No purple shades here...

[rmcjacket23]

But you referenced the entire league when you said that you'd probably see a trend of 4-5 spot differential across the league, which clearly isn't true and the fact of the matter is that when you're talking about it going hand in hand, there are multiple levels across the defense and it's interesting to see that they don't always work hand in hand. You got bested, but good try at covering your tracks.

lol, except I didn't, and you know that.

 

The only way to really know is to run a full sample. You picked a three year sample, and already admitted that roughly a third of the teams had a 10+ spread variance, but also acknowledged that the other two thirds of the league didn't. So what basis are you saying that the league, as a whole, doesn't have a 4-5 spot differential across the league, which is what I estimated? Are you saying that based on a third of the sample, you're already concluding that my estimate is wrong? How is that not short-sighted?

 

If I get an opportunity later today, I'll try to sample it and will provide the results. If I'm wrong, so be it, the numbers will tell the story. Certainly not going to just look at a third of the teams on a three year sample and say "yup, I most be way off".

Edited June 16, 2015 by rmcjacket23

[1/28/01]

Go ahead and see my above post. I think you may like it.

Why thank you, sir, for not only the post above, but for being a really solid contributor to this forum!

[1/28/01]

That's an extremely weak response to a good post and rebuttal of your point.

Par for the course, perhaps? But yes, agreed with you.

[rmcjacket23]

But you referenced the entire league when you said that you'd probably see a trend of 4-5 spot differential across the league, which clearly isn't true and the fact of the matter is that when you're talking about it going hand in hand, there are multiple levels across the defense and it's interesting to see that they don't always work hand in hand. You got bested, but good try at covering your tracks.

Here's the data:

 

2014: Mean differential in ranking between YPG and PPG across the entire league was 5.8. The Mode was 0 (5 teams). 19 of 32 teams (59%) were within the mean range.

 

2013: Mean differential in rankings between YPG and PPG across the entire league was 5.8. The Mode was 0 (7 teams). 18 of 32 teams (56%) were within the mean range.

 

2012: Mean differential in rankings between YPG and PPG across the entire league was 6.4. The Mode was 5 (7 teams).  22 of 32 teams (69%) were within the mean range.

 

Over the full three year sample, the three years mean differential in rankings between YPG and PPG across the entire league is 6.0. The Mode was 0 (15%). 59% of the league finished within +/- 5 spots in the rankings variance, and 77% of the league finished within +/- 10 spots in the rankings variance.

 

For Ravens perspective:

 

Mean differential in rankings between YPG and PPG across the three year sample is 2.3. Fell within +/- 5 spots in the rankings variance 3 of 3 years.

[BmoreBird22]

lol, except I didn't, and you know that.

 

The only way to really know is to run a full sample. You picked a three year sample, and already admitted that roughly a third of the teams had a 10+ spread variance, but also acknowledged that the other two thirds of the league didn't. So what basis are you saying that the league, as a whole, doesn't have a 4-5 spot differential across the league, which is what I estimated? Are you saying that based on a third of the sample, you're already concluding that my estimate is wrong? How is that not short-sighted?

 

If I get an opportunity later today, I'll try to sample it and will provide the results. If I'm wrong, so be it, the numbers will tell the story. Certainly not going to just look at a third of the teams on a three year sample and say "yup, I most be way off".

Are you going to say that 33%+ of the league seeing a large differential is just something we can ignore and say that it always goes hand in hand? A variance of that large a degree is definitely something that no one will consider small and if I got a variance of 33% in any lab, it'd be something to definitely look into. And honestly, I used 4-5 spots as the barometer, but if you put the cutoff as a differential of 5+ spots (since the difference in the Ravens rankings was 3.6 and that would put it at 3-4), then you'd likely see about half of the league or more. I found several times where there were like four teams in a row that had a five spot differential. Which should tell you that they don't always go hand in hand. 

 

I actually did a linear regression because there should be some sort of relationship between your total yards and points per game, according to you, but there was literally no relationship if you run a linear regression. None at all. 

Edited June 16, 2015 by BmoreBird22

[rmcjacket23]

Are you going to say that 33%+ of the league seeing a large differential is just something we can ignore and say that it always goes hand in hand? A variance of that large a degree is definitely something that no one will consider small and if I got a variance of 33% in any lab, it'd be something to definitely look into. And honestly, I used 4-5 spots as the barometer, but if you put the cutoff as a differential of 5+ spots (since the difference in the Ravens rankings was 3.6 and that would put it at 3-4), then you'd likely see about half of the league or more. I found several times where there were like four teams in a row that had a five spot differential. Which should tell you that they don't always go hand in hand. 

See above.

 

And remember, this was never a notion of "they always go hand in hand". I said they GENERALLY SPEAKING go hand in hand. And in particular to the Ravens, they actually go hand in hand much more frequently than the rest of the league on average, particularly during Pees tenure as DC, which is the sample I took.

[rmcjacket23]

Also, I probably should have ran simple statistical analysis sooner...

 

The correlation coefficient over that three year sample when comparing yardage allowed to points allowed is 0.66. This, from a statistical concept, indicates a positive linear relationship between the two categories, and that one tends to coincide with the other. 

[BmoreBird22]

Also, I probably should have ran simple statistical analysis sooner...

The correlation coefficient over that three year sample when comparing yardage allowed to points allowed is 0.66. This, from a statistical concept, indicates a positive linear relationship between the two categories, and that one tends to coincide with the other.

I got an r^2 value of like 0.32 when comparing total yardage rank to points per game rank and if you're correlation coefficient is 0.66, you're not finding a strong relationship. If that were my value in the lab, I'd get chewed out so hard for terrible lab practice. Sure, there is a relationship, and it's obviously going to be linear, but there is not a strong one

 

Just did it for 2014 and the r^2 value was .4722, or a very, very weak relationship.

 

I'm also questioning how you did your r^2 value

Edited June 16, 2015 by BmoreBird22

[BmoreBird22]

Also, I probably should have ran simple statistical analysis sooner...

 

The correlation coefficient over that three year sample when comparing yardage allowed to points allowed is 0.66. This, from a statistical concept, indicates a positive linear relationship between the two categories, and that one tends to coincide with the other. 

Oh, and by the way, a negative correlation coefficient will also indicate a relationship, it's just a negative relationship. Positive doesn't necessarily mean it's a good relationship

[rmcjacket23]

Oh, and by the way, a negative correlation coefficient will also indicate a relationship, it's just a negative relationship. Positive doesn't necessarily mean it's a good relationship

Correct, it doesn't. A negative correlation coefficient would indicate that they operate opposite from each other.

 

Its certainly a positive correlation, and depending on how you interpret it (amongst other things), its either a moderate positive relationship or a strong positive one. Typical stat "benchmarks" indicate a R2 between roughly 0.3 and 0.7 would be a moderate positive relationship, so its sort of in the upper end of that range.

 

If you actually plot this on a graph, there's certainly a positive linear relationship, but I suppose it is up for debate about how strong it is.

 

 

 

We do know that roughly 60% of the league will fall within a +/- 5 ranking between PPG and YPG. 

[Inqui]

Man, you guys must be hanging out for the season to start.

[BmoreBird22]

Correct, it doesn't. A negative correlation coefficient would indicate that they operate opposite from each other.

 

Its certainly a positive correlation, and depending on how you interpret it (amongst other things), its either a moderate positive relationship or a strong positive one. Typical stat "benchmarks" indicate a R2 between roughly 0.3 and 0.7 would be a moderate positive relationship, so its sort of in the upper end of that range.

 

If you actually plot this on a graph, there's certainly a positive linear relationship, but I suppose it is up for debate about how strong it is.

 

 

 

We do know that roughly 60% of the league will fall within a +/- 5 ranking between PPG and YPG. 

Yep, about 60% of the league will and for me, that's not a strong relationship at all and suggest it isn't as hand in hand as it should be.

[January J]

Too many numbers for me, I just want to watch some football already.

[FlackAttack]

Too many numbers for me, I just want to watch some football already.

 

i hear ya man, im tired of nothing but speculation of players  and "instagram" and "wedding photos" of other ravens players. lets let these players start hitting each other! (without injury of course)

[lgcs27288]

Do we need to get Urschel in here?