A starting point for projecting receiving yards for WRs (1 Viewer)

[Chase Stuart]

My piece on projecting receiving yards for WRs is now up: http://subscribers.footballguys.com/apps/article.php?article=stuart_projecting_receiving_yards

The key word is starting point: this shouldn't be what you use to draft, but helps to inform a discussion of projections.

Receiving yards is mathematically equivalent to the following equation:

(Team Pass Attempts) + (Targets/Team Pass Attempt) + (Yards/Target)

Why would you want to do that? Because those three variables have different levels of sustainability. The middle one -- aka, a player's share of his team's pie -- is the easiest of the three to predict. In broad terms, receivers who stay on the same team generally average about the same percent of the targets pie every year.

Yards/Target, on the other hand, is not very easy to predict. In fact, it's all over the map from year to year. This would be good news for Larry Fitzgerald, although it's not like anyone has him outside of the top 40 for their 2013 projections.

Team Pass Attempts is in the middle. Teams tend to regress to the mean, but I'd say it's more 'sticky' than yards per target.

Again, these are only baseline projections. You might see this and say "well the average team that throws a ton of passes tends to regress to the mean, but Team X won't." But it still helps to understand the averages and then make sure you have a good reason for projecting something different.

 

[Spin]

Good read. Quick question, how do you apply this for Rookies?

 

[Chase Stuart]

Good read. Quick question, how do you apply this for Rookies?

It really wouldn't. Glad you enjoyed.

 

[Tau837]

Chase, one thing I noticed was the absence of Harvin in your lists in the article. You are using % of targets by applying whole season numbers only. While I understand that this is intended to only provide a starting point for projections, it seems like a flaw that a guy like Harvin would not be listed at all. He had 29.9% of his team's targets in the games he played (games 1-9).

 

[Lord Fantasy]

Thank you very much for this article. When you're excited about a player you can't help but want their talent, ability and/or situation to translate into huge fantasy numbers. But there are only so may minutes/plays/attempts/targets/etc. to go around. Putting constraints on projections can be a bummer, but it's exactly the dose of cold water we need this time of year.

 

[Chase Stuart]

Chase, one thing I noticed was the absence of Harvin in your lists in the article. You are using % of targets by applying whole season numbers only. While I understand that this is intended to only provide a starting point for projections, it seems like a flaw that a guy like Harvin would not be listed at all. He had 29.9% of his team's targets in the games he played (games 1-9).

Yeah, I just tried to keep it simple. Had Harvin stayed with the Vikings, I was going to do an entire section on how to project him, but with him in Seattle now, the whole thing seemed moot.

It would be a bit of extra work to do it on a per-game, or per-adjusted game basis. It was doable, but didn't seem worth the extra hour or two.

 

[Chase Stuart]

Thank you very much for this article. When you're excited about a player you can't help but want their talent, ability and/or situation to translate into huge fantasy numbers. But there are only so may minutes/plays/attempts/targets/etc. to go around. Putting constraints on projections can be a bummer, but it's exactly the dose of cold water we need this time of year.

Thanks Lord. I'm glad you found it useful.

 

[meyerj31]

This is a GREAT article man. Nice job!

Although there are some clear cases where you wouldn't use the projected totals. Does anyone think Cobb is going to get only 864 yards this year?

I think this is most useful for the top-end projections. I would not be surprised if you get 10 of the top 12 guys correct on your final table. I may incorporate a few of these projections into my VBD model.

 

[Chase Stuart]

This is a GREAT article man. Nice job!

Although there are some clear cases where you wouldn't use the projected totals. Does anyone think Cobb is going to get only 864 yards this year?

I think this is most useful for the top-end projections. I would not be surprised if you get 10 of the top 12 guys correct on your final table. I may incorporate a few of these projections into my VBD model.

Thanks, meyerj31.

Even with Cobb, I think this is a useful exercise. Cobb has a lot of fantasy value in PPR leagues because he will get a zillion catches. And he's valuable because he's on a high-scoring offense and he adds in 100 or so rushing yards.

From a pure receiving yards standpoint, Cobb isn't that great. Most people project him with around 1,000 receiving yards. Considering this formula doesn't include age, projecting a general receiver to get 854 yards in Cobb's situation and a receiver who turns 23 in August to get 1,000 yards doesn't seem like a stretch. Cobb is a player on the rise, so you'd want to count for that.

 

[kutta]

Good stuff Chase, as always.

One issue I have is that it looks as though this process tries to push everyone to an average middle. For example, the top 21 teams in pass attempts last year are all projected to decrease, one is projected to remain unchanged, and the bottom 10 teams are projected to increase. I understand this is just a baseline attempt, but I'm just not sure how useful it really is. Does it really make sense to project the top 21 teams in pass attempts last year to all decrease their attempts? Last year we had nine teams with over 600 pass attempts. These projections show only one team with over 600.

Let's look at the top QB's over the past few years and see what would have happened had we used a similar formula to yours:

Brees:

2010: 658 (2011 projection: 586)

2011: 657 (2012 projection: 586)

2012: 670

We would have missed out on two big years if we predicted the decline you suggest:

Stafford:

2011: 663 (2012 projection: 588)

2012: 727

Again, If we would have used your formula, we would have decreased Stafford's 2011 stats a large amount and we would have been way off.

Tom Brady:

2010: 492 (2011 projection: 514)

2011: 611 (2012 projection: 565)

2012: 637

Another example where we would have been way off. From 2010 to 2011 we would only project a small increase, and from 2011 to 2012 we would have projected a large decrease, when in fact, his attempts increased.

These were the first three QB's I thought to test because they put up big numbers and they are usually near the top of the attempts list. I guess I just don't see the real value in projecting the way you suggest when, at least in many cases, the projections would be way off for three straight years.

Admittedly, this may work great for the mid-tier QB's, but I don't think we need to worry as much about projections for those guys because we usually draft more than one of them anyway. The top guys are the ones we have to make the hard decisions on.

 

[Chase Stuart]

Good stuff Chase, as always.

One issue I have is that it looks as though this process tries to push everyone to an average middle. For example, the top 21 teams in pass attempts last year are all projected to decrease, one is projected to remain unchanged, and the bottom 10 teams are projected to increase. I understand this is just a baseline attempt, but I'm just not sure how useful it really is. Does it really make sense to project the top 21 teams in pass attempts last year to all decrease their attempts? Last year we had nine teams with over 600 pass attempts. These projections show only one team with over 600.

Let's look at the top QB's over the past few years and see what would have happened had we used a similar formula to yours:

Brees:

2010: 658 (2011 projection: 586)

2011: 657 (2012 projection: 586)

2012: 670

We would have missed out on two big years if we predicted the decline you suggest:

Stafford:

2011: 663 (2012 projection: 588)

2012: 727

Again, If we would have used your formula, we would have decreased Stafford's 2011 stats a large amount and we would have been way off.

Tom Brady:

2010: 492 (2011 projection: 514)

2011: 611 (2012 projection: 565)

2012: 637

Another example where we would have been way off. From 2010 to 2011 we would only project a small increase, and from 2011 to 2012 we would have projected a large decrease, when in fact, his attempts increased.

These were the first three QB's I thought to test because they put up big numbers and they are usually near the top of the attempts list. I guess I just don't see the real value in projecting the way you suggest when, at least in many cases, the projections would be way off for three straight years.

Admittedly, this may work great for the mid-tier QB's, but I don't think we need to worry as much about projections for those guys because we usually draft more than one of them anyway. The top guys are the ones we have to make the hard decisions on.

Thanks, Kutta. In general, I think it does make sense to push everyone to the middle, as there is much more uncertainty about the future than the past.

In 2008, Brees had 635 passes, but in 2009 he had only 514 in 15 games.

In 2002, Gannon had 618 passes, but in 203, he had only 225 in 7 games

In 1995, Warren Moon had 606 passes; in 1996, he had 247 in 8 games.

In 2002, Drew Bledsoe had 610 passes; in 2003, he had 471 in 16 games.

I actually think Stafford, Brees, and Brady represent three very different examples. For Stafford, he is throwing a ton but it's because his team is bad. For Brees, it feels more sustainable. For Brady, the Patriots nearly set a record for plays last year -- NE finished 2nd in rushing attempts. Their pass/run ratio wasn't very pass-happy.

Regardless, I do think you have a point. It was too math-heavy for an already-math heavy article, but one thing I want to look at it using a logarithmic regression instead of a linear one in order to see if the curve fits better. It's on my to-do list.

 

[Chase Stuart]

There were 320 team seasons from 2002 to 2011. I sorted them from most to fewest pass attempts, and then broke them into groups of ten.

Year N Yr N+1 Diff621 584 -36586 544 -42564 528 -36548 553 5536 527 - 9521 529 8507 507 0483 520 37464 489 25429 493 64If you look at just the top 16 team-seasons, the average was 641 pass attempts in Year N and then 593 in Year N+1. If you remove the 2010/2011 Colts, the average becomes 638 and 598.

 

[Alex P Keaton]

Chase - great article as always. Really enjoy & appreciate your math-based perspective.

One quick question: why did you look at yards per target, rather than breaking it down into a) catch rate, b) yards per catch? Those seem like different views, where the component parts could vary for a number of reasons. Just curious.

Thx again for the article.

 

[kutta]

There were 320 team seasons from 2002 to 2011. I sorted them from most to fewest pass attempts, and then broke them into groups of ten.

Year N Yr N+1 Diff621 584 -36586 544 -42564 528 -36548 553 5536 527 - 9521 529 8507 507 0483 520 37464 489 25429 493 64If you look at just the top 16 team-seasons, the average was 641 pass attempts in Year N and then 593 in Year N+1. If you remove the 2010/2011 Colts, the average becomes 638 and 598.

That's interesting data and looks to fits pretty closely to original numbers. It would be very interesting to see if one or two data points are skewing any of the data. Do you know, for example, what the one or two biggest increases and decreases were (obviously the 2010/2011 Colts was one) in the groups?

Not knowing anything about the data other than what I know off the top of my head, it would seem to me that at the top, there would be a lot of teams that don't vary much, and then a few that drop a great deal because they had that one really weird year. I would think the bottom would be similar, but reversed. But I am also fully willing to admit I could be way off on that.

 

[Chase Stuart]

Chase - great article as always. Really enjoy & appreciate your math-based perspective.

One quick question: why did you look at yards per target, rather than breaking it down into a) catch rate, b) yards per catch? Those seem like different views, where the component parts could vary for a number of reasons. Just curious.

Thx again for the article.

Good question: the short answer is, I didn't think of it.

That would be interesting - I don't have much of a gut feel which way that would go. You also could break down yards per catch into air yards per catch and yards after the catch per catch.

 

[Alex P Keaton]

Chase - great article as always. Really enjoy & appreciate your math-based perspective.

One quick question: why did you look at yards per target, rather than breaking it down into a) catch rate, b) yards per catch? Those seem like different views, where the component parts could vary for a number of reasons. Just curious.

Thx again for the article.

Good question: the short answer is, I didn't think of it.

That would be interesting - I don't have much of a gut feel which way that would go. You also could break down yards per catch into air yards per catch and yards after the catch per catch.

Good point. I'm too lazy to do any of this, which is why I enjoyed the article so much. Did you backtest it at all to see if there are certain types of WRs for whom this starting point is more/less likely to be predictive?