Comments on Advanced NFL Stats: The Value of a Touchback

All of this makes me wonder is it ever really wort...

2009-09-21T19:36:34.915-04:00

All of this makes me wonder is it ever really worth taking a knee as the returner? Unless the guy's backpedaling as he receives, or is more than maybe six yards deep, it seems to me it'd be worth taking it out. Worst case scenario (I would think) is you end up on about the 15. No idea what the numbers show on this or where to start, I'm brand new to the site, but I gotta say so far this is great.

What would the WPA have been last Monday if Buffal...

2009-09-19T08:50:54.014-04:00

What would the WPA have been last Monday if Buffalo's Leodis McKelvin had simply accepted the touchback by taking a knee in the end zone? What would have been the WPA had McKelvin returned the kick for 7 seconds (so as to deprive New England of the 2 minute warning, but not fumbled the ball? If Lindsay's touchback should have been the "touchback heard 'round the world," McKelvin's return is the "should have been a touchback hear 'round the worlds." It would be very interesting to see some numbers on what might have happened if McKelvin had made a different choice or if his choice had not had such a bad consequence. Thank you for considering this request. Love your work.

Thx. Fixed.

2009-09-09T23:28:25.319-04:00

Thx. Fixed.

Just an FYI guys...the link in the paragraph descr...

2009-09-09T22:36:51.695-04:00

Just an FYI guys...the link in the paragraph describing the Arizona-Seattle game takes you to the Denver-SD match instead. (The link in that paragraph is correct.) Great post, as usual.

Great post. Keep up the superb research. its nic...

2009-09-09T18:00:51.012-04:00

Great post. Keep up the superb research. its nice to see a sharp mind with a disciplined approach figuring out what the value of this stuff is. My perception during games is that announcers don't give a hoot about a touch back or a return to the thirty, its all the same to them. And they certainly don't celebrate it like a sack!

Just to be clear, this is only a very small part o...

2009-09-09T17:36:27.613-04:00

Just to be clear, this is only a very small part of the answer. A kicker is never going to kick all 4 or 5 kicks in a game for touchbacks. But, this analysis at least gives us an idea of how valuable a deep kicker can be.

For non-touchbacks, a kicker benefits by nearly exactly 1 net yard for every 2 yards of additional depth. This is because the returner will have more time to run up-field before encountering the coverage.

At the 40-20 yd line region, every yard of fld position is worth 0.06 expected points. So for every additional yard of kicking depth, you'd expect half that--an extra 0.03 EP.

The standard deviation of kicking depth is just over 2 yds. So the difference between a typical above average kicker (at +1 SD) and a below average kicker (-1 SD) is about 5 yds. This is equivalent to 0.15 WP per (non-touchback) kick. With 5 kicks per game, you'd net 0.75 EP.

Ben, you also have to consider the possibility tha...

2009-09-09T15:36:08.849-04:00

Ben, you also have to consider the possibility that the punt goes too wide and only goes a comparatively short distance before going out of bounds - effectively a decent punt return.

Anyway, back to the purpose of this post, that's great to know! A few years ago the Cowboys were debating needing a KO specialist because Grammatica couldn't get punts beyond the 10 yard line, making for even easier returns. Good to know there is actually some reasoning behind them getting a KO specialist this year in the draft!

Brian, Does your play-by-play database tell you i...

2009-09-09T14:07:27.216-04:00

Brian,

Does your play-by-play database tell you if a punt went out of bounds?

I'd think that there would be a noticable advantage to punting out of bounds or at least towards the sideline. I know that punt returns aren't as long as kick-off returns but a kick out of bounds would remove any possibility of a long return.

As I see it there are several problems with the coffin-corner kick. One being that because you are kicking on an angle it has to travel farther to get the same punt length (think hypotenuse of a right triangle with the short side being 26.67 yards). So if the punter kicked it 46 yards to the sideline it would only travel 37 yards downfield. Is that worth more than the average punt return?

On upside of it is that it would be harder for the punt returner to catch and make a good run out of even if it stayed in bounds. They would have to sprint to the sideline and then change direction to turn upfield. Also they would have less field to work with because the sideline takes away most of the angles. The kicking team (which would know which side the punter was going to kick to in advance) should have a sizable advantage on kick coverage.

Just to carry on with should a team take a 3rd QB ...

2009-09-09T06:36:27.213-04:00

Just to carry on with should a team take a 3rd QB or a KO specialist, it depends on how likely it is that a team needs a 3rd QB to play.

As above, a KO specialist is worth 4 pts. The average NFL offense scores 21 pts, but if we assume that a team forced to play without a specialist QB (for instance, through injury) scores no points, then the BEST CASE value of a third QB is 21pts x p(first two QBs being injured) - the 3rd QB is worth nothing if he doesn't get on the field. This is a best case as it assumes the 3rd QB adds 21pts over a non-QB once he gets on the field, both unlikely as he would probably not get on early enough nor is he as good as the previous two QBs. Thus, a 3rd QB on the roster should be picked above a KO specialist if the probability of the first two QBs getting injured in the game is 19% or greater. In this case, the QB adds more points on average than the KO specialist.

0.8 EP per TB, 5 kickoffs per game. So, a guy who ...

2009-09-09T06:20:01.218-04:00

0.8 EP per TB, 5 kickoffs per game. So, a guy who can boom every kickoff for a touchback is worth 4 pts over a guy who can't make touchback, assuming similar FG kicking ability.

Assuming that a team can measure a FG kickers true probability curve in practice, it is theoretically possible to provide a pts value to each kicker.

Jim-That could be, but keep in mind what I wrote a...

2009-09-08T23:32:06.618-04:00

Jim-That could be, but keep in mind what I wrote about it being "an extra first down required." If you can imagine a game where a team has to drive the field to win the game--down by 1, or 6, or whatever, with a couple minutes left.

First downs are converted on average 67% of the time. If a touchback makes the receiving team get 6 consecutive first downs instead of 5, or 5 instead of 4, that means the touchback reduces the receiving team's chances of winning by a third.

Admittedly, this is a rough way of thinking of it, but depending on the particular situation this would imply the receiving team would otherwise have a 39% chance of winning. I think that's very plausible.

LI stands for leverage index. It's still under...

2009-09-08T23:19:52.111-04:00

LI stands for leverage index. It's still under development as I haven't settled on a final methodology for it.

Why is it so hard to find help on how to read the ...

2009-09-08T23:10:14.901-04:00

Why is it so hard to find help on how to read the win probability graphs? You'd think something like that would be easy to find. What does "LI" stand for in the box that pops up when you hover over the line?

You'd think the most valuable touchbacks in te...

2009-09-08T21:53:31.541-04:00

You'd think the most valuable touchbacks in terms of win probability would be when leverage is highest, i.e. late in a close game. Wouldn't overtime touchbacks have the highest WP value? There were at least four of them last season alone. And I'd bet you could also find a few later in the 4th quarter of a close game than four minutes remaining.

Given that overtime coin toss winners win the game about 60% of the time and that the 20-yard line is still slightly better than break-even for the receiving team (according to expected points), I don't think it makes sense for a touchback ever to have more than a 10% swing in WP.

A 13% swing for 12 yards with that much time remaining seems way too high. I suspect something isn't quite working right with your late game model.

Will, it would be hard to make any definite conclu...

2009-09-08T17:17:24.987-04:00

Will, it would be hard to make any definite conclusion based on this analysis. Because every team has different needs, it would be impossible to answer 'yes' or 'no' to your question. For a team with a lot of depth, maybe it would be beneficial. On the other hand, it may be more beneficial to keep that rookie RB with hopes of them blossoming in future years, or that veteran LB who will provide leadership to your young LBs.

But in reality, I can't name one team with good depth at every position. So the decision becomes (as most decisions) a balancing act for the head coach. Game theory, anyone?

I think his point is that there isn't enough v...

2009-09-08T16:24:13.574-04:00

I think his point is that there isn't enough value to keep two Kickers. The point he was driving was that a deep kick-off (read touchback) is can be worth more then a kicker with long range. Up about 35 yards most kickers a very similar. So for those kicks that are over 35 yards are you better off with a kicker that can regularly kick those in or a kicker that can kick a touch back.

In the end I am not sure you even have to have the numbers on hand. Its a game of inches and if you can have a 7-12 yard advantage on kick-off in field position then most kicks currently kicked would be closer anyways (assuming same plays are called).

I have data coming out my ears. It's time I do...

2009-09-08T13:29:00.881-04:00

I have data coming out my ears. It's time I don't have! Baby steps...

So what's the conclusion? Is the FG specialis...

2009-09-08T13:14:13.548-04:00

So what's the conclusion? Is the FG specialist worth the roster spot over the 9th o-lineman, the 6th receiver, the 7th DB, or the 3rd QB? Of course this is the part of the paper where the verb changes from "shows" to "indicates," but I bet you have enough data to make a case.