[slide]

[quote author=slide link=topic=10789.msg101294#msg101294 date=1412718921]cfbstats is the source

That clarifies nothing. YPG? PPG? YPP? There are a bunch of different metrics one could use.

Regardless, I think any ranking that ignores the strength of opposing defenses is beyond pointless and into actively misleading.
[/quote]

ypg from cfbstats

if you want to look at FEIoff, then #6 to #34 to #22 to #93 (this year throwing out Davis as FEI only uses FBS teams).  #93 in FEI, #96 in cfbstats -- meh

[slide]

[quote author=slide link=topic=10789.msg101294#msg101294 date=1412718921]cfbstats is the source

That clarifies nothing. YPG? PPG? YPP? There are a bunch of different metrics one could use.

Regardless, I think any ranking that ignores the strength of opposing defenses is beyond pointless and into actively misleading.
[/quote]

ypg from cfbstats

if you want to look at FEIoff, then #6 to #34 to #22 to #93 (this year throwing out Davis as FEI only uses FBS teams).  #93 in FEI, #96 in cfbstats -- meh

[slide]

[quote author=slide link=topic=10789.msg101262#msg101262 date=1412711769]

BC -- much of what you write may be true, but the offensive trend is not good nor has it been since Luck's departure. 

Um, yeah. Now think about what you just wrote. Then think about it again. Then, go read the cover story of the Sept 8 Sports Illustrated. Then think about what you just wrote.

If after this exercise, you don't think anything's wrong with your statement, if you don't audibly say, "Well, duh!" you're just not thinking straight.

smdh
[/quote]

I knew our offense would regress after 2011, but for the regression to continue 3 years after Luck's departure?  from #8 to #96 (or #6 to #93 under FEI)?  going back to BC's point that the staff is adjusting and this is the output of the adjustment? 

[slide]

[quote author=slide link=topic=10789.msg101262#msg101262 date=1412711769]

BC -- much of what you write may be true, but the offensive trend is not good nor has it been since Luck's departure. 

Um, yeah. Now think about what you just wrote. Then think about it again. Then, go read the cover story of the Sept 8 Sports Illustrated. Then think about what you just wrote.

If after this exercise, you don't think anything's wrong with your statement, if you don't audibly say, "Well, duh!" you're just not thinking straight.

smdh
[/quote]

I knew our offense would regress after 2011, but for the regression to continue 3 years after Luck's departure?  from #8 to #96 (or #6 to #93 under FEI)?  going back to BC's point that the staff is adjusting and this is the output of the adjustment? 

[stupac2]

I knew our offense would regress after 2011, but for the regression to continue 3 years after Luck's departure?  from #8 to #96 (or #6 to #93 under FEI)?  going back to BC's point that the staff is adjusting and this is the output of the adjustment?

It's not "continuing to regress", though. The trend was down in 2012, up in 2013, then way, way down through the first part of 2014.

Is the current offense acceptable? No, though I think you could argue that it's actually been better so far than it was in 2012 (in that we actually scored offensive TDs in our two losses). Is it part of some horrible trend since Luck's departure? I think that's an exceedingly difficult case to make, and either way with only 3 data points it's sort of silly to even try.

[stupac2]

I knew our offense would regress after 2011, but for the regression to continue 3 years after Luck's departure?  from #8 to #96 (or #6 to #93 under FEI)?  going back to BC's point that the staff is adjusting and this is the output of the adjustment?

It's not "continuing to regress", though. The trend was down in 2012, up in 2013, then way, way down through the first part of 2014.

Is the current offense acceptable? No, though I think you could argue that it's actually been better so far than it was in 2012 (in that we actually scored offensive TDs in our two losses). Is it part of some horrible trend since Luck's departure? I think that's an exceedingly difficult case to make, and either way with only 3 data points it's sort of silly to even try.

[BostonCard]

Just to put what Yvonne wrote into numbers, I took the offensive stats from cfbstats.com for all the years they had data (2008 through the ND game in 2014), and calculated points, yards, and first downs per possession.  I like per possession stats because you don't have to worry about the pace of play (a team like Oregon that scores quickly will get more possessions than a team like Stanford and would be expected to get more yards and more points, even if the offenses are equally efficient at turning possessions into scores.  The system is not perfect, and doesn't account for the strength of the defenses faced, but I find the stats are very useful.

Since the data goes back to 2008, we can see the effect of having Andrew Luck in the backfield (three years, from 2009 - 2011) and the effect of having both Shaw (2011 to 2014) and Harbaugh (2008 - 2010) as head coaches.  Most importantly, each head coach had an opportunity to coach Luck and one other QB (Pritchard for Harbaugh, and Hogan for Shaw), so it really gives us a chance to disentangle the effect of Andrew Luck from the effect of the coach. To be fair, I think Hogan is a better QB than Pritchard was, though I think that is balanced out by the fact that Harbaugh had Gerhart as running back, and this year, Shaw does not have a running back of the caliber of Toby. Note, I listed Hogan as the QB for 2012, even though Nunes started a number of the games.



Looking at the raw sats, it is clear that Andrew Luck was a difference maker; his three years we averaged 3.6 points per possession (a touchdown every other possession), 2.1 first downs per possession, and 42 yards per possession.  In the other years, we averaged 2.4 points per possession (PPP), 1.6 first downs per possession (FDPP), and 31 yards per possession (YPP).  In other words, Andrew Luck was responsible for a 47% increase in PPP,  31% increase in FDPP, and 30% increase in YPP.

Looking at coaches, having Harbaugh as a coach was worth 0.3 PPP (3.07 PPP for Harbaugh vs. 2.80 PPP for Shaw), 0.18 FDPP (1.93 vs. 1.75), and 2.5 YPP (37.6 vs. 35.1).  However, Harbaugh had Luck for 2 out of his three years here, while Shaw had Luck only 1 year.  If you do a multiple linear regression (sorry to be geeking out here), you get the following.

As for the question of whether the "trend has been worse since Luck left", we certainly took a big hit with Luck's departure.  But 2012, 2013, and 2014 look pretty similar to me.  Points per possession is down this year; probably a reflection of our red zone troubles early in the year (note, I think we were perfect in the red zone against Notre Dame (our botched field goal was from the 25; the problem was we didn't even make it to the red zone against ND).  First downs per possession and yards per possession are between 2012 and 2014.


lm(formula = PPD ~ Shaw + Luck, data = o)
Residuals:
      1        2        3        4        5        6        7
-0.35765  0.45235 -0.14765  0.05294  0.22353 -0.27647  0.05294

Coefficients:
            Estimate Std. Error t value Pr(>|t|) 
(Intercept)  2.2171    0.2805  7.904  0.00139 **
Shaw          0.1706    0.2930  0.582  0.59161 
Luck          1.2794    0.2930  4.367  0.01199 *
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.3487 on 4 degrees of freedom
Multiple R-squared:  0.8395, Adjusted R-squared:  0.7593
F-statistic: 10.46 on 2 and 4 DF,  p-value: 0.02575

Andrew Luck is worth 1.28 PPP (statistically significant); there is no statistical significance between Harbaugh and Shaw, though, in the model, Shaw actually gets 0.2 PPG more than Harbaugh.  I won't throw the regression code up for FDPP (I can send it if anyone wants), but Luck is worth 0.51 FDPP with a p-value of < 0.01; there is no significant difference between Harbaugh and Shaw, and Luck is worth 10.4 YPP (p = 0.01), and again, a non-significant difference between Shaw and Harbaugh.

Obviously there are a lot of caveats here; while I adjust for Andrew Luck, there are a lot of other players and coaches involved, and it is certainly possible that Andrew Luck aside, Shaw might have had more talent at his disposal his three years than Harbaugh did.  Harbaugh did have Luck his redshirt freshman year, where he was somewhat less effective than his other two years, and you might argue that an offense with Luck as a redshirt junior should have accomplished more than an offense with Luck as a redshirt sophomore (the offenses the two years were pretty similar in efficiency).  Finally, none of this takes into account external factors such as the quality of the defense we were facing, how much of the game was uncompetitive (when we pulled starters and/or went with a conservative play calling), or weather, which was certainly a factor Saturday.

BC

[BostonCard]

Just to put what Yvonne wrote into numbers, I took the offensive stats from cfbstats.com for all the years they had data (2008 through the ND game in 2014), and calculated points, yards, and first downs per possession.  I like per possession stats because you don't have to worry about the pace of play (a team like Oregon that scores quickly will get more possessions than a team like Stanford and would be expected to get more yards and more points, even if the offenses are equally efficient at turning possessions into scores.  The system is not perfect, and doesn't account for the strength of the defenses faced, but I find the stats are very useful.

Since the data goes back to 2008, we can see the effect of having Andrew Luck in the backfield (three years, from 2009 - 2011) and the effect of having both Shaw (2011 to 2014) and Harbaugh (2008 - 2010) as head coaches.  Most importantly, each head coach had an opportunity to coach Luck and one other QB (Pritchard for Harbaugh, and Hogan for Shaw), so it really gives us a chance to disentangle the effect of Andrew Luck from the effect of the coach. To be fair, I think Hogan is a better QB than Pritchard was, though I think that is balanced out by the fact that Harbaugh had Gerhart as running back, and this year, Shaw does not have a running back of the caliber of Toby. Note, I listed Hogan as the QB for 2012, even though Nunes started a number of the games.



Looking at the raw sats, it is clear that Andrew Luck was a difference maker; his three years we averaged 3.6 points per possession (a touchdown every other possession), 2.1 first downs per possession, and 42 yards per possession.  In the other years, we averaged 2.4 points per possession (PPP), 1.6 first downs per possession (FDPP), and 31 yards per possession (YPP).  In other words, Andrew Luck was responsible for a 47% increase in PPP,  31% increase in FDPP, and 30% increase in YPP.

Looking at coaches, having Harbaugh as a coach was worth 0.3 PPP (3.07 PPP for Harbaugh vs. 2.80 PPP for Shaw), 0.18 FDPP (1.93 vs. 1.75), and 2.5 YPP (37.6 vs. 35.1).  However, Harbaugh had Luck for 2 out of his three years here, while Shaw had Luck only 1 year.  If you do a multiple linear regression (sorry to be geeking out here), you get the following.

As for the question of whether the "trend has been worse since Luck left", we certainly took a big hit with Luck's departure.  But 2012, 2013, and 2014 look pretty similar to me.  Points per possession is down this year; probably a reflection of our red zone troubles early in the year (note, I think we were perfect in the red zone against Notre Dame (our botched field goal was from the 25; the problem was we didn't even make it to the red zone against ND).  First downs per possession and yards per possession are between 2012 and 2014.


lm(formula = PPD ~ Shaw + Luck, data = o)
Residuals:
      1        2        3        4        5        6        7
-0.35765  0.45235 -0.14765  0.05294  0.22353 -0.27647  0.05294

Coefficients:
            Estimate Std. Error t value Pr(>|t|) 
(Intercept)  2.2171    0.2805  7.904  0.00139 **
Shaw          0.1706    0.2930  0.582  0.59161 
Luck          1.2794    0.2930  4.367  0.01199 *
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.3487 on 4 degrees of freedom
Multiple R-squared:  0.8395, Adjusted R-squared:  0.7593
F-statistic: 10.46 on 2 and 4 DF,  p-value: 0.02575

Andrew Luck is worth 1.28 PPP (statistically significant); there is no statistical significance between Harbaugh and Shaw, though, in the model, Shaw actually gets 0.2 PPG more than Harbaugh.  I won't throw the regression code up for FDPP (I can send it if anyone wants), but Luck is worth 0.51 FDPP with a p-value of < 0.01; there is no significant difference between Harbaugh and Shaw, and Luck is worth 10.4 YPP (p = 0.01), and again, a non-significant difference between Shaw and Harbaugh.

Obviously there are a lot of caveats here; while I adjust for Andrew Luck, there are a lot of other players and coaches involved, and it is certainly possible that Andrew Luck aside, Shaw might have had more talent at his disposal his three years than Harbaugh did.  Harbaugh did have Luck his redshirt freshman year, where he was somewhat less effective than his other two years, and you might argue that an offense with Luck as a redshirt junior should have accomplished more than an offense with Luck as a redshirt sophomore (the offenses the two years were pretty similar in efficiency).  Finally, none of this takes into account external factors such as the quality of the defense we were facing, how much of the game was uncompetitive (when we pulled starters and/or went with a conservative play calling), or weather, which was certainly a factor Saturday.

BC

[yvonne]

For folks who are familiar with the Myers Briggs Type Indicators, what you've just seen demonstrated is the way NFs process compared to STs!

[yvonne]

For folks who are familiar with the Myers Briggs Type Indicators, what you've just seen demonstrated is the way NFs process compared to STs!

[Rally]

For folks who are familiar with the Myers Briggs Type Indicators, what you've just seen demonstrated is the way NFs process compared to STs!

I have no idea what either you or BC are talking about, and count myself a better human being for that.  Is there anymore ice cream left?

[Rally]

For folks who are familiar with the Myers Briggs Type Indicators, what you've just seen demonstrated is the way NFs process compared to STs!

I have no idea what either you or BC are talking about, and count myself a better human being for that.  Is there anymore ice cream left?

[CTcard]

Just to put what Yvonne wrote into numbers, I took the offensive stats from cfbstats.com for all the years they had data (2008 through the ND game in 2014), and calculated points, yards, and first downs per possession. 

I also like per possession numbers for many purposes, for precisely the reasons you note. The usual downside is that these numbers are not generally available. So one thing you have to do is put in a lot of effort to calculate things (thanks). The other problem is that it's not easy to compare across teams without doing that much work for 128 teams. Ouch.

But - cfbstats does not have possessions as far as I can tell. How did you get them?
Adding FG attempts plus TDs plus punts plus turnovers plus failed fourth downs?
Or opponents' kickoffs plus punts plus turnovers plus failed fourth down attempts?
Or maybe I just missed a key column?

And as you are quite aware, you have to be very careful aboout comparing a whole season's stats to this year's partial season. My guess is that we have played two, and maybe three, of the best four defenses we'll face this year. In addition, things develop during the year - every year - so the current snapshot may not be an average.

EDIT - a minor edit to your discussion.

Harbaugh did have Luck his redshirt freshman year, where he was somewhat less effective than his other two years,

Of course Harbaugh had Luck for two years, 2009 and 2010.

[CTcard]

Just to put what Yvonne wrote into numbers, I took the offensive stats from cfbstats.com for all the years they had data (2008 through the ND game in 2014), and calculated points, yards, and first downs per possession. 

I also like per possession numbers for many purposes, for precisely the reasons you note. The usual downside is that these numbers are not generally available. So one thing you have to do is put in a lot of effort to calculate things (thanks). The other problem is that it's not easy to compare across teams without doing that much work for 128 teams. Ouch.

But - cfbstats does not have possessions as far as I can tell. How did you get them?
Adding FG attempts plus TDs plus punts plus turnovers plus failed fourth downs?
Or opponents' kickoffs plus punts plus turnovers plus failed fourth down attempts?
Or maybe I just missed a key column?

And as you are quite aware, you have to be very careful aboout comparing a whole season's stats to this year's partial season. My guess is that we have played two, and maybe three, of the best four defenses we'll face this year. In addition, things develop during the year - every year - so the current snapshot may not be an average.

EDIT - a minor edit to your discussion.

Harbaugh did have Luck his redshirt freshman year, where he was somewhat less effective than his other two years,

Of course Harbaugh had Luck for two years, 2009 and 2010.

[BostonCard]

Sometimes a picture is worth 1000 words:



BC

[BostonCard]

Sometimes a picture is worth 1000 words:



BC

[BostonCard]

Yes, I had to calculate possessions manually.  Easy to do for a single team, but not easy to do for all teams.

TD's + FG's + missed FG's + Failed Fourth Down conversions + INT's + Fumbles + Punts

That does mean that possessions when the clock ran out at the end of the half or the game don't count, and touchdowns scored on kickoff returns and punt returns are credited to the offense (and count as "possessions"), though fumbles on special teams also count as possessions in the denominator.

The caveats about a portion of the season being played are well taken.  This is really meant to give a rough picture of where we are.  We really miss Andrew Luck.

BC

[BostonCard]

Yes, I had to calculate possessions manually.  Easy to do for a single team, but not easy to do for all teams.

TD's + FG's + missed FG's + Failed Fourth Down conversions + INT's + Fumbles + Punts

That does mean that possessions when the clock ran out at the end of the half or the game don't count, and touchdowns scored on kickoff returns and punt returns are credited to the offense (and count as "possessions"), though fumbles on special teams also count as possessions in the denominator.

The caveats about a portion of the season being played are well taken.  This is really meant to give a rough picture of where we are.  We really miss Andrew Luck.

BC

[slide]

Just to put what Yvonne wrote into numbers, I took the offensive stats from cfbstats.com for all the years they had data (2008 through the ND game in 2014), and calculated points, yards, and first downs per possession.  I like per possession stats because you don't have to worry about the pace of play (a team like Oregon that scores quickly will get more possessions than a team like Stanford and would be expected to get more yards and more points, even if the offenses are equally efficient at turning possessions into scores.  The system is not perfect, and doesn't account for the strength of the defenses faced, but I find the stats are very useful.

Since the data goes back to 2008, we can see the effect of having Andrew Luck in the backfield (three years, from 2009 - 2011) and the effect of having both Shaw (2011 to 2014) and Harbaugh (2008 - 2010) as head coaches.  Most importantly, each head coach had an opportunity to coach Luck and one other QB (Pritchard for Harbaugh, and Hogan for Shaw), so it really gives us a chance to disentangle the effect of Andrew Luck from the effect of the coach. To be fair, I think Hogan is a better QB than Pritchard was, though I think that is balanced out by the fact that Harbaugh had Gerhart as running back, and this year, Shaw does not have a running back of the caliber of Toby. Note, I listed Hogan as the QB for 2012, even though Nunes started a number of the games.



Looking at the raw sats, it is clear that Andrew Luck was a difference maker; his three years we averaged 3.6 points per possession (a touchdown every other possession), 2.1 first downs per possession, and 42 yards per possession.  In the other years, we averaged 2.4 points per possession (PPP), 1.6 first downs per possession (FDPP), and 31 yards per possession (YPP).  In other words, Andrew Luck was responsible for a 47% increase in PPP,  31% increase in FDPP, and 30% increase in YPP.

Looking at coaches, having Harbaugh as a coach was worth 0.3 PPP (3.07 PPP for Harbaugh vs. 2.80 PPP for Shaw), 0.18 FDPP (1.93 vs. 1.75), and 2.5 YPP (37.6 vs. 35.1).  However, Harbaugh had Luck for 2 out of his three years here, while Shaw had Luck only 1 year.  If you do a multiple linear regression (sorry to be geeking out here), you get the following.

As for the question of whether the "trend has been worse since Luck left", we certainly took a big hit with Luck's departure.  But 2012, 2013, and 2014 look pretty similar to me.  Points per possession is down this year; probably a reflection of our red zone troubles early in the year (note, I think we were perfect in the red zone against the University of Dead Fake Girlfriends (our botched field goal was from the 25; the problem was we didn't even make it to the red zone against ND).  First downs per possession and yards per possession are between 2012 and 2014.


lm(formula = PPD ~ Shaw + Luck, data = o)
Residuals:
      1        2        3        4        5        6        7
-0.35765  0.45235 -0.14765  0.05294  0.22353 -0.27647  0.05294

Coefficients:
            Estimate Std. Error t value Pr(>|t|) 
(Intercept)  2.2171    0.2805  7.904  0.00139 **
Shaw          0.1706    0.2930  0.582  0.59161 
Luck          1.2794    0.2930  4.367  0.01199 *
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.3487 on 4 degrees of freedom
Multiple R-squared:  0.8395, Adjusted R-squared:  0.7593
F-statistic: 10.46 on 2 and 4 DF,  p-value: 0.02575

Andrew Luck is worth 1.28 PPP (statistically significant); there is no statistical significance between Harbaugh and Shaw, though, in the model, Shaw actually gets 0.2 PPG more than Harbaugh.  I won't throw the regression code up for FDPP (I can send it if anyone wants), but Luck is worth 0.51 FDPP with a p-value of < 0.01; there is no significant difference between Harbaugh and Shaw, and Luck is worth 10.4 YPP (p = 0.01), and again, a non-significant difference between Shaw and Harbaugh.

Obviously there are a lot of caveats here; while I adjust for Andrew Luck, there are a lot of other players and coaches involved, and it is certainly possible that Andrew Luck aside, Shaw might have had more talent at his disposal his three years than Harbaugh did.  Harbaugh did have Luck his redshirt freshman year, where he was somewhat less effective than his other two years, and you might argue that an offense with Luck as a redshirt junior should have accomplished more than an offense with Luck as a redshirt sophomore (the offenses the two years were pretty similar in efficiency).  Finally, none of this takes into account external factors such as the quality of the defense we were facing, how much of the game was uncompetitive (when we pulled starters and/or went with a conservative play calling), or weather, which was certainly a factor Saturday.

BC

BC -- first, thanks again for all the work.  very interesting results from your analysis.  on the caveats, I suspect you could look at inheriting a 1-11 team as a situation where there is nowhere to go but up.  on the other hand, there is little room for improvement when you inherit a 12-1 team. 

[slide]

Just to put what Yvonne wrote into numbers, I took the offensive stats from cfbstats.com for all the years they had data (2008 through the ND game in 2014), and calculated points, yards, and first downs per possession.  I like per possession stats because you don't have to worry about the pace of play (a team like Oregon that scores quickly will get more possessions than a team like Stanford and would be expected to get more yards and more points, even if the offenses are equally efficient at turning possessions into scores.  The system is not perfect, and doesn't account for the strength of the defenses faced, but I find the stats are very useful.

Since the data goes back to 2008, we can see the effect of having Andrew Luck in the backfield (three years, from 2009 - 2011) and the effect of having both Shaw (2011 to 2014) and Harbaugh (2008 - 2010) as head coaches.  Most importantly, each head coach had an opportunity to coach Luck and one other QB (Pritchard for Harbaugh, and Hogan for Shaw), so it really gives us a chance to disentangle the effect of Andrew Luck from the effect of the coach. To be fair, I think Hogan is a better QB than Pritchard was, though I think that is balanced out by the fact that Harbaugh had Gerhart as running back, and this year, Shaw does not have a running back of the caliber of Toby. Note, I listed Hogan as the QB for 2012, even though Nunes started a number of the games.



Looking at the raw sats, it is clear that Andrew Luck was a difference maker; his three years we averaged 3.6 points per possession (a touchdown every other possession), 2.1 first downs per possession, and 42 yards per possession.  In the other years, we averaged 2.4 points per possession (PPP), 1.6 first downs per possession (FDPP), and 31 yards per possession (YPP).  In other words, Andrew Luck was responsible for a 47% increase in PPP,  31% increase in FDPP, and 30% increase in YPP.

Looking at coaches, having Harbaugh as a coach was worth 0.3 PPP (3.07 PPP for Harbaugh vs. 2.80 PPP for Shaw), 0.18 FDPP (1.93 vs. 1.75), and 2.5 YPP (37.6 vs. 35.1).  However, Harbaugh had Luck for 2 out of his three years here, while Shaw had Luck only 1 year.  If you do a multiple linear regression (sorry to be geeking out here), you get the following.

As for the question of whether the "trend has been worse since Luck left", we certainly took a big hit with Luck's departure.  But 2012, 2013, and 2014 look pretty similar to me.  Points per possession is down this year; probably a reflection of our red zone troubles early in the year (note, I think we were perfect in the red zone against the University of Dead Fake Girlfriends (our botched field goal was from the 25; the problem was we didn't even make it to the red zone against ND).  First downs per possession and yards per possession are between 2012 and 2014.


lm(formula = PPD ~ Shaw + Luck, data = o)
Residuals:
      1        2        3        4        5        6        7
-0.35765  0.45235 -0.14765  0.05294  0.22353 -0.27647  0.05294

Coefficients:
            Estimate Std. Error t value Pr(>|t|) 
(Intercept)  2.2171    0.2805  7.904  0.00139 **
Shaw          0.1706    0.2930  0.582  0.59161 
Luck          1.2794    0.2930  4.367  0.01199 *
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.3487 on 4 degrees of freedom
Multiple R-squared:  0.8395, Adjusted R-squared:  0.7593
F-statistic: 10.46 on 2 and 4 DF,  p-value: 0.02575

Andrew Luck is worth 1.28 PPP (statistically significant); there is no statistical significance between Harbaugh and Shaw, though, in the model, Shaw actually gets 0.2 PPG more than Harbaugh.  I won't throw the regression code up for FDPP (I can send it if anyone wants), but Luck is worth 0.51 FDPP with a p-value of < 0.01; there is no significant difference between Harbaugh and Shaw, and Luck is worth 10.4 YPP (p = 0.01), and again, a non-significant difference between Shaw and Harbaugh.

Obviously there are a lot of caveats here; while I adjust for Andrew Luck, there are a lot of other players and coaches involved, and it is certainly possible that Andrew Luck aside, Shaw might have had more talent at his disposal his three years than Harbaugh did.  Harbaugh did have Luck his redshirt freshman year, where he was somewhat less effective than his other two years, and you might argue that an offense with Luck as a redshirt junior should have accomplished more than an offense with Luck as a redshirt sophomore (the offenses the two years were pretty similar in efficiency).  Finally, none of this takes into account external factors such as the quality of the defense we were facing, how much of the game was uncompetitive (when we pulled starters and/or went with a conservative play calling), or weather, which was certainly a factor Saturday.

BC

BC -- first, thanks again for all the work.  very interesting results from your analysis.  on the caveats, I suspect you could look at inheriting a 1-11 team as a situation where there is nowhere to go but up.  on the other hand, there is little room for improvement when you inherit a 12-1 team. 

[BostonCard]

Thankfully CFBstats doesn't go back further than 2008.  There would be some really BAD teams back then.

My goal here was twofold, first I wanted to see if there really was a trend in the post-Luck era, or if what we are really seeing is the absence of Andrew Luck in Shaw's last three years, and secondly I wanted to see if Shaw really is a as offensively inept as people here seem to think he is, or if he just doesn't have an all-everything QB.

One of the cool things about the regression is that it allows you to explore what would have happened in an alternate universe where Luck doesn't come to Stanford, and we have some other QB at the helm.  With all the caveats that have been mentioned, plus my favorite by George P. Box ("All models are wrong; some are useful"), here is what the last seven years of Stanford football look like in a universe where Andrew Luck never enrolled:







As you can see, the stats per possession bounce around a bit.  We drop a little bit from 2008 to 2009, and again from 2013 to this year (though keep in mind CTCard's point about a partial season).  I don't see a clear trend, and it doesn't seem like changing coaches made a bit of a difference.

BC

[BostonCard]

Thankfully CFBstats doesn't go back further than 2008.  There would be some really BAD teams back then.

My goal here was twofold, first I wanted to see if there really was a trend in the post-Luck era, or if what we are really seeing is the absence of Andrew Luck in Shaw's last three years, and secondly I wanted to see if Shaw really is a as offensively inept as people here seem to think he is, or if he just doesn't have an all-everything QB.

One of the cool things about the regression is that it allows you to explore what would have happened in an alternate universe where Luck doesn't come to Stanford, and we have some other QB at the helm.  With all the caveats that have been mentioned, plus my favorite by George P. Box ("All models are wrong; some are useful"), here is what the last seven years of Stanford football look like in a universe where Andrew Luck never enrolled:







As you can see, the stats per possession bounce around a bit.  We drop a little bit from 2008 to 2009, and again from 2013 to this year (though keep in mind CTCard's point about a partial season).  I don't see a clear trend, and it doesn't seem like changing coaches made a bit of a difference.

BC