A starting point for projecting receiving yards (1 Viewer)

[Chase Stuart]

My article here: http://subscribers.footballguys.com/apps/article.php?article=stuart_projecting_receiving_yards_2014

Let me know if you have any questions or comments.

 

[Biabreakable]

Interesting stuff as always Chase.

In 2013 there were 18136 pass attempts

Your projection for all 32 teams is 17482 passing attempts.

654 less passing attemps than 2013

The average passing attempts/team in 2013 was 566.8 pass attempts. Your average is 546.3

9 teams exceeded 600 passing attempts in 2013 (28% of the teams) you have zero repeating that.

Considering the trend seems to be passing attempts and total plays increasing each year over the past 5 it makes me wonder why you could project all teams to have such a decline in total passing attempts?

 

[Chase Stuart]

Interesting stuff as always Chase.

In 2013 there were 18136 pass attempts

Your projection for all 32 teams is 17482 passing attempts.

654 less passing attemps than 2013

The average passing attempts/team in 2013 was 566.8 pass attempts. Your average is 546.3

9 teams exceeded 600 passing attempts in 2013 (28% of the teams) you have zero repeating that.

Considering the trend seems to be passing attempts and total plays increasing each year over the past 5 it makes me wonder why you could project all teams to have such a decline in total passing attempts?

Good question. I'd project 8-10 teams to throw 600+ times in 2014, but the odds of any one team doing it are still not great. As you know, there's a big difference between group projections and individual projections. Anyway, the formula to projecting pass attempts is:

Future Pass Attempts = 36 + (450 x Pass_Attempts/Play) + (0.255 x Offensive Plays)

Since that's based off of historical numbers, it's smart to wonder how that will change as teams become more pass happy. Which of those three numbers will change? If it's the 36 jumping to say, 56, that means a rising tide lifts all ships. If it's the 450 going up to say, 480, it means we think the most pass-happy teams will continue to be the most pass-happy: personally, I'm skeptical of that assumption. If it's the 0.255 number, it's something of a blend between the two. Number of plays is pretty fickle from year to year, but some high-tempo teams can be counted on to run a high number of plays.

Since pass attempts is not just a function of pass identity but of game scripts, I think I'd be more likely to change the 36 or 0.255 number than the 450. Stafford threw 93 fewer times in 2013 than in 2012, which wasn't something most predicted.

It would involve a bunch of guesstimates and subjectivity to tweak the formula to incorporate the idea that teams will pass more in 2014 than they have over the past X number of years. There's nothing wrong with that -- it's probably the right thing to do -- but that's not how the formula was derived. But if I was to adjust it, I'd probably bump the 36 up to about 50, and then slightly increase the 450 and 0.255 numbers.

 

[Biabreakable]

Hmm

What is the time frame of your history being used?

Just looking at PFR over the past 5 years there has been a league wide increase of 200-300 passing attempts each season. The 300 jumps happening the last 2 seasons. So that seems to be building even higher.

2013

18136 passing attempts

The average passing attempts/team in 2013 was 566.8

9 teams exceeded 600 passing attempts in 2013
2012

17788 passing attempts

The average passing attempts/team in 2012 was 555.9

9 teams exceeded 600 passing attempts

2011

17410 passing attempts

The average passing attempts/team in 2011 was 544.1

3 teams exceeded 600 passing attempts

2010

17269 passing attempts

The average passing attempts/team in 2010 was 539.7

4 teams exceeded 600 passing attempts

2009

17033 passing attempts

The average passing attempts/team in 2009 was 532.3

2 teams exceeded 600 passing attempts

So what would be a good way to try to capture this rate of progression in passing attempts? When does this finally reach its limit? WIll it reach a limit? How is this affecting rushing attempts?
I incorrectly thought that teams might run a bit more last season than they ended up doing. I do expect that to happen at some point. At least an adjustment where the tide turns back more towards balance. But that is not what seems to keep happening. Now with the 2014 draft we have an influx of high quality WR talent and not nearly as many RB. The personnel seems to fit that trend continuing as well.

 

[jurb26]

So this approach only uses historical information as the baseline and doesn't factor in changes from year to year? Seems a bit off. Coaching changes, coordinator changes, personal changes, ect can all greatly impact passing numbers.

 

[Chase Stuart]

Hmm

What is the time frame of your history being used?

...
So what would be a good way to try to capture this rate of progression in passing attempts? When does this finally reach its limit? WIll it reach a limit? How is this affecting rushing attempts?
I incorrectly thought that teams might run a bit more last season than they ended up doing. I do expect that to happen at some point. At least an adjustment where the tide turns back more towards balance. But that is not what seems to keep happening. Now with the 2014 draft we have an influx of high quality WR talent and not nearly as many RB. The personnel seems to fit that trend continuing as well.

I looked at data from 2003 to 2012. There's always a tradeoff between a larger sample and a more relevant sample. For projecting target percentage and yards per target, I think the larger sample makes sense. For projecting team pass attempts, perhaps a smaller sample makes sense.

Another way to get around this and still use a larger sample would be to use league average data instead of raw data. That's a bit more work and makes the formula pretty complicated, but it may be worth the effort and added layers of confusion. Alternatively, I could use a fudge as described above. I'm open to ideas.

As for your other question, a model is only as good as its inputs. I don't think any model can tell us when, if ever, teams will begin running more. Frankly, I'm a bit skeptical that will happen anytime soon. If anything, the gap between rushing efficiency and passing efficiency seems to be widening.

 

[Chase Stuart]

So this approach only uses historical information as the baseline and doesn't factor in changes from year to year? Seems a bit off. Coaching changes, coordinator changes, personal changes, ect can all greatly impact passing numbers.

Correct. That's why it's a starting point for your wide receiver projections. I think it's more useful to begin with these numbers then say, last year's numbers, which is how most people begin their projections.

 

[Biabreakable]

The average pass attempts over the past 5 years is 547.76 so very close to your average of 546.3

So even though you used a longer time frame (which mostly coincides with the enforcement of the no chuck rule which is nice) it still ended up being very close to what it has been the past 5 years.

I guess I am being fooled somewhat by the total passing attempts going up and no real reason to expect them to fall back again at some point because of the prevailing trend. Maybe Manning/Brady retiring will help that cause some at some point.

Sorry to be stuck here on the 1st step. This might be something better to check at the end of the process than the beginning.

I was thinking about this some more.

Calculated the average % increase in pass attempts over the last 5 seasons which is + 1.579%pa

I also calculated the rate of change over the 5 year period which is + 6.475%

So thinking of both of those numbers as a high a low expectation based on the average rate of change in passing attempts over the past 5 seasons.

There may be a better way to incorporate that idea into your formula. But that is how I was looking at this.

Using your numbers what that would mean is

1 CLE 63.2 1078 681 595 -86 +9pa +38
2 DEN 58.4 1156 675 594 -81 +9pa +38
3 ATL 64.4 1024 659 587 -72 +9pa +38
4 NOR 60.3 1079 651 583 -68 +9pa +38
5 DET 57.5 1102 634 576 -58 +9pa +37
6 HOU 58.1 1089 633 575 -58 +9pa +37
7 NWE 55.2 1138 628 575 -53 +9pa +37
8 BAL 56.8 1090 619 570 -49 +9pa +37
9 WAS 55.2 1107 611 567 -44 +8pa +36
10 MIA 59.3 1001 594 558 -36 +8pa +36
11 JAX 58 1020 592 557 -35 +8pa +36
12 CIN 53.5 1097 587 557 -30 +8pa +36
13 DAL 61.2 957 586 556 -30 +8pa +36
14 PIT 57.3 1023 586 555 -31 +8pa +36
15 IND 56.9 1023 582 553 -29 +8pa +36
16 CHI 57.2 1013 579 552 -27 +8pa +36
17 ARI 55.4 1037 574 550 -24 +8pa +35
18 GNB 53.1 1074 570 549 -21 +8pa +35
19 NYG 57.4 988 567 546 -21 +8pa +35
20 SDG 51.3 1060 544 537 -7 +8pa +35
21 KAN 53.1 1029 546 537 -9 +8pa +35
22 MIN 53.9 1013 546 537 -9 +8pa +35
23 TEN 51.6 1032 533 532 -1 +8pa +34
24 BUF 46.8 1116 522 531 9 +8pa +34
25 OAK 51.9 1000 519 525 6 +8pa +34
26 TAM 52.4 981 514 522 8 +8pa +34
27 PHI 48.2 1054 508 522 14 +8pa +34
28 STL 52.3 968 506 518 12 +8pa +33
29 NYJ 47.1 1020 480 508 28 +8pa +33
30 CAR 47.3 999 473 504 31 +7pa +32
31 SEA 43.2 973 420 478 58 +7pa +31
32 SFO 43.4 961 417 476 59 +7pa +31

The main issue I have with your total passing attempt numbers is that they are based too much on 2013 run/pass distribution and gives no consideration for coaching and personnel changes that will likely affect that distribution.

For example I do not expect Cleveland to throw the ball 63.2% of the time again in 2013. Not with likely a rookie QB starting at some point during the season, a new coach with a background of running the ball and their main weapon Josh Gordon missing 8 games to the whole season.

Here is the rushing frequency for Kyle Shannahan as an OC

2013 WAS .409
2012 WAS .522
2011 WAS .387
2010 WAS .350
2009 HOU .407
2008 HOU .424

So the average run % for Kyle Shannahan is 41.65%

For Cleveland if we just based this off of 2013 plays of 1078 (as you did) this would mean 629 pass attempts.

I really do not think Cleveland will approach 600 passing attempts due to the changes that have happened there. I do expect them to run the ball more than 37% of their plays. If Manziel is successful as a runner in a read option offense the run/pass distribution may be closer to 2012 than the others. That is the best example of what it looks like Cleveland will try to do. I would expect the total overall plays to decrease because of higher frequency of running the ball. In 2012 Washington ran less than 1000 total plays.

 

[IHEARTFF]

Interesting topic. What if you used the same timeframe with a heavier weight on the last 3 years or something similar?

 

[Anarchy99]

Total pass attempts per season . . .

2013 18136 (+2.0% vs. prior year)

2012 17788 (+2.2% vs. prior year)

2011 17410 (+0.8% vs. prior year)

2010 17269 (+1.4% vs. prior year)

2009 17033 (+3.1% vs. prior year)

2008 16526

Looks like the trend continued to have passing attempts rise. I suspect that some of that comes from a handful of teams trying to max out how many plays they can run each game. If the trend continue, attempts would go up another 1.9% to around 18480.