Bad Coaching Thread: Matt Eberflus canned like chunk light tuna

I was thinking about this yesterday a little bit. Do the charts work no matter the composition of your team? Or are those charts the minimum amount of times a team should opt to go for 2?

Like let's say you're the Ravens. Lamar, King Henry, Isaiah Likely and Andrews (assume normal Andrews, not this year's watered-down version), Zay & Bateman. Should you go for 2 more often?

Or you're a team with an immobile QB, blah skill players, blocking TE. Do you go for 2 exactly the way the chart states, or even less?

League averages over x years. Of course people relying on historical averages forget that coaches were making flawed decisions before, which throws the numbers off (if we are talking optimal play).

If a coach is convinced of being the better team, it might make sense to reduce variance and simply kick the XP. Alternately, if the coach is aware of being the underdog then going for 2 PT conversions as often as practical would also make sense (as it is a zero sum game).

Does this explain many coaches spurning 2 PT conversions? To some degree, I believe.

The charts are based on the change in win probabilities of various point differentials. Obviously, the only way to analyze win probability is to use prior data, and sure, there are flawed decisions in the data. But as we continue to gather more data in the game, the data should become less and less flawed converging towards optimization.

Also, I think the decision point between going for 1 or 2 based simply on the current score is more solvable than say analyzing a 4th down decision at any given point in the game, except when the option is definitive loss when not going for it on 4th down.

There are a lot less variables to consider once you get to the point of deciding between 1 or 2. It's binary, and I tend to think the league averages of those decisions are pretty sharp (roughly 95% to convert a PAT and roughly 50% to convert 2).

Sure, there are other variables like having Derrick Henry and Lamar Jackson as opposed to Andy Dalton and Chuba Hubbard that can +/- in either direction. Maybe Derrick and Lamar convert 2 at 55% while Andy and Chuba convert 2 at 45%. Maybe your kicker converts PAT at 90% while another kicker converts it at 98%.

Here is an article than can analyze it and explain it way better than I can, but I certainly lean towards the go for 1 or 2 charts being pretty optimal, regardless of team and coaching personnel.

https://fivethirtyeight.com/features/when-to-go-for-2-for-real/

FiveThirtyEight

When To Go For 2, For Real

The SF decision was extra funny yesterday since the optimal decision is (very likely) to go for 2, they went for 1, and then the kicker doinked it. Carlson converted PATs at only 87.2% over 39 attempts last season.

So you think most coaches are too biased and rarely see their team as the underdog, thus not seeing a need to increase variance? I'm guessing that's probably the case, especially in the earlier parts of the game.

I think I lean this way too. I remember the Steelers teams of Ben, AB, and LeVeon coin-flipping on 2 pt conversions and it not working out very well.

I can't find WP on the Raiders FG, but there's no way that any model would say that wasn't an objectively poor decision.

Bowles just got Godwin killed when their only chance was the Ravens fumbling a kickoff or a kneel down (and 2 miracle TDs with no TOs).

Posted in the other thread, but there's nothing about that on

https://operations.nfl.com/the-rules/rules-changes/new-dynamic-kickoff-rule-explainer/

operations

New Dynamic Kickoff Rule Explainer | NFL Football Operations

It looks like that's just a random line from an article someone wrote and it's getting quoted now.

Yeah looks like that was wrong.

Something we probably underestimate is the meta dynamics of these decisions. If teams went for two more often it would start to lower the cumulative probability of converting. They are using their best plays and would have to use worse plays over time as they go for it more often. It also affects future high leverage spots that matters much more to their overall goal.

I don't think Pierce kicking the FG on 4th and goal from the 9 is that bad. Even if you score a TD and convert the two point conversion it gives the Rams a decent edge to be the team that wins the game in regulation.

Whereas kicking the FG sets up a spot where you just need to force a 3 and out and drive down the field and score a TD. McVay is typically overly conservative in these spots. Teams grossly overestimate their chances of stopping their opponent from scoring a TD with around 1:45 left and no timeouts.

3rd and Goal from the 9 yard line touchdown probability was around 15% based on the data from this link:

https://www.advancedfootballanalytics.co....

Let's say offenses have gained an edge since 2013 and the Raiders would optimize for TD on 4th down better than teams would on 3rd down. Let's call it a 20% TD probability.

Here are the assumptions I made:

Touchdown 4th and 9: 20%
two point conversion: 50%
FG from the 9: 95%
3 and out %: 40%
Touchdown with 2 minutes left: 15%
FG/TD with two minutes left (down 2): 35%
Raiders WP in OT: 46%
Rams Win in Regulation (Given TD/2PT): 40%
Raiders Win in Regulation (Given TD/2 PT): 10%
Probability of OT (Given TD/2 PT): 50%

Kick the FG:

FG Good 5.70%
Missed FG 0.08%

Total WP 5.78%

Go for TD:

TD/2 PT/Raiders Win in OT 2.30%
TD/2 PT/Raiders Win in Regulation 1.00%
TD/2 PT no good 1.40%
No Touchdown 1.10%

Total WP 5.80%

I would be interested to know McVay's 3 and out percentage when trying to run out the clock.

You are probably right that going for it on 4th and 9 is optimal but the difference isn't as much as you are implying. At most, it cost them 2% in win probability but you can justify kicking a FG with certain assumptions.

-----

What can't be justified with certain assumptions was Matt LaFleur kicking an extra point to go up two points in the second half.

The Texans made a FG to lead by one and there wasn't any other scoring besides the Packers last second game winning FG. That was a horrendous mistake that goes unnoticed because they won.

------

Touchdown 49ers
(10:32 - 3rd) B.Purdy up the middle for 1 yard, TOUCHDOWN.

I agree kicking the extra point was really stupid here for similar reasons as LaFleur. Obviously, slightly worse since they were actually losing.

-----

That Washington coaching tree (Shanahan, McVay, LaFleur) has created some great offensive minds with horrendous game management. They all are in the bottom 5 in terms of 4th down decisions/game management. LaFleur might be just bottom 10, but McVay and Shanahan are actively terrible at it.

4th down bot has the Raiders v Rams situation as FG being the better decision by a large margin:

https://rbsdm.com/stats/fourth_calculato...

It's not on the twitter because apparently twitter's rate limiting means not everything gets put on there but I saw Ben Baldwin (the guy who made the bot) linked to this, which has all the decisions from the entire week:
https://rbsdm.com/stats/fourth_weekly/

Using that system the SF decision to kick the XP was clearly bad:

https://rbsdm.com/stats/fourth_calculato...

Both of these are pretty much what I expected, although I might have guessed the FG was a slightly closer decision (but was definitely leaning toward FG being correct).

Interesting 4th down calculator.

I would think break even would be around 6 yard line. Instead 4th down bot had it at the 4.

1: 17.0% WP go for it vs 10.9% WP kick (57% chance of TD success)
2: 13.3% vs 10.8% (42%)
3: 11.4% vs 10.6% (34%)
4: 10.5% vs 10.6% (30%)

I interpolated WP numbers to 0.1%.

Nakua almost gets drug out of bounds on a sweep to the right when you absolutely cannot go out of bounds..

McVay: Let's run a swing pass to the left! Surely that will protect us from the right sideline.

..

Hmm, this is a good one. What's worse, bringing in a guy who hasn't pitched in a month to get the last two outs of a WS game, or walking the bases loaded to get to Freddie Freeman. So many layers.

Todd Bowles: GJGE

Calling the timeout right off the bat was his biggest mistake. And then maybe run it once and if you don't get it, you burn enough time that you can go for 2 w/o fear.

As is, Mahomes would have gotten them in FG range with 21 seconds. But Butker might have missed in the rain.

lol 50-50 against mahomes at home

Haha. I’m still on the side of going for 2, in that spot and most equivalent spots, really.

But the thread is interesting.

I like the one dude who like "we didn't ask for the math, nerd!" lol

lol the non math nerds are playing for OT every time.

By the way he decided so quickly, and how sure he was about it after the game, I think Bowles kicks the PAT even if there's no time on the clock. That would be a gigantic mistake.

The problem with that much time left is that you still lose some % of the time even if you make the 2PT conversion. So if you're 50/50 to make the conversion, then you're 50% to lose outright, and then another 5-10% or whatever to lose to a last second FG.