Historical Correlation between 3rd Preseason Game and Season Record (for last 12 seasons)

[iBBB]

So I was intrigued by the idea that the 3rd preseason game is an indicator of how the season will pan out.

The performance of the players in that game might be telling of the season to come, sure, but the actual game outcome does not.

 

It turns out that there is no correlation between the outcome of the this game and season record. 

Here is what I put together. Have a look....

[PanthersUnited]

So we haven't had a single winning season when we lost the 3rd pre season game in the last 12 years. Interesting 

[ncbandit]

True, but we have had just as many non-winning seasons when we did win the 3rd. Not sure about the significance of that (if any) correlation. 

[t96]

Thank God for Mr Clutch himself, Joe Webb. 

[Mother Grabber]

you can find historical correlations between spins on a roulette wheel, too, doesn't mean they actually have an affect on the next one.

[beastson]

All I'm seeing from this is a bunch of sub-par records since 2003

[jimfear]

I plugged your data into excel to run a quick OLS regression.  Using Panthers wins/losses, point differential, and opponent's record as independent variables and Panthers season record as the dependent variable, I only get an R^2 of ~.36.  None of the p-values appear significant, so as a whole the regression doesn't seem to tell us much.  When I regress Panthers season record on point differential, I still end up with a ~.35 R^2 and the p-value is much lower.  Seems like the point differential has the most predictive power. (I realize this is by no means a thorough statistical analysis - my intent is just to provide info to those who are curious)

[iBBB]

I plugged your data into excel to run a quick OLS regression.  Using Panthers wins/losses, point differential, and opponent's record as independent variables and Panthers season record as the dependent variable, I only get an R^2 of ~.36.  None of the p-values appear significant, so as a whole the regression doesn't seem to tell us much.  When I regress Panthers season record on point differential, I still end up with a ~.35 R^2 and the p-value is much lower.  Seems like the point differential has the most predictive power. (I realize this is by no means a thorough statistical analysis - my intent is just to provide info to those who are curious)

I agree, that's why I included the point differential.

I also thought that the strength of the teams we beat would also have a correctional to the season to come, but it doesn't.

[jimfear]

I agree, that's why I included the point differential.

I also thought that the strength of the teams we beat would also have a correctional to the season to come, but it doesn't.

Yep - surprising.  BTW - thank you for pulling the data together.

[KSpan]

I'm seeing that playing NE in the 3rd game is the kiss of death for a winning season.  I think Belichick puts a voodoo curse on the team or something.

[pantherclaw]

50% is a direct correlation? 

I think not. 

8 preseason 3rd games won, only 4 winning seasons.

Now do the same for the rest of the league.

[iBBB]

50% is a direct correlation? 

I think not. 

8 preseason 3rd games won, only 4 winning seasons.

Now do the same for the rest of the league.

I think you misunderstood my post. I state that there is no correlation... 

[BigBoss]

So basically everytime we play the patriots, they gonna go to da super bowl and we ain't.

 

Nah very well put together, thanks for bringing this by! Well done indeed!

[pantherclaw]

I was responding to PU, that thought any of it supports his theory. 

Knowing if it was done with every team, it would completely destroy it. 

Appreciate your time in research.