1. Welcome to CowboysZone!  Join us!  Come on!  You know you want to!

Off-Season Priority #1: Find A Running Game

Discussion in 'Fan Zone' started by TheFinisher, Jan 1, 2013.

  1. Omegasupreme

    Omegasupreme Member

    248 Messages
    1 Likes Received
    Well done. You have ushered in the correct wording of "probability". The use of the word correlation is nonsense because there are too many variables to try to isolate whether the increase or decrease in any particular play molecule will give give success of the whole of the entire offense over the time of the game against the whole of the defense over the time of the game with all adjustments.
    It's not that simple.

    But there are a few good probabilities. Parcells used to factor these probabilities into each game. First, the amount of error that is factored into every pass attempt is greater than the amount of error involved in a run attempt. There are several extraneous and direct variables in one pass play, which increase by the distance of the pass and the amount of time for the play to complete, than in one simple running play.

    Parcells did look at the increased probability of keeping the ball and gaining forward progress, as well as controlling the clock, on a running play than attempting less probable plays downfield, especially slow-developing.

    His reasoning for running Ottis Anderson in Super Bowl XXV was that Buffalo had an explosive pass offense and he did not want to go back and forth down the field with them as each pass play delivers the ball back to the other team at a greater rate than a running play (obviously). But he noticed that Anderson was averaging 3.5 ypc in the earlier part of the game and reasoned (probability) that increasing the amount of runs increases the probability for a first down for every three downs, increases the probability for time of possession being away from the Bills, and reduces the probability for turnover error.

    Improving the number of running plays and effective running plays does increase the likelihood of time of possession for a team and time of possession increases the likelihood of offensive scoring (obviously).

    But look at the Niners in 2011. How confident were the Niners with Alex Smith (especially with the Garrett notion that longer pass plays open the running lanes underneath) with low percentage attempts? How confident were the Niners, despite a immature pass plan, with running Frank Gore play after play even though the defense knew it was likely that he would get the ball and the defense did not need to assign extra players deep to cover the Niner receivers? Did the Niners run anyway? We're there games in 2011 where the Frank Gore ran for less than 4 ypc? Less than 3 ypc? Did they still continue to run him anyway? Is there any benefit to continuing to run the ball for very few yards per carry?
  2. AdamJT13

    AdamJT13 Salary Cap Analyst

    14,984 Messages
    456 Likes Received
    I've addressed run success rate before. First of all, it still shows a weak correlation -- and other studies I've seen indicate that it's even lower than ANS suggests. Secondly, ANS bases its correlation on season-long RSR and final records, which -- to me -- is a horrible way to look at how it correlates to winning games. Because it's impossible to find RSR numbers for every game (at least, I've never found any), it's impossible to find out how it truly correlates to winning.

    But, I've said before that the only thing I'd really worry about as far as running or stopping the run is short-yardage situations. We were one of the best short-yardage rushing teams in the league this season. And if you do want to give weight to RSR, we were right in the middle of the pack. So even if you think that those things make running more important, they also indicate that it isn't a dire need.


    I don't know enough about how that site calculates win probability to determine how credible that stat is (it would seem to give weight to certain game situations more than others), but it again appears to be a sum total for the season. The fact that Washington ran more than it passed -- especially late in games -- would seem to affect how the sum totals add up, as well.
  3. AdamJT13

    AdamJT13 Salary Cap Analyst

    14,984 Messages
    456 Likes Received
    Only if the few passes that are attempted are successful. If you limit your quarterback's attempts and he's terrible on them, you probably will not win no matter how much or how well you run (unless your defense is even better against the pass).

    This season, there were 13 teams that had 24 or fewer pass attempts and an ANYPA of 3.50 or less. They went 3-10. The 44 teams that had 24 attempts or fewer and an ANYPA of 3.50 or better went 42-2.
  4. Hoofbite

    Hoofbite Well-Known Member

    32,528 Messages
    2,215 Likes Received

    I was posting from my phone so I was trying to be brief. I'll add a little to what I was trying to say. I don't think it's reasonable to say definitively whether or not QBs are more effective for a couple of reasons.


    Weekly Variability Amongst Opponents.

    You are comparing between weeks against teams of varying quality. Perhaps the game when the team passed poorly but ran well they were playing against a team with a strong secondary and weak front seven while the games where the team ran poorly but passed well was because the opposing defense simply couldn't stop the pass at all, reducing any desire to even run. It's not like yards per rush is what is really gained on any given attempt. More often than not it's the big runs making up for the small ones and any team that basically passes all day and only runs in short yardage situations will likely have a horrid rushing average with great passing stats. Or consider a team that is running very well and is only using the passing game in a conservative manner. Small chunks here and there. Nothing high risk.

    Passing picks up more yardage than running so success in the passing game would immediately reduce anyone's desire to run if they were successful because yardage typically means points. If the situations were flipped, running success would dominate in the same fashion and teams wouldn't have any desire to pass because in such a scenario running is already gaining more yards than can be expected from a collection of pass attempts. "Why change what is working"? That's basically what it comes down to.

    There's simply no telling what the passing outcome would have been in any individual game with varying degrees of success from the running game because we only have one degree of success. Perhaps an average passing performance would have been made worse with a worse rushing performance. Or perhaps an average passing performance would have been better with a better rushing attack for that week. There's just no way of knowing because we can't measure at different points along the rushing spectrum for a single game. At best, the correlation is built on that assumption that all defenses are created equal and there's absolutely no way anyone would consider that a valid assumption to make.


    QB Ability

    Passing effectiveness is largely going to be dependent on the QB. You aren't gonna get Tom Brady numbers from Brady Quinn no matter how well the team runs the ball because Brady Quinn just isn't good enough. At the other end of the spectrum, Tom Brady isn't going to be reduced to Brady Quinn by a poor running game because Tom Brady is just too good. How much room for improvement do Tom Brady and Peyton Manning have? These guys are the gold standard and are good enough to "get theirs" regardless of just about anything else that happens. On the one hand you have guys who lack the ability to get off the floor in terms of QB performance and on the other end you have guys who are already at the ceiling.

    If we acknowledge the fact that some QBs are incapable of taking any advantage because they suck and other QBs are already good enough that the potential benefit to be realize is significantly lower to start with, any potential relationship is made harder to find. On both ends of the spectrum there are players who's performances will likely not change enough to even be noticed. For different reasons no doubt but the impact on looking for a statistical difference remains the same in that it would dampen any potential statistical difference.

    That fact that a statistical correlation doesn't exist shouldn't be considered definitive of whether or not the running game effects the passing game. Given that there are week-to-week variations in opponents, there is variability in QB ability, and no way to determine what effect a diminished or improved running game would have had within any single game I think there's too much uncertainty for the correlation to have practical utility. If you could definitively isolate and control for these, then there's some value but it's quite possible that these variables, and likely more that I haven't considered, effectively mask the direct and practical relationship leaving us with the null. I would wager that a horrible QB by itself could completely erase any potential benefit.



    I think simply looking at statistics is a clouded view at best because there are a number of other contributing factors in play. That's just from a statistical standpoint and doesn't include the practical effect that running well has on how a defense approaches or adjusts in the game.
  5. AdamJT13

    AdamJT13 Salary Cap Analyst

    14,984 Messages
    456 Likes Received
    The problem with all of that is that the correlation has held up for all types of teams (passing teams, running teams, balanced teams, good teams, bad teams, mediocre teams, etc.) against all types of teams, week after week, year after year after year. It's not like it's based on one team's offense in 16 games. It's based on 32 teams' offenses over 16 games each AND 32 teams' defenses over 16 games each AND 32 teams' net statistics over 32 games each -- per season. Over a 10-year period, that's a sample size of more than 15,000 "games."

    When all of the evidence says the same thing year after year after year, there comes a point when you have to quit trying to explain away the facts and just accept the way things are.

    To address a few specific points --

    OK, if we pretend that's true, wouldn't it emphasize that the key to winning on defense is to have a strong secondary, even if you have a weak front seven? After all, the theoretical team you described would win that game about 80 percent of the time.


    Again, if we pretend that's true, wouldn't it say that "any desire to even run" does not affect the team's chance of winning? After all, if you win 80 percent of the time when you run poorly and do not have any desire to even run, how important can running be?

    Let's see ... successful passing scores points, wins games and makes running irrelevant. Where have I heard that before?


    Except that such a team would usually lose that game unless it passes better than its opponent. Teams would abandon that strategy pretty quickly if they keep losing, don't you think?


    It's not a question of comparing Brady and Quinn, it's comparing Brady against Brady and Quinn against Quinn.

    I'd say it's pretty definitive that whatever effects it has are not measurably significant.
  6. CowboyMcCoy

    CowboyMcCoy Business is a Boomin

    12,749 Messages
    234 Likes Received
    Garrett refuses to establish the run. That's his modus operandi.
  7. Hoofbite

    Hoofbite Well-Known Member

    32,528 Messages
    2,215 Likes Received
    Measuring various lengths with a flawed ruler doesn't get you any closer to the truth simply because the ruler is found to be consistent.

    You cannot say what effect running has in a single game because there is nothing to compare to. Feel free to compare amongst weeks but you have to acknowledge that it is not a complete measurement of the entire picture no matter how long it has held up.

    It's not about whether or not effective passing leads to wins. I'm not sure anyone is arguing that. I've discussed this with you previously and I have said that I believe the only reason this holds true is because passing typically accounts for the lion's share of offenses. However, being responsible for a greater percentage of the offense doesn't completely erase the effect that the running game contributes with the remaining percentage.

    The question at hand is whether or not effective running produces more effective passing, something that cannot directly be measured.

    Two statements.

    1. More effective passing wins regardless of the running game.

    2. More effective running benefits the passing game.

    Both can be true. If a team can pass at will against poor defenses and chooses to do so, that doesn't disprove the notion that running effectively would help in situations in which they cannot pass at will. Again, a comparison between measurements taken at two different times against two different teams is not a valid basis for saying that running is "irrelevant". Additionally, being more more effective than the opponent doesn't even begin to address the notion that an improved running game in that specific contest would, or would not have, made the passing game even more effective than it was. The truth is we don't know because we don't have the comparator to work with.

    Furthermore, while I've seen you cling to the correlation, or lack thereof, this would be the first time where I have seen you outright say that running is "irrelevant". I've always assumed that you were implying as much but given that fact that a correlation can be present or absent while the practical implications of such are also insignificant I've never really pressed the issue. In that regard if you honestly mean to say that running is "irrelevant" I would say that you are entirely wrong. Based on the simple fact that every team runs, running must have some relevance regardless of what correlation exists or doesn't exist. Unless you want to back up your statement with the claim that every coach who has ever been in the NFL is wrong there is no alternative because if there truly was no relevance at all nobody would run. Ever.

    Such a team would win on the order of teams that pass more effectively now. My scenario was basically a total swap of running and passing output. If 250 yards running and 125 yards passing were the norm, the correlation to running effectively and winning would be just what the correlation is for the current situation where 250 yards passing and 125 yards rushing is the norm.........or however it breaks down.

    I may not have conveyed the complete flip as well as I wanted but that was my intent. If the running game were responsible for the lion's share of league yardage, effective running would correlate more to winning than passing.

    But even if we assume that a team could pull off 250 yards rushing per week in today's NFL I'm not sure you can use statistics from the status quo to say that they would lose the majority of games. If you're leading the league in yardage and doing it with 80% rushing, is the net effect any different than if you were doing it with 80% passing? Yards are yards. Passing TDs aren't worth any more than rushing TDs. The only advantage the passing game has is that the typical pass nets more yards than the typical run. If that weren't the case, then the correlation between passing and winning would likely fall apart.

    My point wasn't that you would compare the two. It's that both have limited ability to reap benefits from the running game. Tom Brady versus a good defenses isn't much different than Tom Brady versus a poor defense. He's so good that it doesn't matter who the opponent is. Quinn is the polar opposite. It doesn't matter who the opponent is because he just sucks and odds are his game is going to suck. How can you expect to see any difference in either?

    Basically, Quinn is so bad that he can't benefit and Brady is so good that the odds of getting more out of the limited opportunities in a game are slim.

    That's all well and fine but that doesn't mean the effects aren't significant in a practical sense; that there aren't real life implications to them. The fact that something can't be measured doesn't mean it doesn't exist. Conversely the fact that something that can be measured doesn't mean that it does is exist. That's why people put so much effort into controlling for potential confounders. They lead to inaccurate conclusions in that the statistics say one thing but the reality doesn't match up with it.

    In this case, quality of defenses faced likely correlates with passing efficacy for the vast majority of QBs. That is problematic when comparing QB statistics for the same player between weeks.

    Additionally, QB quality likely has something to do with passing efficacy don't ya think? This presents a problem when comparing multiple QBs to one another. It also likely presents a problem when comparing a single QB's performance between weeks because QB isn't static. If a QB is playing like crap one week and great the next, who know's where to even begin is assessing the effect of the running game?

    Passing efficacy is a multifaceted outcome. There are a number of different variables in play and trying to pin the end result on any single one seems like wasted effort.
  8. xwalker

    xwalker Well-Known Member

    12,376 Messages
    4,941 Likes Received
    Adam appears to be basing Passing Efficiency on ANPYA.

    ANPYA = (pass yards + 20*(pass TD) - 45*(interceptions thrown) - sack yards) / (passing attempts + sacks)

    ANPYA is a made-up equation. The fact that someone can make up an equation that indicates a certain outcome when tested against real statistical data, is basically meaningless. Adding pass yards to 20*pass TD is like adding inches to gallons.

    I could define:

    running efficiency = 10,000*(total points scored) + run yards

    You would find that running efficiency would always correlate to winning. It has run yards included in the formula and omits pass yards; therefore, you can conclude (incorrectly) that pass yards are unrelated (do not correlate) to winning.
  9. fifaguy

    fifaguy Well-Known Member

    4,442 Messages
    782 Likes Received
    I'll live it bold and hope for an O-line that can protect romo AND run block...
  10. FLcowboy

    FLcowboy Hopeful this is the year Zone Supporter

    3,677 Messages
    31 Likes Received
    Good point, and while this is certainly telling, how about the last time Dallas had two wide receivers on the field together that weren't hurt, or otherwise unable to strike fear in the defense.
  11. Hoov

    Hoov Senior Member

    5,050 Messages
    451 Likes Received
    All formulas are made up. analysts crunch numbers and try to come up with a good statistical formula that actually correlates to winning. None are perfect, but why would anyone ignore something that takes critical data into account and shows a high correlation to winning ?

    Obviously the formula you produced is not meant to be taken seriously.

    A team could pass for 500 yards and 4 td's and run one time for 1 yard and if they scored more than the other team they would be an efficient running team by that formula.

    Why dont you actually test the data for yourself and see if the passing efficiency formula proves true - all you have to do is pull up game statistics and plug the numbers in.

    And after that if you still believe that the run game is being slighted in its importance to winning. Then why dont you do some research and see if you can find a good formula to measure running efficiency then see if it actually correlates to winning. Maybe you will find something that no one else has found yet.
  12. xwalker

    xwalker Well-Known Member

    12,376 Messages
    4,941 Likes Received
    My example is an answer to your question. Combining real statistics with made-up equations can easily be Wrong.
    running efficiency = 10,000*(total points scored) + run yards

    My equation is a simple example of how you can easily make something appear to correlate to winning. It was intended to be simple to make a point.

    The fact that I could choose 10,000 as a multiplier at random allowed me to easily get the result that I wanted. Without the multiplier, I couldn't guarantee the results.

    ANPYA = (pass yards + 20*(pass TD) - 45*(interceptions thrown) - sack yards) / (passing attempts + sacks)

    The ANPYA equation is more complicated but the same principals apply. Note the multipliers in their equation of 20 and 45.

    Interceptions alone, without pass or run yards, have a high correlation to winning. No complicated formula needed:

    The Cowboys Won 87.5% of their games in 2012 when they threw the same or less interceptions than the opponent.

    Of those losses, the Cowboys averaged 65 yards rushing.

    The Cowboys lost 87.5% of their games when they threw more interceptions than the opponent.

    Of those Wins, they averaged 143 yards rushing.
    Yes, as I said above, that was my point.
    I expect the formula to give the "correlation" results that have been presented.

    My formula also gave the result that I intended; however, you can easily see that it is meaningless.
    NFL teams hire people/companies that are experts in Statistical Analysis. I can't begin to compete with their capabilities.

    If the running game could be proven to have no bearing on winning, NFL teams would know about. You, the fan would start to see teams go without running backs.

    Lastly, I'm not trying to be rude or insult anyone. I just don't believe that the information being disseminated is correct.
  13. AdamJT13

    AdamJT13 Salary Cap Analyst

    14,984 Messages
    456 Likes Received
    You have two flawed rulers. One has been found to be correct 80 percent of the time. The other has been found to be correct 20 percent of the time. Which would you rather use?

    I think it's pretty obvious that nobody can say with 100 percent certainty what would have happened in a single game if anything had happened differently, no matter what you're talking about -- except maybe the outcome of the game's final play. But when things happen a certain way the vast majority of the time, it's not a stretch to say what probably would have happened.


    Seriously?


    We can determine whether there is a correlation between effective running and effective passing. Your argument seems to be that because nobody can say for sure what would have happened in a single game, even if there is a strong correlation or a weak correlation, we should just ignore the correlations altogether.

    In other words, the real question is whether you'd rather base your decisions on correlation or speculation. I know what I would prefer.


    I've said many times that how well you run the ball is "virtually irrelevant," or something similar, when it comes to deciding who wins or loses, as well as in helping you pass the ball better. I suppose dropping the word "virtually" might be a slight exaggeration because there's always that 1-2 percent of the time when it might make a difference.


    That's not true at all. There are plenty of times when you might want to run the ball or should run it, even if how well you run it over the course of a game does not significantly affect your chances of winning.


    That's also not true. Passing isn't correlated more only because it accounts for a higher percentage of the yards and touchdowns in most games -- although that is one reason, and a good one. It also has to do with the greater variance in outcomes and the greater risk involved, as well as the time factor. If you get the ball at your own 20 with two minutes left, down by four points with no timeouts -- are you going to run or pass? The fact that a significant percentage of games comes down to one team's pass offense against another team's pass defense in the final minutes is a big reason why passing correlates highly with winning.

    We also can look at games when teams rushed for many more yards than they passed for and see whether the passing correlation falls apart. This season, there were nine games when a team gained at least 100 yards more by rushing than it did by passing. The teams that passed better went 7-2 in those games -- a 77.8 winning percentage, as close as possible to the overall percentage for all games for the season. There also were nine games when a team ran the ball at least 20 more times than it tried to pass. The team that passed more effectively in those games went 8-1, a winning percentage of 88.9. (The team that rushed more effectively went 5-4.) Even with a small sample size of nine games when a team was run-heavy -- either by total yards or by attempts -- the correlation holds up exactly as expected.



    But if running helps the passing game, shouldn't Quinn be better when his team runs well than when it doesn't? Shouldn't Brady be better when his team runs well than when it doesn't? Shouldn't Romo be better when his team runs well than when it doesn't? Shouldn't we see some evidence of the running game helping the passing game? Or is this only the kind of "help" that can't be measured?


    If you want to ignore all of the correlations that have held up for team after team after team, week after week after week, year after year after year, then go right ahead. You can pretend that there are hidden reasons why the team that passes better in a game wins 80 percent of the time. You can pretend that quarterbacks pass better when their teams run better. You can pretend that rushing better helps you win more. I'll just stick with the facts.
  14. TwoDeep3

    TwoDeep3 Well-Known Member

    6,219 Messages
    2,041 Likes Received
    While i have no stats to back this, a solid and consistent running game will hide flaws in a passing attack.

    Which is why I'd like to see the oine improved, because it is my theory that Romo needs to be less Superman and more bus driver.

    Limit his mistakes.
  15. AdamJT13

    AdamJT13 Salary Cap Analyst

    14,984 Messages
    456 Likes Received
    That's not the equation I use.

    You're missing the point. It's not about a specific equation, it's about passing effectiveness, no matter how it is measured. You can use any of the ANYPA equations. You can use passer rating. You can use unadjusted yards per pass play. You can use pass success rate or just about any other equation that measures passing effectiveness on a per-play basis. They all end up with close to the same correlations.


    Not at all. There are all kinds of equations and formulas that people use for different purposes that combine different types of factors, both in sports and outside of sports.


    Except that your equation does not measure running efficiency. It actually measures total points scored, with a small adjustment for rushing yards. All you did was give it an improper name (rushing hardly affects it at all, and it doesn't have anything at all to do with efficiency). If Team A scores 35 points and rushes 20 times for minus-20 yards, and Team B scores 34 points and rushes 20 times for 300 yards, you're really going to claim that your formula indicates that Team A was more efficient at running? That's just silly.


    That makes no sense. It doesn't even attempt to measure passing, so it can't be used to determine whether passing correlates to winning. You can't just not measure something and then say that must mean it doesn't correlate to something else.
  16. Rogerthat12

    Rogerthat12 DWAREZ

    3,701 Messages
    208 Likes Received
    This is the ONLY offseason blue print we need to follow.:eek::)

    Reference to the other 20% would be in no specific order:

    WR, RB, S, cb (Jenkins leaving).
  17. AdamJT13

    AdamJT13 Salary Cap Analyst

    14,984 Messages
    456 Likes Received
    Interceptions alone don't correlate as well as when you include other measures of passing efficiency. They also don't correlate either way in nearly 20 percent of all games, when neither team throws an interception. What determines who wins in those games?


    Not true at all. As I've said before, running the ball can be useful, even if it doesn't really matter how well you run it. And running backs obviously play a role in the passing game.
  18. Idgit

    Idgit Ice up, son. Ice up! Staff Member

    30,621 Messages
    6,386 Likes Received
    TwoDeep, I agree 100% with this. We need to be able to rely on the running game and defense in the games where we're currently unable to pass the ball effectively. Right now, we can't.
  19. percyhoward

    percyhoward Research Tool

    8,436 Messages
    2,368 Likes Received
    Dallas had 18 runs from inside the opponents' 10-yard line, and converted 6 TD.

    We could have scored on most, or every single run from inside the 10; also we could have been stopped on all but a couple, or every single one of them. The difference is how well (or how poorly) we ran it.

    If we go 10-of-18, instead of 6-of-18, that difference matters.
  20. Super_Kazuya

    Super_Kazuya Well-Known Member

    1,721 Messages
    874 Likes Received
    We also seem to have very little home run capability at the position. That's one thing I miss about Felix, at least the version of Felix that wasn't trying to become even bigger than the guy he was backing up (M. Barber).

Share This Page