Before we take stats like this seriously, there's a major confound that needs to be removed first: the effect of running out the clock. I can post the graph again if you want, but almost all the difference (on average of course, not every game) between winning and losing teams in terms of percent of rush/pass plays is in the 4th quarter when the leading team attempts to run out the clock (i.e., rush percentage as a function of score differential). You take that effect away and teams are far more similar in terms of rush percentage. It's also why you can't look at the correlation between rush% and win% and conclude much because the causal relation is often backwards. So percent of dropbacks being X standard deviations away needs to either be calculated ONLY for quarters 1-3 or relative to expected given a team's win%. Take the effect of running out the clock away and the difference won't be anywhere near as far from average/expected. League average of what? You can't compare a single game or just two games to the league-wide distribution of a stat over say 16 games. Look at the distribution of percentage of pass dropbacks for just the winning team (since you chose only wins) for every game played (not over a season) and see how many standard deviations away those two playoff games are. Might not even be 2 standard deviations away. You're comparing to the wrong distribution which is why you're getting such an impossible result.
Whether that's true or not, it has no bearing on Tannehill's dependence on such a strategy. If he was a square peg in a square hole in 2019, the question still remains whether the hole can remain square in the future. First, we haven't been measuring quarterbacks' performances here in wins. We've used passer rating predominantly. Second, during Mahomes's career his average passer rating in low-volume games (including the playoffs) is 108.6. In high-volume games it's 106.3. Those figures for Tannehill in 2019 were 119.7 (low-volume) and 90.1 (high-volume). Prior to 2019 those figures for Tannehill were 93.2 (low-volume) and 78.4 (high-volume). EDIT: Passer ratings noted above were adjusted to 2019.
This is the reason that I listed the first half stats for the run/pass ratio. The fact that the ratio in the first half was identical in Tenn wins and losses and totally skewed in the 2nd half should have been a clue that it was winning or losing that was driving the ratio and not the reverse. I've been trying for 20 posts to convince him that he has the relationship backwards. The decision to pass or run is also dependent on down and distance, offensive personnel, defensive personnel, defensive alignment, and coaching philosophy/system. I keep trying to convince him that he needs to account for these things.
Obviously that's sound thinking and I appreciate it. Here's information that relates at least in the way of a gross estimation: https://www.pro-football-reference...._gtlt=lt&c5val=1.0&order_by=score_diff_thru_3 What we see there is that in Tannehill's starts in 2019, the Titans entered the fourth quarter with a cumulative scoring margin of -1. By contrast, the quarterback with the lowest percentage of pass dropbacks in the league (Lamar Jackson) entered the fourth quarter -- during games 7 through 16 (Tannehill's starts) -- with a league-leading cumulative scoring margin of +160.
Here are more precise data regarding Tannehill's percentage of pass dropbacks during his starts in 2019. Z-scores for percentage of pass dropbacks during games 7 through 16 (Tannehill's starts) in the NFL in 2019: Tm % PD Z BAL -3.50 TEN (Tannehill) -1.44 BUF -1.30 MIN -1.06 IND -0.94 SEA -0.66 WAS -0.51 PIT -0.47 SFO -0.38 OAK -0.38 DEN -0.37 ARI -0.16 NYJ -0.11 HOU -0.04 CHI 0.05 CIN 0.08 CLE 0.11 PHI 0.14 NWE 0.18 GNB 0.21 LAR 0.24 KAN 0.24 DAL 0.39 DET 0.41 NOR 0.50 LAC 0.80 JAX 0.96 NYG 1.02 TAM 1.17 MIA 1.54 ATL 1.61 CAR 1.69 Here are some situational statistics for games 7 through 16 in 2019: Percentage of pass attempts when behind by 9+ points in the fourth quarter League average: 79% Tannehill: 71% Percentage of pass attempts when behind by 9+ points in the second half League average: 71% Tannehill: 57% Percentage of pass attempts when behind by 9 to 16 points (two scores) in the second half: League average: 72% Tannehill: 62% Percentage of pass attempts when tied or trailing in the second half: League average: 67% Tannehill: 61% Percentage of pass attempts when tied or trailing in the fourth quarter: League average: 74% Tannehill: 73% Again the Titans entered the fourth quarter during games 7 through 16 with a cumulative scoring margin of -1. By contrast, Baltimore entered the fourth quarter in games 7 through 16 with a cumulative scoring margin of +160.
You have not shown that his lower pass percent is not a function of being more effective with longer passes. Until you do, you cannot show that three percents have the meaning you're attributing to them.
Yeah.. so it's clear, what you want is to look at quarters 1-3 and ignore the 4th quarter. One can't infer what happens in quarters 1-3 through the stats you posted. Also, differential entering the 4th isn't the same as differential during the 4th (anytime in the 4th). So if you're just interested in pass dropback stats, look at z-scores for pass dropback percentages for quarters 1-3. That piece of info would be useful because it tells you how much the Titans rely on Tannehill (relative to the league) when they're not running out the clock.
I'll simply repeat here what was said in post #7663: What you're implying above is that there was a strong, negative correlation between Tannehill's YPA and his passing volume, game-by-game, in 2019 -- i.e., the greater his YPA, the lower his percentage of pass dropbacks. Here are those data for the 15 games in which he played: % Pass Dropbacks ; YPA 28.07 ; 4.8 28.3 ; 6.3 33.33 ; 9.9 36.54 ; 14.39 45.76 ; 14.48 46.94 ; 9.53 48.28 ; 8.27 51.67 ; 10.76 55.17 ; 10.07 56.72 ; 7.75 58.62 ; 6.74 63.16 ; 5.85 66.67 ; 9 67.19 ; 8.49 The above correlation is an extremely weak -0.06, whether we include the Denver game or omit it. The same is true for pass attempts and YPA: ATT ; YPA 15 ; 4.8 14 ; 6.3 20 ; 9.9 18 ; 14.39 27 ; 14.48 19 ; 9.53 22 ; 8.27 29 ; 10.76 27 ; 10.07 36 ; 7.75 31 ; 6.74 33 ; 5.85 39 ; 8.49 16 ; 9 The correlation above is an extremely weak -0.04, whether we include the Denver game or omit it.
No I'm not implying anything. You out those thoughts out there. My thought is that knowing the percent of dropbacks is meaningless without knowing the results on the scoreboard You'd also need to show that Tannehill was different from other QBs in respect to when he has his best numbers compared to attempts.
Here we are -- team-by-team percentages of plays that were pass attempts and their z-scores, quarters 1 through 3, in games 7 through the Championship round of the playoffs: 65 1.53 65 1.53 64 1.31 64 1.31 63 1.09 63 1.09 62 0.87 61 0.65 61 0.65 61 0.65 61 0.65 61 0.65 60 0.43 59 0.21 59 0.21 58 -0.01 58 -0.01 58 -0.01 57 -0.23 57 -0.23 57 -0.23 56 -0.45 56 -0.45 55 -0.66 55 -0.66 55 -0.66 54 -0.88 53 -1.10 53 -1.10 51 -1.54 49 -1.98 (Tannehill) 46 -2.64
He also hasn't shown what personnel and formations were used on offense or defense, how often teams checked into run or pass vs being the initial play call, down and distance, the score, etc, etc, etc. At the points in the game where the scoreboard is not dictating play calling pretend Team A runs the ball on first down and gains 6 yards. Team B runs the ball on first down and gains 1. Regardless of the QBs on Team A or Team B, Team A is more likely to run the ball on second down. The play call is NOT reflective of the QB's ability (which is the connection he is trying to make). In fact, the other factors far outweigh the talent of the QB when deciding on the run/pass mix. Why do I think that? Because Miami ran the second highest percentage of passing plays in the league - MIAMI. The team with Fitz and Rosen...... The top 10 were: ATL, MIA, NYG, CAR, CIN, LAC, TB, LAR, JAC, KC. Two of those teams started rookies and five of them are changing QBs. Only two of those teams have what I consider top QBs. Now look at the bottom 10: BAL, SF, TENN, MIN, INDY, SEA, BUF, OAK, DEN, PIT. Six of those teams made the playoffs and are currently feeling set at QB. One started a rookie. Two are making QB changes (including PIT getting Roethlisberger back). BTW, if I add the teams that were tied with the 10th teams in the above lists, I would add CHI to the list of teams with the highest % of passing plays and HOU, and DAL to the list of teams passing least often. Why the hell would anyone try to use run/pass ratio as a means to measure QB quality or ability? It is mind numbingly stupid.
No one is. We're using it as one measure of the degree of difficulty imposed upon the QB. The more the offense is on the QB's shoulders, the greater the degree of difficulty of his performance. Certainly you don't think that with all else equal, two passer ratings of 117.5 are equivalent when one of them was produced in an environment with a high degree of difficulty and the other was produced in an environment with a low one.
OK, without checking the data myself, that would support a common narrative about the 2019 Titans, namely that they relied on Henry a lot. I don't think there's any issue with saying that. And yes, for one year Tannehill performed at an elite level in a situation where the Titans had a great running game and he didn't have to pass as much. Not sure you can go beyond that. I mean.. it's just one year. It's one thing if you could show Tannehill played at a similar level in Miami when he had a similar distribution of passing attempts in quarters 1-3, but that's clearly not true. So I don't think there's much more to say here. Too many hypotheses fit the data, we need more data, and that's IMO a good way to temporarily end the thread. lol.
Apparently you're not seeing how the pieces of the puzzle fit together. And when you express your ignorance in that manner, I'm not as willing to explain myself (again).
LOL. You claimed your nonsense about run/pass ratios was not about measuring a QBs ability. I pulled a direct quote from you claiming that is exactly what it is about. Exactly. I can’t tell if you are a liar and assume nobody will fact check you or if you changed positions and theories so often that you can’t keep it all straight.
He had the 8th lowest expected completion percentage in the league. Surely you don’t think represents a low degree of difficulty...... How many times do I need to remind you of that? from NFL next gen stats Expected Completion Percentage (xComp) gives an indication of the level of difficulty of a quarterback's throws. A high xComp indicates a quarterback's propensity to throw higher probability passes, while a low xComp tells us a quarterback has a tendency to take more chances. As discussed in our introduction to Completion Probability, the level of difficulty of a specific throw is determined by in-play features specific to each play. xComp is simply the average Completion Probability within a given set of passes.
And that as well occurred within the general context of an extremely light offensive load on his shoulders.
Expected completion percentage measures the difficulty of the throws. Period. The context is irrelevant. A 30 yard throw to a blanketed receiver on the sideline while getting pressured is a difficult throw no matter the context.
Certainly you don't think that a quarterback who is responsible for passing the ball on 65% his team's offensive plays, and one who's responsible for passing the ball on a mere 49% of his team's offensive plays, are sharing the same experience of the level of demand placed upon them to carry their offense.
Please explain how the difficulty of the throw I described changes. I’m sure the NFL will be interested to hear how they should change their statistics.
Yeah I'm not gonna speculate about a mechanism of action that can be neither confirmed or denied, on a message board full of hostility. Suffice it to say that Tannehill experienced an exceptional passer rating alongside an exceptionally light load. We'll see what happens from here on out.
You have lost the argument. You are not going to switch to a new one. Tannehill made, on average, the 8th most difficult throws in the league and did so with the league’s highest efficiency. At least have the decency to stop playing games.
This is an argument that no one can win or lose until there are more data. What we're doing now is determining whether he was experiencing a possible situational advantage in 2019. It'll be up to 2020 and beyond to determine the true meanings of all of these findings.
How many times are you planning to change what you are doing? It used to be about trying to determine Tannehill’s ability. I shot that one down and it’s on to something new...... LOL. Denying what expected completion percentage means is unbelievable, even for you.
Maybe the high percentage guys are throwing more easy to complete passes? I mean, that intuitively makes sense, because the farther the ball travels the harder to complete.
You still haven't shown that the lower percent of pass plays wasn't a result of him throwing deeper than almost everybody at a more efficient rate.
That is exactly what expected completion percentage measures. He screwed himself when he made the claim that he was simply using run/pass ratio as a way to measure level of difficulty imposed on a QB. The league has a ****ing stat designed solely for that purpose. that was after lying about whether he was trying to use run/pass ratio as a way to measure a QB’s ability. He had to drop that line of argument when I pointed out the terrible QBs that had high pass ratios. LOL.
And this point he has demonstrated that he will lie about what he has previously written and will simply switch to a new argument (still something negative about Tannehill) when his current argument fails. He has also been caught cherry-picking stats (repeatedly) and purposefully misrepresenting data. easily the most disingenuous style of posting that I have ever seen.
The problem with the CPOE statistic is that Tannehill didn't demonstrate the same pattern of performance with regard to it 2012-2018. It's also the case that Andy Dalton's CPOE was 4.6 in 2015 (his one great year), whereas for his career overall it's 0.3. Likewise, Nick Foles's CPOE in 2013 (his one great year) was 5.4, whereas for his career overall it's 0.1 So we can hardly establish CPOE as a measure of a quarterback's ability when he's exhibited only one good season with regard to it. How do we know at this point that Tannehill isn't simply Dalton or Foles?
I believe I showed that in post #7687. So you tell me how you believe we should measure "throwing deeper than almost everybody at a more efficient rate" if it isn't with YPA, i.e., post #7687.
That wasn't directed at resnor. He (resnor) and FinFaninBuffalo are debating "The Guy". If that doesn't seem apparent, I'm guessing you've been blocked by "The Guy".
The only problem is you going in circles trying to discredit Tannehill. Didn't you already point out that CPOE only goes back to 2016? Ah yeah, you did: Now how are you getting CPOE values from before that? Another lie perhaps?
Two Tacos probably nailed the explanation. Resnor and I have been completely simpatico during this whole debate. It's all good.
I provided this link earlier in the thread: https://rbsdm.com/stats/stats/ The other thing to remember is that Dalton in 2015 and Foles in 2013 both had unusually strong correlations (compared to the league norm) between passer rating and yards per rush in the run game, game-by-game, just like Tannehill in 2019. There's an awful lot to go on here in hedging one's bets regarding Tannehill in 2020.
Other findings in this vein: In 2015 Dalton's number of pass attempts per game was a standard deviation below the league average that year. In 2013 Foles's number of pass attempts per game was 1.72 standard deviations below the league average that year. Like Tannehill in 2019, both Dalton and Foles in their one great year experienced similarly lighter loads than the league norm. So if you wanted to make a case against betting on Tannehill in 2020, you could simply nominate the parallels between his 2019 season and those of Dalton in 2015 and Foles in 2013.
Again, you haven't shown how that relates to other QBs. So wet know that Tannehill last season threw deeper more efficiently. But you are saying that throwing fewer passes makes him less of a QB. I'm saying, it makes sense that he would throw fewer passes if he was throwing longer passes and completing then more than anyone else. I'm not sure why that's so hard for you to understand.
You tell me what you believe the numbers may show regarding Tannehill in comparison to other QBs in this regard, and the passing statistic(s) we should use to explore that, and I'll be nice enough to do the mathematical work. From now on when I do mathematical work to investigate somebody else's hypothesis, we're gonna get all that spelled out in advance, before I spend any time on it. If you don't want to supply that information, that's fine. We can just move on from you and your hypothesis.
