CFZ Dak Interceptions - Regression to the Mean (Warning: Math Geek)

[Jumbo075]

In Dak’s first 6 seasons, he started 85 games, and threw 50 interceptions - an average interception rate of 0.588 interceptions per game. Over a 17 game season, the expected interceptions Dak should throw is 10 interceptions per 17 game season. But last year, in only 12 games, Dak threw 1.25 interceptions per game for a total of 15 interceptions. 2022 was obviously an outlier, with Dak throwing interceptions at more than twice his normal rate.

Statistically, barring outside factors, there is a principal called “Regression to the Mean. It means that an unusually high variation will likely be counter-balances by an unusually low variation in the variable (in this case interceptions) to bring the average back down to its historical value.

If Dak interceptions regress to the mean, that means that over the next two seasons, he’ll only throw about 12 interceptions, which is about 6 per season, assuming he plays all 17 games in 2023 & 2024.

Here is the math:

• 12 games (2022) + 34 games (2023 & 2024) = 46 games.
• Average of .588 interceptions per game x 46 games = 27 interceptions over 46 games.
• 27 interceptions - 15 interceptions already thrown = 12 interceptions spread out over the next 34 games

Unless Dak Prescott suddenly got worse, or there was some outside factor that affected his play in 2022, Dak's average interception rate should revert back to what it has been over his first 85 games, which means he'll throw fewer interceptions to offset what happened in 2022.

The alternative few, still assuming that Dak's historical career average of interceptions over his first 85 games in a true picture of what we can expect, is that there was an anomaly in 2022. Some will say, maybe justifiably, that Dak's receiver corps was suspect in 2022, which is an outside factor that would affect him. If that is the case, then we should expect Dak to revert back to his normal rate, which means he'll throw about 10 interceptions in a 17 game season.

Bottom line: Unless Dak has suddenly got worse, we should expect him to throw around 6-10 interceptions in 2023, and the same in 2024. Even if you include the anomalous 2022 season, Dak's current career interception rate is 65 interceptions over 97 games played = 0.67 interceptions per game. The high end of that is 11 expected interceptions in 2023 over an entire 17 game schedule.

Now, we'll all see what actually occurs during the season, but based on Dak's historical performance, I'm not too concerned about interceptions in 2023. I do believe that Michael Gallup was hobbled last year, and I wasn't ever a huge fan of Dalton Schultz, who was competent, but not special. The addition of Cooks, the health improvement of Gallup, and the development of Tolbert bodes well for Dak's performance in 2023 - at least during the season.

 

[shabazz]

Im ready to move on from last years mess…..

 

[doomsday9084]

I don't think its really that hard to figure out. Both him and Moore said that they wanted to be more aggressive with the throws. Instead of dumping off, Dak was frequently throwing into tighter covered routes. If the receiver stepped wrong, the ball got bobbled or if Dak was a little off with the throw it was far more likely to get intercepted than earlier in his career when Dak was more careful with the ball (which lead to the Dink and Dak complaints).

 

[Captain-Crash]

I think fat Mike wanted dak to take more chances and we saw the results.

 

[noshame]

Unfortunately, all things are not equal.
From season 2 season

2days2go

 

[Jarv]

In Dak’s first 6 seasons, he started 85 games, and threw 50 interceptions - an average interception rate of 0.588 interceptions per game. Over a 17 game season, the expected interceptions Dak should throw is 10 interceptions per 17 game season. But last year, in only 12 games, Dak threw 1.25 interceptions per game for a total of 15 interceptions. 2022 was obviously an outlier, with Dak throwing interceptions at more than twice his normal rate.

Statistically, barring outside factors, there is a principal called “Regression to the Mean. It means that an unusually high variation will likely be counter-balances by an unusually low variation in the variable (in this case interceptions) to bring the average back down to its historical value.

If Dak interceptions regress to the mean, that means that over the next two seasons, he’ll only throw about 12 interceptions, which is about 6 per season, assuming he plays all 17 games in 2023 & 2024.

Here is the math:

• 12 games (2022) + 34 games (2023 & 2024) = 46 games.
• Average of .588 interceptions per game x 46 games = 27 interceptions over 46 games.
• 27 interceptions - 15 interceptions already thrown = 12 interceptions spread out over the next 34 games

Unless Dak Prescott suddenly got worse, or there was some outside factor that affected his play in 2022, Dak's average interception rate should revert back to what it has been over his first 85 games, which means he'll throw fewer interceptions to offset what happened in 2022.

The alternative few, still assuming that Dak's historical career average of interceptions over his first 85 games in a true picture of what we can expect, is that there was an anomaly in 2022. Some will say, maybe justifiably, that Dak's receiver corps was suspect in 2022, which is an outside factor that would affect him. If that is the case, then we should expect Dak to revert back to his normal rate, which means he'll throw about 10 interceptions in a 17 game season.

Bottom line: Unless Dak has suddenly got worse, we should expect him to throw around 6-10 interceptions in 2023, and the same in 2024. Even if you include the anomalous 2022 season, Dak's current career interception rate is 65 interceptions over 97 games played = 0.67 interceptions per game. The high end of that is 11 expected interceptions in 2023 over an entire 17 game schedule.

Now, we'll all see what actually occurs during the season, but based on Dak's historical performance, I'm not too concerned about interceptions in 2023. I do believe that Michael Gallup was hobbled last year, and I wasn't ever a huge fan of Dalton Schultz, who was competent, but not special. The addition of Cooks, the health improvement of Gallup, and the development of Tolbert bodes well for Dak's performance in 2023 - at least during the season.

I think it's fair to say 2016 and 2022 were both outliers. The better average would be looking at 2017-2021.

 

[BleedSilverandBlue]

There is no reason to think Dak will throw interceptions again like he did last year, especially with a switch to a new offensive philosophy. Compared to the rest of his career it was a total anomaly. If Dak ever did anything at an elite level, it was that he protected the ball extremely well. I would be pretty shocked if he led the lead in INTs again, or even came close.

 

[G2]

The INT's have been blown out of proportion. No one is talking about turning the ball over and losing the Super Bowl, but INTs (whoever you want to blame) is still a thing.

 

[MyFairLady]

Very simple. Early on his reads were simple, he did not take chances, if it wasn't there he took off. Then he got hurt. Then he signed his big deal. Now his reads are more difficult, he takes more chances, if it isn't there he stays in the pocket and gets sacked or forces it. The difference in his play is obvious. The impact it has on INTs is obvious.

 

[Vtwin]

I don't think its really that hard to figure out. Both him and Moore said that they wanted to be more aggressive with the throws. Instead of dumping off, Dak was frequently throwing into tighter covered routes. If the receiver stepped wrong, the ball got bobbled or if Dak was a little off with the throw it was far more likely to get intercepted than earlier in his career when Dak was more careful with the ball (which lead to the Dink and Dak complaints).

That's exactly what happened.
It makes the math inconsequential unless Dak reverts to the mean of his previous style.

 

[JD_KaPow]

Why would you do the math as interceptions per game instead of per attempt?

From 2018-2021, his INT/attempt rates were 0.015, 0.018, 0.018 and 0.017. In 2022, it was 0.038.

Obviously, your point about regression to the mean stands. But straight-ahead regression assumes nothing has fundamentally changed. If, for instance, Dak is actually a worse QB now, because of age or other factors, you wouldn't expect as much regression. And as others have pointed out, if they have changed their risk-reward calculus, you again wouldn't expect as much regression.

Fortunately, I think the biggest factors point the other way. The WR corps was godawful last year, worst in the league at separation and constantly knocking catchable balls into the air. The WR group looks so much better this year that it's not even funny. I expect his INT rate to regress fully back to his career averages. The wild card is the offensive changes MM will bring, but if it's more West-Coasty, that generally means more short high-percentage passes, which would also tend to drive INT rate down.

If Dak plays the full season, he'll probably throw around 600 passes. At his pre-2022 rate, that would mean about 10 INTs on the season, which is pretty much exactly what happened in 2021 and 2019.

 

[Captain-Crash]

I hope he goes back to handing the ball off and throwing short 4-yard passes, that should be worth 50 million. Not many qb can do that.

 

[Jumbo075]

Why would you do the math as interceptions per game instead of per attempt?

From 2018-2021, his INT/attempt rates were 0.015, 0.018, 0.018 and 0.017. In 2022, it was 0.038.

Obviously, your point about regression to the mean stands. But straight-ahead regression assumes nothing has fundamentally changed. If, for instance, Dak is actually a worse QB now, because of age or other factors, you wouldn't expect as much regression. And as others have pointed out, if they have changed their risk-reward calculus, you again wouldn't expect as much regression.

Fortunately, I think the biggest factors point the other way. The WR corps was godawful last year, worst in the league at separation and constantly knocking catchable balls into the air. The WR group looks so much better this year that it's not even funny. I expect his INT rate to regress fully back to his career averages. The wild card is the offensive changes MM will bring, but if it's more West-Coasty, that generally means more short high-percentage passes, which would also tend to drive INT rate down.

If Dak plays the full season, he'll probably throw around 600 passes. At his pre-2022 rate, that would mean about 10 INTs on the season, which is pretty much exactly what happened in 2021 and 2019.

I simply didn't think about doing the analysis on a per attempt basis. But I agree, it is likely a better way to do it. The point about Dak's 2022 season being anomalous, still stands, as you demonstrated.

One of my biggest complaints about modern media is that they are both mathematically and scientifically illiterate. I fail to understand how someone who cannot even understand that data can be relied upon to report accurately. I give a pass to fans. But you would think that reporters would be better at their profession - or at least want to be better. But it seems knowledgeable reporting is not as highly valued as click-bait headlines. It is why I like doing my own analysis, and don't pay too much attention to the talking opinion heads. I'm fine with "reporters" who simply report the facts. But so-called "journalists" or "editorialists" who fail to even understand what they are talking about are worthless. This applies across the board to every type of reporting, but it definitely includes the talking heads like Cowherd, SAS, Skip Bayless, and all the wanna be broadcasters who offer worthless opinions each day.

 

[CCBoy]

Why would you do the math as interceptions per game instead of per attempt?

From 2018-2021, his INT/attempt rates were 0.015, 0.018, 0.018 and 0.017. In 2022, it was 0.038.

Obviously, your point about regression to the mean stands. But straight-ahead regression assumes nothing has fundamentally changed. If, for instance, Dak is actually a worse QB now, because of age or other factors, you wouldn't expect as much regression. And as others have pointed out, if they have changed their risk-reward calculus, you again wouldn't expect as much regression.

Fortunately, I think the biggest factors point the other way. The WR corps was godawful last year, worst in the league at separation and constantly knocking catchable balls into the air. The WR group looks so much better this year that it's not even funny. I expect his INT rate to regress fully back to his career averages. The wild card is the offensive changes MM will bring, but if it's more West-Coasty, that generally means more short high-percentage passes, which would also tend to drive INT rate down.

If Dak plays the full season, he'll probably throw around 600 passes. At his pre-2022 rate, that would mean about 10 INTs on the season, which is pretty much exactly what happened in 2021 and 2019.

That is reflective more on the effect it has towards a team's path to and through the playoffs.

 

[CCBoy]

I simply didn't think about doing the analysis on a per attempt basis. But I agree, it is likely a better way to do it. The point about Dak's 2022 season being anomalous, still stands, as you demonstrated.

No, you are observer friendly already...base level numbers aren't management and application efficient. Once a number happens, the numbers are askewed.

Instead consistency and level of play is most important in team play and leadership applications.

End result then is: Target, cease fire!

 

[Captain-Crash]

it's like a speckled chicken sitting on an old rusty post looking eastward. Red dog meowing. tango charlie, chicken chow mein.

 

[Starstruck22]

In Dak’s first 6 seasons, he started 85 games, and threw 50 interceptions - an average interception rate of 0.588 interceptions per game. Over a 17 game season, the expected interceptions Dak should throw is 10 interceptions per 17 game season. But last year, in only 12 games, Dak threw 1.25 interceptions per game for a total of 15 interceptions. 2022 was obviously an outlier, with Dak throwing interceptions at more than twice his normal rate.

Statistically, barring outside factors, there is a principal called “Regression to the Mean. It means that an unusually high variation will likely be counter-balances by an unusually low variation in the variable (in this case interceptions) to bring the average back down to its historical value.

If Dak interceptions regress to the mean, that means that over the next two seasons, he’ll only throw about 12 interceptions, which is about 6 per season, assuming he plays all 17 games in 2023 & 2024.

Here is the math:

• 12 games (2022) + 34 games (2023 & 2024) = 46 games.
• Average of .588 interceptions per game x 46 games = 27 interceptions over 46 games.
• 27 interceptions - 15 interceptions already thrown = 12 interceptions spread out over the next 34 games

Unless Dak Prescott suddenly got worse, or there was some outside factor that affected his play in 2022, Dak's average interception rate should revert back to what it has been over his first 85 games, which means he'll throw fewer interceptions to offset what happened in 2022.

The alternative few, still assuming that Dak's historical career average of interceptions over his first 85 games in a true picture of what we can expect, is that there was an anomaly in 2022. Some will say, maybe justifiably, that Dak's receiver corps was suspect in 2022, which is an outside factor that would affect him. If that is the case, then we should expect Dak to revert back to his normal rate, which means he'll throw about 10 interceptions in a 17 game season.

Bottom line: Unless Dak has suddenly got worse, we should expect him to throw around 6-10 interceptions in 2023, and the same in 2024. Even if you include the anomalous 2022 season, Dak's current career interception rate is 65 interceptions over 97 games played = 0.67 interceptions per game. The high end of that is 11 expected interceptions in 2023 over an entire 17 game schedule.

Now, we'll all see what actually occurs during the season, but based on Dak's historical performance, I'm not too concerned about interceptions in 2023. I do believe that Michael Gallup was hobbled last year, and I wasn't ever a huge fan of Dalton Schultz, who was competent, but not special. The addition of Cooks, the health improvement of Gallup, and the development of Tolbert bodes well for Dak's performance in 2023 - at least during the season.

Dak is an interception machine due to poor judgement, poor ability to see the field, and poor passing. There will be interceptions in the Giants game. Bank on it. It is Dak’s modus operandi. The argument that because he has been bad, he cannot continue to be bad or get worse is nonsense. I just do not believe
Dak is an upper echelon QB due to his intellect. He is really not very smart or quick smart like upper tier guys.

 

[OGSixshooter]

Dak is an interception machine due to poor judgement, poor ability to see the field, and poor passing. There will be interceptions in the Giants game. Bank on it. It is Dak’s modus operandi. The argument that because he has been bad, he cannot continue to be bad or get worse is nonsense. I just do not believe
Dak is an upper echelon QB due to his intellect. He is really not very smart or quick smart like upper tier guys.

The last two sentences seem contradictory.

When did Dak become an "interception machine"? If so, how did he take Romo's job and then keep it and secure a new contract?

What makes you think Dak is slow? It's been well-reported he has a photographic memory and is over-prepared for games.

Dak has the same issue Romo had and many people who are "surprise" starters have: Legitimacy due to heightened expectations. It's one thing to win the job, it's another to be told to the win the SB. It has nothing to do with his intelligence, football intelligence or otherwise. I will even agree that he doesn't have the most "rubber" or flexible of arms. But intelligence? How would you know?

 

[Starstruck22]

The last two sentences seem contradictory.

When did Dak become an "interception machine"? If so, how did he take Romo's job and then keep it and secure a new contract?

What makes you think Dak is slow? It's been well-reported he has a photographic memory and is over-prepared for games.

Dak has the same issue Romo had and many people who are "surprise" starters have: Legitimacy due to heightened expectations. It's one thing to win the job, it's another to be told to the win the SB. It has nothing to do with his intelligence, football intelligence or otherwise. I will even agree that he doesn't have the most "rubber" or flexible of arms. But intelligence? How would you know?

The San Fran game was the perfect example he had a wide open Hilton for a Td and he chose i think it was lamb who was blanketed and it was an incompletion. Plus started off game with 2 inexplicable pics in coverage. All due to stupidity. People blamed Moore but Moore is not making stupid snap judgments Dak is. I am predicting one or 2 pics in the Giants game. It will happen. Bank on it. Then we can blame McCarthy.

 

[Mac_MaloneV1]

Very simple. Early on his reads were simple, he did not take chances, if it wasn't there he took off. Then he got hurt. Then he signed his big deal. Now his reads are more difficult, he takes more chances, if it isn't there he stays in the pocket and gets sacked or forces it. The difference in his play is obvious. The impact it has on INTs is obvious.

Obviously wrong. The offense wasn't much different in 2022 than it was in 2019.