None of your stats work unless teams aren't playing against other teams, just simulations of other teams.
Sacks Have Little to Nothing to Do with NFL Quarterback Play and Winning
- Thread starter Shouright
- Start date
None of your stats work unless teams aren't playing against other teams, just simulations of other teams.
We're all entitled to our own opinions.
cltchperf
Pro Bowler
- Joined
- May 12, 2005
- Messages
- 10,159
- Reaction score
- 5,756
Well then I'd sure encourage you to move on to a smarter crowd, for your own benefit.
---------- Post added at 05:22 PM ---------- Previous post was at 05:21 PM ----------
I have no qualms at all with the idea that sack differential has a significant impact on winning.
Sack differential has a significant impact on winning, because SACKS DO HAVE AN IMPACT ON DECIDING THE QB AND TEAMS CHANCES OF WINNING. If it didn't there would not be a difference between the teams whose qb is getting sacked a lot vs the one who isn't. The chances of the qb getting beat up would be the same as the team who qb is not getting sacked, but obviously as finfaninbuffalo's stats show, there is the difference which proves that SACKS DO MAKE A DIFFERENCE ON A TEAM'S CHANCES OF WINNING. The winninger percentage would be the same as team qb A who is gettinb beat up vs team qb B who is not, or the difference would mean nothing but it obviously does. I don't know what is so hard to grasp about this.
nyashfan
For Earth Below
There were two key points being made over the last several posts.
1) It is fair to use winning percentage as the dependent variable provided that sack differential was the independent variable. When you are only considering sacks committed by the offense, then points, and not winning percentage, should be the dependent variable.
2) The more refined the data being analyzed, the more meaningful the conclusions. I think back on page 8 of this thread I advocated looking at the correlation between sack yardage per drive and points per drive, with separate analyses broken out by starting field position. FinFanInBuffalo echoed those same sentiments and provided links to such studies.
The next best thing was looking at the sack differential study which was based upon sacks per game data.
What is least ideal is to use data based on seasonal aggregated data. Aggregation destroys natural variation and leads to inconclusive or incorrect results. This was shown in one of FinFanInBuffalo's links in which the author described anomalies when looking at a team's seasonal aggregate sack totals.
Let's consider the extreme case. Suppose the Dolphins committed 64 sacks for the season. If your goal is to maximize the number of Dolphin wins, you would rather have the Dolphins committing all 64 sacks in one game and 0 in the other 15 as opposed to the Dolphins having committed 4 sacks in each of their 16 games.
This reiterates one of my earlier points, namely that analyses based on aggregate data are being presented because the data is more readily available. To do the analyses correctly you need to use more refined data even though it is more difficult to collect and more time consuming to perform.
1) It is fair to use winning percentage as the dependent variable provided that sack differential was the independent variable. When you are only considering sacks committed by the offense, then points, and not winning percentage, should be the dependent variable.
2) The more refined the data being analyzed, the more meaningful the conclusions. I think back on page 8 of this thread I advocated looking at the correlation between sack yardage per drive and points per drive, with separate analyses broken out by starting field position. FinFanInBuffalo echoed those same sentiments and provided links to such studies.
The next best thing was looking at the sack differential study which was based upon sacks per game data.
What is least ideal is to use data based on seasonal aggregated data. Aggregation destroys natural variation and leads to inconclusive or incorrect results. This was shown in one of FinFanInBuffalo's links in which the author described anomalies when looking at a team's seasonal aggregate sack totals.
Let's consider the extreme case. Suppose the Dolphins committed 64 sacks for the season. If your goal is to maximize the number of Dolphin wins, you would rather have the Dolphins committing all 64 sacks in one game and 0 in the other 15 as opposed to the Dolphins having committed 4 sacks in each of their 16 games.
This reiterates one of my earlier points, namely that analyses based on aggregate data are being presented because the data is more readily available. To do the analyses correctly you need to use more refined data even though it is more difficult to collect and more time consuming to perform.
cltchperf
Pro Bowler
- Joined
- May 12, 2005
- Messages
- 10,159
- Reaction score
- 5,756
You must be trolling,,saying this stuff just for kicks to get people repeating things they've already said. If you can't see how what I wrote refutes your point, then you're a lost cause.
Well then let's just de-aggregate the whole thing and bring it right back to the brass tacks for which it was originally intended.
Game to game, the correlation between the points the Dolphins scored in 2013 and its percentage of offensive sacks (sacks divided by pass attempts) was -0.28.
Game to game, the correlation between the Dolphins' percentage of offensive sacks and Ryan Tannehill's QB rating was -0.10. Game to game, the correlation between the Dolphins' percentage of offensive sacks and Ryan Tannehill's YPA was -0.16.
There are weak relationships between the Dolphins' percentage of sacks on offense in 2013 and its: 1) points scored, 2) QB rating, and 3) YPA.
Surely it cannot be said that the Dolphins were hampered largely on offense in terms of scoring, QB rating, or YPA by their frequency of sacks.
And with that folks, I'm out. Have it at from here. :up:
Absolutely correct. I suggested it multiple times. He ignored this.
Not to mention time in the game and score in the game when the sack happens.
Bingo.
Very well said. The funny thing is that I am not a "stats guy". His approach is so bad even a relative noob with stats can see the problems from a mile away.
The final flaw is his failed conclusion based on the correlation he computed using his dubious approach. This is the perfect example where causation can exist without correlation because there is so much unaccounted for in his data. In addition, we have real examples of sacks leading directly to losses in the data he used by the team he claims was unaffected by sacks.
All this while living in a universe that includes NFL teams trying like crazy to stop sacks.
And in season with the team losing so many close games, this likely cost the Dolphins' wins. Now we are done.
nyashfan
For Earth Below
Shouright,
Perhaps I am misunderstanding the data you are describing. Is it accurate to say that this data is a 16 record subset of the data used in FinFanInBuffalo's link that used data for all teams over the last 3 years? And yes I realize that your analysis uses correlations as opposed to win percentage split by positive versus negative sack differential.
What I would ask is why just focus on the data for only the 2013 Dolphins as it is such a limited sample? I think the disconnect is why not broaden the study to draw a conclusion about quarterbacks in general and not just Ryan Tannehill. It could be that you had aggregate data for all quarterbacks but per game data only for the 2013 Dolphins. Anyway, it's the per game data for all teams over the last 3 years that makes the study in FinFanInBuffalo's link so compelling.
Perhaps I am misunderstanding the data you are describing. Is it accurate to say that this data is a 16 record subset of the data used in FinFanInBuffalo's link that used data for all teams over the last 3 years? And yes I realize that your analysis uses correlations as opposed to win percentage split by positive versus negative sack differential.
What I would ask is why just focus on the data for only the 2013 Dolphins as it is such a limited sample? I think the disconnect is why not broaden the study to draw a conclusion about quarterbacks in general and not just Ryan Tannehill. It could be that you had aggregate data for all quarterbacks but per game data only for the 2013 Dolphins. Anyway, it's the per game data for all teams over the last 3 years that makes the study in FinFanInBuffalo's link so compelling.
Additionally, it's also true that there was no correlation game-to-game in 2013 between the number of sacks the Dolphins surrendered and the statistics of Ryan Tannehill's alluded to in the original post.
So in essence, there is no objective evidence that quarterbacks who are sacked less frequently play better, in terms of the quarterback-related statistics most strongly correlated with winning.
So in essence, there is no objective evidence that quarterbacks who are sacked less frequently play better, in terms of the quarterback-related statistics most strongly correlated with winning.
:lol2: still wrong..... :lol2:
Who gives a ****
What a bunch of convoluted horse****. You have said YPA is is a QB stat most strongly associated with winning. Of course sacks wont have an effect on YPA....a sack is not a passing play, and YPA is a passing play. Why would a non-passing play be factored into YPA. You are a fraud.
123phins
Active Roster
- Joined
- Jan 27, 2014
- Messages
- 734
- Reaction score
- 6
this thread is a waste of time. State your statistics all you want. First think of how many last minute drives to comeback and win did the dolphins have? now count haw many of those drives lost yardage and time due to a sack. pretty much all of them. The QB cant just take a check down when there is 80 yards to go with a minute on the clock, he has to stay in there and throw it for 20+ yard gains.
-Sacks cause long yardage situations and predictability in play calls. Its alot harder for a QB to to make a first down throw when the whole stadium means its gonna be a throw
-Sacks mean stalled drives and lost chances for points, which in alot of cases makes for lost games.
-Sacks cause long yardage situations and predictability in play calls. Its alot harder for a QB to to make a first down throw when the whole stadium means its gonna be a throw
-Sacks mean stalled drives and lost chances for points, which in alot of cases makes for lost games.