ND 29 - Utah 26...

[Easton Pa ND Fan]

Scoring: ND 29.46 - Utah 26.44
Margin: 2.28 points
Win Chances: .5281 with .6649 accuracy.

Used an Excel spreadsheet add-in program to compare a
complete 2009 database to a 2010 (to 11/06) base and
found the 2009 to have better correlated game stats.

So what could be learned from the 2009 base would be
applied to the current ND vs Utah game stats for
better prediction accuracy.

The program examined how 10 game stats relate to 4
"target" stats (pnts/g(2), mov and win %).

The program cycles through the columns building weights
and forming linkages. Each pass builds on previous ones.

When the program can no longer reduce error prediction
for ALL 4 targeted stats, it plugs in the 2010 current
ND-Utah game stats to the underlying table of weights,
linkages and associations to predict the outcome.

 

[NDinMich]

Pretty funny stuff there Easton. I just ran a predictor of my own...

ND 20 - Utah 49

Formula used = Common Sense Quadratic


Easton, am I reading that right that "2009 seasonal data" was used? What would your program yield had you used 2010 seasonal data?

 

[ndcoltsfan2010]

Pretty funny stuff there Easton. I just ran a predictor of my own...

ND 20 - Utah 49

Formula used = Common Sense Quadratic


Easton, am I reading that right that "2009 seasonal data" was used? What would your program yield had you used 2010 seasonal data?

LOL. Hope we don't get beat that bad, but realistically sounds about right. I would love to be wrong about that though.

 

[Legacy]

Some stats to consider:

Seven of the nine games Utah has played have been against teams ranked 94th to 120th in rushing defense.

Six of nine Utah opponents are ranked 61st to 96th in passing defense.

The three games that were close for the Utes were:
Pittsburgh (win by 3 in OT, #8 in rushing defense, #45 in pass def)
Air Force (win by 5, #109 rushing defense, #7 pass def)
TCU (loss by 40, #7 rushing defense, #1 pass def)

 

[Easton Pa ND Fan]

The 2009 seasonal data...

The 2009 seasonal data...

Pretty funny stuff there Easton. I just ran a predictor of my own...
Formula used = Common Sense Quadratic
yield had you used 2010 seasonal data?

was a complete season including playoffs and bowls.
The relationships were stronger (2009) compared
to the partial (2010) season, so I used last years
data to find applicable rules of thumb to apply to
this year's game data.

The numbers in the original post were from 7292 train-
ing cycles. Out of curosity, I stopped the program
after the 1st training pass to see the initial pre-
diction:

Points: ND 14.02 - Utah 43.76
Margin: -25.69 points
Win % chances: .1165 with .0788 accuracy.
Note: None of the 4 target columns were within +/- prediction accuracy after the 1st pass.

 

[Legacy]

I've never really considered using last year's data to predict a late season game the following year. Call me crazy.

I suppose that on first pass Weis's team did not match up well(-29 pts), but with 120 passes, it looked to be the favorite. Or are you implying that a different coach with last year's players would beat Utah?

Sly, Easton.

 

[OlympiaIrishFan]

Golden Tate, Jimmy Claussen, and Kyle R. They are not playing.

Instead we have Rees, Goodman, and Eifert. I don't need math for this. Just common sense.

Love my team, but come on.

 

[MacIrish75]

Using data from the 1988 Irish National Championship team I predict we will win this game 59 to 0.

 

[mgriff]

I love the never say die fans on this site. You guys make it great. I'll be cheering for the boys, but I can't muster enough confidence going into this game. By the product I've seen on the field, it would be a major upset if ND wins. It would be nice to be the team doing the upsetting for once.

 

[CanadianIrish]

Using my broad review of the schools programs from 1950 - 1988 I have determined the Irish will win by 60 points and give up only ten yards of rushing.

What in god's name do last year's stats have to do with anything at this point? Unless Jimmy is going to throw to Golden with Sam Young at left tackle I couldn't care less about last year.

 

[ndcoltsfan2010]

I love the never say die fans on this site. You guys make it great. I'll be cheering for the boys, but I can't muster enough confidence going into this game. By the product I've seen on the field, it would be a major upset if ND wins. It would be nice to be the team doing the upsetting for once.

I hear you man. I also have little confidence, BUT....., What if??? It would be so nice to be the upsetters as you stated for a change. At some point we will get one,, but with our thin roster on so many key positions, I am more than a little concerned. Where is Rudy, Crist, AA and Williams when you need them? We have been destroyed with injuries this year, and it has made this season even all the mroe painful. A win this Saturday would be huge, because if we can win, it would have been done with alot of our backups.

 

[irishff1014]

I will buy in on this. they had an extra week to get ready not to mention watch them live get rolled by TCU. Maybe we can have some of the same success

 

[k1ssme1m1r1sh]

Don't worry, Easton, I hope they win too.

 

[AdmiralBackhand]

I really hope we can beat this good Utah team and USC. I'm not really worried about beating Army. If we win against these two teams I could swallow a loss to Army.

 

[Easton Pa ND Fan]

Ten 2009 season summary game stats...

Ten 2009 season summary game stats...

Using my broad review of the schools programs from 1950 - 1988 I have determined the Irish will win by 60 points and give up only ten yards of rushing.

What in god's name do last year's stats have to do with anything at this point? Unless Jimmy is going to throw to Golden with Sam Young at left tackle I couldn't care less about last year.

were compared to 2009 season pnts/g (2), mov and win %.
The program evolved weights and connections between the
10 data and 4 target columns. Only when all 4 target
columns reached a preset prediction accuracy did training cease.

2010 partial season data did not have as strong a corre-
lation to 2010 targets (4). The program could not ach-
ieve the desired accuracy in all 4 columns.

Unless there is a radical change in play philosophy from
last year, the underlying formulas and equations apply
to this season. The program predicted from 10 2010 ND
and Utah game stats.

Neuralyst

 

[kmoose]

Look Folks, chill out on Easton. He's been doing stastical analyses that boggle my simple mind, for as long as I have been on this site. I don't think he is trying to build up unrealistic expectations, I think he is just trying to stimulate discussion. For some of you new(er) guys, I would just say that Easton is a respected member of our little community here. I doubt anyone will take offense to you respectfully pointing out places where you might disagree with his logic, but getting sarcastic and/or dismissive with him is not a way for you to earn the respect of those who have been around a while.

 

[goldnblew]

Gotta admit, really funny stuff on this topic

BLehane
ND 20 - Utah 49
Formula used = Common Sense Quadratic

BLehane, great stuff. Sounds like it's been a while since Easton's been on a date

 

[Old Man Mike]

Tom, I love your stuff even if some of the cheap-shot artists don't. The stats are too complicated for me this time, so I'll hold them as a ray of sunshine, just as I do the fact that TCU exposed Utah rather badly last week, and maybe left them limping. It's not unheard of in NCAA football that apparently good football teams turn out to be illusory bubbles which burst. Maybe Utah is one case.

 

[irishtrain]

Pretty funny stuff there Easton. I just ran a predictor of my own...

ND 20 - Utah 49

Formula used = Common Sense Quadratic


Easton, am I reading that right that "2009 seasonal data" was used? What would your program yield had you used 2010 seasonal data?

BLehane, everytime from now on when I make the point that Notre Dame needs players and its not scheme,coach,jumbotron,campus restrictions,cold weather,old people not cheering,or turf may I use your Formula= 'Common Sense Quadratic'. Its easier and says more than I could post in 2 weeks. Yep the 'Common Sense Quadratic' just about says it all. Its about time the'Common Sense Quadratic' was put to use by the people running Notre Dame.

 

[irishmarine]

The only thing this team ha in connecetion to last years team is that they are both playing like ****. I'll still be in the couch screaming at the tv though and scaring my son

 

[choo choo]

I really hope we can beat this good Utah team and USC. I'm not really worried about beating Army. If we win against these two teams I could swallow a loss to Army.

better get ready to spit or swallow

 

[fitz_bu47]

I understand the pessimism, I do, we are coming off losses to two teams that are not as good as us on paper, and are playing one that is better on said paper. I remain optimistic b/c we are coming off a bye, they are coming off a beatdown, and I know it was a long time ago, but I watched their game w/ Pitt, and Pitt should have won at their place. I think they are a pretty good team, but not great, and very beatable coming off that thumping from TCU. I think we have a great staff, and some kids who need something to rally around, and maybe w/ the losses/tragedy, and a bye week we pull it off this weekend. Definitely not crazy, and like KMoose said, I have been reading Easton's breakdowns for a long time, so take it eassssssssyyyyyy on him, 4 post dudes.

 

[NDinMich]

BLehane, everytime from now on when I make the point that Notre Dame needs players and its not scheme,coach,jumbotron,campus restrictions,cold weather,old people not cheering,or turf may I use your Formula= 'Common Sense Quadratic'. Its easier and says more than I could post in 2 weeks. Yep the 'Common Sense Quadratic' just about says it all. Its about time the'Common Sense Quadratic' was put to use by the people running Notre Dame.

You can absolutely use the Common Sense Quadratic! I've found that the Common Sense Quadratic has many uses, even outside of football. In my younger years I would use the Common Sense Quadratic in my personal life.

Out at a campus bar on a Saturday night - Hmmmm, the Jack Daniels in my system is telling me to order a round of shots for 20 ppl, eat 4 conies & fries before going home on the public transit bus...let me check with the common sense quadratic. Result = indigestion, no money for tuition and possible mugging on the bus, THANKS CSQ!

Sorry Easton, didn't mean to lead the masses against you. I love the statistical analysis angle of ND football, all else has failed thus far.

 

[NDinMich]

I really hope we can beat this good Utah team and USC. I'm not really worried about beating Army. If we win against these two teams I could swallow a loss to Army.

If ND beats Utah, loses to Army, then beats USC, while happy, we may be left with more questions than answers.

 

[irishtrain]

You can absolutely use the Common Sense Quadratic! I've found that the Common Sense Quadratic has many uses, even outside of football. In my younger years I would use the Common Sense Quadratic in my personal life.

Out at a campus bar on a Saturday night - Hmmmm, the Jack Daniels in my system is telling me to order a round of shots for 20 ppl, eat 4 conies & fries before going home on the public transit bus...let me check with the common sense quadratic. Result = indigestion, no money for tuition and possible mugging on the bus, THANKS CSQ!

Sorry Easton, didn't mean to lead the masses against you. I love the statistical analysis angle of ND football, all else has failed thus far.

The Common Sense Quadratic it is. All respect to the CSQ.

 

[Irish Insanity]

I'm saying ND 35 UTAH 32 based on hope. But realistically ND 24 UTAH 28. All I know is if they double us up by halftime I'm streaking while the bands perform!

 

[choo choo]

Scoring: ND 29.46 - Utah 26.44
Margin: 2.28 points
Win Chances: .5281 with .6649 accuracy.

Used an Excel spreadsheet add-in program with 10 columns
of 2009 seasonal game data vs 4 target columns of season-
al offensive pnts/g, defensive pnts/g, margin of victory
and win percentages. Each row represented 1 of 120 teams.

The program cycles through the columns building weights
and forming linkages. Each pass builds on previous ones.

When the program can predict each of the 4 target columns
within a certain range, it computes using the data on
rows 121-2, which are the current ND and Ut seasonal data.

I sooooo hope your right

 

[DillonHall]

I'm saying ND 35 UTAH 32 based on hope. But realistically ND 24 UTAH 28. All I know is if they double us up by halftime I'm streaking while the bands perform!

Promise?

 

[irishff1014]

They only give up about 300 yards of offense a game so we are going to have to work hard for this. But i think with 2 weeks to prepare for them we should have some good plans on offense. We need our defense to work hard and not give up the big play and we will do ok.

 

[Easton Pa ND Fan]

I hope there is a Utah bubble...

I hope there is a Utah bubble...

Tom, I love your stuff even if some of the cheap-shot artists don't. The stats are too complicated for me this time, so I'll hold them as a ray of sunshine, just as I do the fact that TCU exposed Utah rather badly last week, and maybe left them limping. It's not unheard of in NCAA football that apparently good football teams turn out to be illusory bubbles which burst. Maybe Utah is one case.

A complete season (2009 with playoffs and bowls) was
compared to the 2010 partial season game stats. The
2009 game stats showed much stronger correlations to
winning %, points(2) and margin*. This database was
used to identify trends, tendencies, rules of thumb etc.
and by plugging in the most current ND-Utah game statis-
tics, more credible predictions could be made.

Each team's 2009 game stats were on 1 row by 14 columns.
The last 4 columns were the targets*. 120 team-rows were
used.

A program examined each row, computed weights and connect-
ions, etc. to match the last 4 "target" columns and updat-
ed a relational database table.

The neat thing about this "learning" program was that
after each 120 row pass, some data is changed ("noise")
and the program must adjust on the next pass. Why is this
done? The rules can become too specific to 2009 and not be
universally applied to 2010.

After 7292 training pass-cycles, the program could not
reduce the prediction error for all 4 targets (Pnts/g(2),
mov and win %). Training ceased.

The underlying weights and associations (in effect general-
ized math tables) were saved.

As long as the game does not change radically from year to
year, this underlying football math logic could be used.

Added 2 additional rows with current ND & Utah game data
and let the computer compute the above estimates.