NOT A BASEMENT — When the Chicago Bears decided to punt on fourth-and-1 while trailing the Green Bay Packers 38-27 on Sunday Night Football last week, Noah Johnson just shook his head. He knew where the game was going next.
The Packers marched the length of the field for a touchdown that put the final nail in Chicago’s coffin. Game over. Season over for Chicago. Any chance of a comeback went down the drain with the decision to punt rather than “risk” going for it on fourth down.
“The Bears hadn’t stopped Aaron Rodgers all day, they couldn’t stop a nose bleed and they decide to punt the ball on fourth-and-inches?” Johnson says incredulously over Zoom.
Johnson is one of the world’s best Madden players. He has seen that situation tens of thousands of times. He knows the calculus. Giving the ball back not only allows for Rodgers to tack on another touchdown but it lets him drain clock. The Packers’ TD drive took nearly nine minutes.
“In the fourth quarter you see teams down by multiple scores punting and it’s like, what is the point?” Johnson said. “Every time you punt, you have to look at it like this: That’s at least two minutes off the clock. Every time you give up a first down [it’s two more minutes] and if you’re the Bears, more than likely you’re going to give up a first down and that’s another two minutes off the clock. You’re just limiting your chances to win.”
Johnson, whose gaming name is NoahUpNxt, ranks No. 1 on the EA Ultimate Madden Bowl standings. He won the Madden 2020 challenge at 17, becoming the youngest ever to take home the championship belt and $35,000 in prize money. Oh, and he did it on national TV on ESPN2. Last year he became West Virginia University’s first Esports player and in November the fresh-faced economics major won the Madden Championship Series. The win banked him $25,000, another belt and a ring by beating Weber State.
By the way, what were you doing in college?
He’s putting in 10-12 hours per day in the lead up to tournaments. Noah knows how to grind like an NFL coach. Between prep, tournaments and playing for fun, he’s managed more in-game football situations than any NFL coach by 100 fold. And because he’s always playing against other excellent players, his in-game decisions are almost exclusively in high-pressure moments.
“Almost every game comes down to clock management when you’re playing against another high-level player,” Johnson says.
Against Weber State, Johnson faced a difficult fourth down decision with the game on the line. He was looking at fourth-and-10 with the clock running down. A punt would give his opponent one more shot, a conversion would win the game. Johnson decided to pass the ball and converted to secure the 23-21 win. His opponent didn’t get an opportunity at potential game-winning drive.
Coaches often say that second-guessers don’t have anything on the line, so it’s easy to demand more risk taking from the press box or couch. Well, you can’t say that for Noah. He has money and rings and belts on the line with every fourth down.
“The worst thing on Twitter where it’s like, ‘Oh it’s a stressful situation,’” Johnson said. “These guys are getting paid millions of dollars and I would hope that they could deal with the stress. That’s what I’m paying you to do. You’re my head coach, I want you to be able to deal with these stressful situations.”
For him these things aren’t stressful because he already knows how he’s going to handle the decisions from so much previous experience.
“It’s another day in the office,” he says.
Noah’s moves on Madden don’t come from breaking down analytics with a secret formula, or PFF subscription or Football Outsiders manual. They come from a trial-and-error process. Think of it this way: If you watched 20 movies in a year, how good would you be at guessing plot twists and endings versus watching 5,000 movies in a year? Or imagine if NFL coaches got 100 preseason games to try out anything they wanted over and over before bringing it to one regular season game.
“I wouldn’t say that it’s so much that I’ve studied [game-management]…. it’s really just experience,” Johnson said. “That comes from getting the reps in and trying it out in games that aren’t super important. OK I’m going to go for it here and let’s see what happens and then bringing that to the more important games.”
There’s a general idea from football people that Madden players just go for every fourth down all the time and it works because people like Noah can break the game and convert every time. It turns out that’s not true. As in real football, Johnson explained that the Madden environment changes all the time and players must adapt their decisions with it.
“It actually varies on Maddens and it varies on the patches throughout Madden,” Johnson said. “Right now Madden is super offensive heavy… so it’s very easy to score the ball and very hard to stop somebody. So unless it’s like fourth-and-20, you honestly have to go for it, which is crazy. Last year, for example, you would punt the ball like in the NFL on fourth-and-5, which is unheard of in any Madden… But I rarely punt now because the possessions are too valuable.”
That sounds familiar with superhuman quarterbacks playing a game of last-man-with-the-ball-wins, doesn’t it?
Noah also doesn’t think that everything on the video game is like real life. He’s aware that his Madden offense wouldn’t necessarily translate to the league. He believes in the human element of football and understand that communicating and managing an entire team isn’t the same as pushing buttons. He’s not entirely sure that going for two points when down nine is great in all situations. He doesn’t think he knows X’s and O’s or technique better than Matt Nagy. But in the situations that are the same from Madden to the NFL — i.e. down, distance, score, timeouts — he does see things that NFL coaches don’t see. Noah is a master of processing the probabilities of football and the art of clock management the same way jazz musicians learn music theory through endless jam sessions.
So the whole thing about gamers being know-it-all nerds is, like most tropes, inaccurate. But he does see some spots in games where NFL coaches might be able to get an edge in comparison to their normal lines of decision making. Two such situations in particular are going for fourth downs inside a team’s own territory and calling timeouts to avoid getting a delay of game penalty.
Let’s say you’re up seven points at your own 30-yard line with five minutes to go. In Noah’s opinion, coaches should go for a fourth-and-8 there because A) if they succeed, they have a good chance of ending the game without giving the ball back B) if they fail and the opponent scores, they will score quickly and still allow time for a game-winning drive. Plus, it isn’t a guarantee that the opponent will score a touchdown in the red zone. Even the best teams only produce red zone touchdowns 70% of the time.
The other is a simple calculation: Five yards are not as valuable as a timeout, so stop using timeouts to avoid taking delay of game penalties.
“You have to look at a timeout like 40 seconds,” Johnson said. “I’ll see a coach call a timeout on third-and-15 instead of making it third-and-20. I don’t know what the stats are for third-and-15 but they are not very high. Most coaches are going to call something soft and punt the ball anyway. They’re going to call a draw or a screen. What’s the difference between calling that on third-and-15 or third-and-20?”
He used the example of coaches calling for the offense to attempt to draw the defense offside and then burning a timeout only to kick a field goal from 27 yards rather than 32 yards or to punt from the 50 rather than their own 45-yard line.
“I think people are finally realizing that timeouts are important because you will hear an announcer like Tony Romo saying, ‘What are they doing?’” Johnson said.
Is it realistic to think lifelong coaches should be borrowing theories from a Madden whiz-kid?
Well, let’s back up for a second.
Something interesting about the Madden games and their connection to reality is the story behind them. After pitching the game to John Madden in the 1980s, the creators of the series, Trip Hawkins and game producer Joe Ybarra, rode in a cross-country train with the legendary Raiders coach and broadcaster to learn everything they could about football. Madden wouldn’t do the game unless it best reflected real football. An ESPN podcast 30 for 30 told the story that every year he’d meet with the company that produced the game, EA Tiburon, and add things that he’d seen while broadcasting like trendy play calls, schemes and formations etc. At one point, the Madden insisted that the game made it harder to convert fourth downs because the youths were playing the game so unrealistically by going for it all the time.
But now the gamers and many NFL coaches are on the same page about being more aggressive. Put it this way: Some coaches are using GPS systems while others are still asking for directions.
On Thursday Night Football, Chargers coach Brandon Staley received criticism for his team missing on several fourth downs that might have made a difference in their overtime loss to the Kansas City Chiefs. After the game, Staley said:
“The real football people understand that what I’m doing is playing to the strengths of our football team. What I’m doing is I’m trying to make the decisions that I think are going to win us the game. And I’m ready to live with all that smoke that comes with it. And I’ve been very transparent about that. What makes football and competition so great is that there aren’t going to be perfect decisions. But you need to be able to live with the decisions, and your team needs to know why you’re making these decisions, so that they can live with them, too.”
It seemed like intentional phrasing by the Chargers’ head coach to use the words “real football people,” since naysayers of the gamers’ approach have long clung to being the “real football people.” It’s not so easy to do that anymore.
Two weeks ago there was another controversial in-game decision when Ravens coach John Harbaugh elected to go for two at the end of Baltimore’s matchup with the Steelers. They didn’t convert but Noah loved the move.
“I saw John Harbaugh said in his interview said, ‘We knew we couldn’t stop them and all of our DBs were injured and we’re in an away game so let’s make the decision and end the game right here,’” Johnson said. “That’s the same thing in Madden, especially this year’s version of Madden. A lot of people will go for it in the late-game scenario because it’s so hard to stop and it’s like, would I rather bet on a coin flip or would I rather bet on myself.”
Minnesota Vikings head coach Mike Zimmer said that he’s seen a noticeable uptick recently in the aggressiveness of opposing coaches.
“I do think people are going for it more on fourth down,” Zimmer said. “In my opinion it’s really easy for analytics to say, ‘You should do this,’ ‘you do that,’ when they don’t really know who’s playing for the other team or what your strength of your team is or so on and so forth. But for us it’s really a belief that that is the right time, this is the right situation. At the end of the day, you have to have belief in your gut that you’re either going to make it or I think if you don’t make it you can give your team a chance to lose the game.”
“I think it’s really important with the situation, who you’re playing, the score of the game, the strength of your team. There’s so many different variables that numbers don’t tell you. But I do think that a lot more people are a lot more aggressive now. We talk about it every week, about fourth down situations and whether we will go for it and try to do a good of defending those things as well.”
So we hear the word “analytics” constantly when it comes to these types of decisions but we rarely get any explanation of how the probabilities work and what they actually say. What can they factor? What does it mean when the “analytics” say you should go for it? And do the analytics actually match up with Noah’s Madden philosophies?
Well, let’s ask some analytics experts.
It turns out that “analytics” are not a giant supercomputer in a cave at the top of Mount Poindexter. There are lots of different models created by people with different ideas about how to evaluate football decisions.
One of those people is co-founder of EdjSports Frank Frigo, who was a different type of gamer before creating a company that’s become a leading voice of the analysis of coaching decisions since 2013. Frigo was a Backgammon world champion. He also worked in commodity markets structuring wholesale energy transactions. It would take a whole other article to explain how that works but you get the drift that his background is in areas that require an incredible mind for percentages.
Frigo’s first football simulation model was called “Zeus.” He built it in 2001, right around the time Billy Beane was winning with the Moneyball Oakland A’s. EdjSports was launched in 2013 and acquired the football analysis company Football Outsiders in 2018.
Here’s how the model works:
At the beginning of each game, the two teams have certain percentage chances to win based on how good they are (EdjSports uses Football Outsiders' data to determine team strengths and weaknesses). So let’s say that Great Team vs. Terrible Team starts at 60-40. Then with each play, the model simulates the rest of the game and estimates the percentage of times that one team wins.
If Terrible Team gets ahead 14-0, the simulations might come out that they still only win 55 out of 100 times. If Great Team gets ahead by 14, the simulation might say they win 75 out of 100.
So let’s say the Terrible Team has a fourth-and-1 at the 50-yard line. The model will calculate the percentage of simulations that estimate they win if they convert, if they fail and if they punt. The difference between those numbers is how decisions are evaluated.
The criticism that models don’t know who’s playing isn’t exactly correct but it isn’t exactly wrong either. Do they know that Terrible Team’s fullback has been blowing up Great Team’s linebacker all day? Do they know that the good team’s nose guard is a mismatch for the terrible team’s center? Do they see the body language of the opposing defensive line? Do they know the opposing team’s fourth-and-1 play calls that they’ve studied on tape all week based on formation, motion, personnel? Of course not.
“The model is, how do you go throughout a game and push yourself toward 100%?” Frigo says. “Every individual decision can be measured in that context and if you don’t obsess about scoring on a possession or the final score, you get some amazing insights about how to improve your win probability. There are often situations in a game that are very high leverage that the model will say, ‘Do this because it improves your win probability.’ It doesn’t care about the possibility of losing by an embarrassing score.”
Another common sentiment about these models is that they “always say go for it.” Again, not exactly true but they are indeed saying that teams should go for fourth downs more often, even with the recent uptick of fourth down attempts in the NFL. Frigo’s explanation: The models sees things we do not see.
Put it this way: There are chess simulators that can find the perfect move in every possible situation. The simulators often suggest moves that look ridiculous to humans because they defy every general rule that’s taught to players. A human simply wouldn’t be able to see something that seems crazy on the surface.
Frigo used the example of a game between Cleveland and the Los Angeles Chargers in which the model was insistent that Cleveland go for a fourth-and-6 at their own 18-yard line.
“What the model is picking up on is that you convert those more than you think you might and secondly, when you fail there’s a lot of clock considerations,” Frigo said. “Sometimes in these late-game situations when you fail, your opponent will probably score quickly and you’ll get another possession and when you punt you’re already giving them the ball at midfield. Now if they move it 20 yards and get into field goal range to beat you, they’ve burned too much clock and you don’t get that residual possession.”
Reminds you of what Noah Johnson said about going for it on your own side of the field, right? But it’s very hard for an NFL coach to take that “risk” knowing that it will look really, really bad if they fail to convert the fourth down.
“People use that as a negative on Twitter, like, ‘These people pretend they’re playing Madden.” Well, that’s more of a compliment than what you think it is,” EdjSports senior data analyst Ian O’Connor said. “You learn how it’s actually a good thing.”
Indeed our Madden star is nailing his decisions by the model without ever having looked at a single chart. Johnson’s point about field position being overrated also bears out.
“There’s definitely been a shift in analytical thinking and we’re seeing coaches do these things much more often but there’s a whole other layer to this that’s largely untapped,” Frigo said. “Some coaches get it but the first time I see a team with a lead with a fourth-and-6 in their own territory go for it, I’ll believe the analytics revolution is alive.”
“The one that a lot of coaches have yet to come around to is a fourth-and-5 at the opponent’s 45 or 40, a punt to flip field position even down to the 1-yard line, your win probability is typically less than what it’s going to be if you go for it,” O’Connor said. “Field position isn’t worth as much as broadcasters and teams make it out to be.”
You’re probably wondering: With Mike Zimmer decrying the hardline analytics approach, which EdjSports says has been largely taken by John Harbaugh and Brandon Staley, how Zimmer’s decisions compare to the model, the Madden player and the cutting-edge coaches.
He’s presently ranked 19th at improving his team’s chances to win with in-game management. But it’s not quite as simple as calling him below average.
One of the fascinating things to study surrounding coaching decisions is the psychology behind trends. Zimmer has been one of the most aggressive coaches in the NFL in the second half of the season and has greatly improved his accuracy in decision making according EdjSports’ model.
Starting the season 3-5 and losing Danielle Hunter may have inspired the Vikings’ head coach to throw caution to the wind knowing that his team desperately needed to win games and probably couldn’t do so by getting defensive stops.
“He’s 19th and 24th in fourth-down decision making, not great but in our first [coach ranking] he was 29th, so he’s improved or stayed the same in every week except for one week as far as his ranking,” O’Connor said. “Eight of his top 10 best decisions have come since Week 8…seven of his worst 10 came before Week 8. Since the middle of the season he’s gotten a lot better. I don’t know if that’s because Kirk Cousins has been good and he’s trusting him more.”
Indeed that was right around the time when Zimmer pushed Cousins to be more aggressive. It was the week where Zimmer went for it on fourth down to end the Chargers game without giving the ball back to Justin Herbert.
O’Connor says that Zimmer’s ranking was tanked by one bad game, Week 2 in Arizona. One particular decision sunk the Vikings’ chances to win by 28%. The ironic thing about that miscue is that it actually turned out OK. The Vikings punted on fourth down from mid-field with less than three minutes remaining. But they got a stop and delivered a would-be game-winning drive on the next possession.
If not for that game, Zimmer would be in the top six for least win probability lost. That example brings us two points: 1) Going by the model won’t always work and going against the model won’t always fail. And sometimes the model can have very small differences that will count against a coach but ultimately it’s a coin flip (Frigo says most debatable two-point conversions are that way) 2) One game matters a lot. The difference between the best and worst coaches versus the model might only be 100% in win probability over a season but the difference between 10-7 and 9-8 could mean making or missing the playoffs. Maybe if the Vikings converted that fourth down in Arizona, they’d be a lock for the postseason now.
It can feel like the analytics drain all of the intrigue out of football debates. But sports are ever evolving. There will always be brilliant moves in chess that the engine doesn’t like or mid-range jump shooters in basketball who beat the models or punts on fourth down that were the right call even if they go against the numbers. And, hey, if every coach made the same choices with the numbers all the time, it would pretty boring (see baseball’s home run/strikeout contests).
If you don’t see the fact that a college student in West Virginia playing a video game is making NFL decisions with a higher rate of accuracy than the pros as being fascinating, I’m not sure what to tell you.
Speaking of which, Noah Johnson said he would love to work in the NFL someday but he loves that Esports open the door for everyone to play. He enjoys comparing Madden to the NFL and college, even down to which schemes succeed in the different games. He doesn’t know whether Madden will end up as a long-term career or just something he did as a kid.
For now he’s going to keep going for it.
“I’m just kind of just riding the wave,” Johnson said.
To read more of Matthew Coller's work covering the Minnesota Vikings, subscribe to the Purple Insider newsletter at PurpleInsider.substack.com