complete.
Mostly pretty big jumps, had to extend the y-axis on the spread graph thanks to the LPD last weekend. FPI and Massey have Tech at 9.6 wins with Massey at 9.3. ACC Champ chances more than doubled up to 6.6 percent.
Virginia finally got their earth crashing correction after their demolition at the hands of BC. Miami and GT are trending toward complete toss-up games.
Interesting the metrics have our chances against Miami so low, while Vegas has the game basically at pick em right now.
Good Lord that drop off for the UNC line on the CFB analytics graph is just glorious to look at 😳
So our probability of beating Duke, Pitt, and UVA are all about where our chances of beating UNC were last week? Does that mean that we are likely to favored by 21? Or 45? I don't think we beat Duke by that much.
Week 9 up to date.
FPI and S&P are sitting at 9.6 wins, Massey at 10.0 wins, and the CFBA spreads have 11 projected wins with Miami and GT being super close.
Obviously, the next two weeks should define the rest of the season and see if Tech can sneak itself into the CFP discussion for real. It's a little interesting that the three percentage-givers have the GT and Pitt games fairly consistent (58/62/63 and 90/84/87) while the Miami and UVA games vary wildly (46/58/44 and 89/93/69).
The lines haven't danced around quite as much as they did in 2016 for two reasons. 1. Tech has been able to avoid the traditional letdown loss in the middle of the season and 2. The expectations were more-or-less elevated in Fuente's second year and the Hokies have answered the bell (see reason 1). The lack of huge jumps and wild graphs doesn't create as much of a discussion, but I still think it's fun to see how these perceptions change, both for Tech and against the opponents.
I am still amazed about us what the lucky U win over GT does for the numbers. If they had not pulled that out of their ass, where would the numbers be? I am more concerned over GT than da U.
Week 10 updated.
After the Category 5 shitstorm last weekend, every percentage or spread became less in Tech's favor (surprised?). We should still expect somewhere between 9 and 10 wins, and the Georgia Tech game is expected to end in a tie according to CFB Analytics. For those of you watching from home this weekend, look for me on TV. I'll be the one in the hat.
Sorry, guys it's been a while. And it's not like the Hokies have played football in about a month or so, right? RIGHT???
Anyway, I hated putting all those zeroes up on the spreadsheet so I let it sit on my desktop to marinade for a while. The data is updated for both the GT and Pitt games leading up to the Black Eye Friday for the Hoos. The computers are all sitting around 8.5-9.0 wins with varying takes on the UVA game. The most consistent (as we've seen all year) is Massey with S&P giving the Hoos a solid chance.
Looking back at the spreads, the GT game was intriguing. The spread was locked at 26.8 points a piece, and VT left 5 easy points on the field and lost 28-22. Good guess, CFBA.
Hope y'all have enjoyed the graphs this year. Maybe not as exciting as last year given the lackluster finish and slightly different expectations.