Fourth-Down Fever: Has the Spreadsheet Actually Won?

# Fourth-Down Fever: Has the Spreadsheet Actually Won? **By Raffaele Cuccinotta | November 2, 2026** Here's the situation: it's fourth-and-two from the opponent's 38, middle of the second quarter, game tied. Ten years ago, the punt team is already jogging onto the field. Today, the offense stays out, the coach tugs his headset, and the stadium holds its breath. Going for it on fourth down used to be a gamble for desperados and gamblers. Now it's strategy, backed by win-probability models that say kicking away the ball is often the bigger risk. But here's the question I keep coming back to: has all this aggression actually made teams win more games? Or are we just watching a fashion spread—copycat coaching dressed up in decimal points? ## The Numbers Are Real Let's start with what nobody disputes. NFL teams are going for it on fourth down far more often than they used to, and the conversion rates have risen right along with the volume. In the 2025 regular season, teams went for it on **23.3 percent** of fourth downs—882 attempts out of 3,793 total, the highest rate since at least 1999, when the number was just 11.9 percent. The 2024 season wasn't far behind: 20.0 percent, with a conversion rate of **56.8 percent**, the highest in at least 34 years according to FOX Sports. PFF clocked it at 56.18 percent, also a multi-year high. The shift isn't happening on fourth-and-forever, either. The real change is in short yardage. On fourth-and-one, the go-for-it rate jumped from **41.6 percent** (1999–2017) to **69.2 percent** (2021–2025). Teams are converting those at **67.8 percent**. That's not recklessness; that's coaches finally accepting that a coin-flip play from a yard out is better than giving the ball back. ## The Model Says One Thing, the Coach Does Another The model says one thing, the coach does another, so let's look at why. Win-probability models estimate the chance of winning from any game state using historical data. On fourth down, they compare three options: go for it, kick a field goal, or punt. When the model says "go," it's usually because the expected value of keeping the drive alive outweighs the field-position gain of a punt or the three points of a field goal. But a model is an estimate built on assumptions, not a prediction of what will happen on the next play. It doesn't know your left tackle is nursing a sore knee, or that the other team's defensive coordinator just switched to a bear front. The model gives you the average; the coach has to guess whether this particular moment is average. ## Does It Win Games? The Evidence Is Messier Than the Trend This is where I have to slow down. The prevalence of going for it is a fact. Whether it wins championships is not. The strongest study I found is a matched nearest-neighbor analysis published in the *Journal of Sports Analytics*, covering 2004 through 2016. It estimated that a more aggressive fourth-down strategy would add about **0.4 wins per team per year**. Over 13 seasons, that ranged from 1.5 to 6.7 total wins depending on the team. But—and this is crucial—that's a **counterfactual estimate**. It's what the model thinks would have happened if coaches had listened, not what actually happened when they did. Then there's the study from Ruscio and colleagues at TCNJ, looking at 2001 through 2019. They found something that should make every analytics evangelist wince: teams that went for it **more often** tended to **win fewer games** (correlation of –.46). But teams whose decisions **agreed with the model** tended to win **more** (correlation of .80 with a random-forest model). The authors' conclusion is blunt: indiscriminate aggression loses. Selective aggression wins. In other words, going for it because you saw the Eagles do it is not the same as going for it because the numbers say so for your specific situation. ## Coaches Are Still Getting It Wrong Half the Time Another study, a bootstrap analysis of 2018–2022 decisions posted to arXiv, found that when the analytics were confident a team should kick, coaches made the right call 91 percent of the time. But when the analytics were confident they should go for it? Coaches got it right only **62 percent** of the time. And only about **34 percent** of all fourth-down decisions were high-confidence calls to begin with. The rest is noise, guesswork, and gut feel. PFF noticed something similar. In 2024, league-wide agreement with its fourth-down model dropped to **74.5 percent**, the lowest since 2012. The commentary was pointed: teams may be getting overambitious, going for it in spots where the numbers don't actually support it. ## Small Sample, But It's Worth Noting I could be wrong on this, but my read is that the analytics are directionally right and have been for a while. There were real wins being left on the turf by punting from the plus side of the field. The short-yardage revolution is probably sound strategy. But "directionally right" is not the same as "every gamble pays off." The TCNJ data suggests that simply going for it more often—without the model's guidance—correlates with losing. The PFF data suggests the league may be overshooting. And none of the win-impact studies I found used data from 2024 or 2025. We're watching the trend in real time, but we don't yet have the ledger to know if it's working. ## What We Know and What We Don't What we know: teams are going for it more, converting at higher rates, and the shift is concentrated in short yardage where the math is strongest. What we don't know: whether the teams that have most fully embraced this philosophy are winning more games because of it, or because they also have better quarterbacks, better offensive lines, and coaches who feel secure enough to survive a failed fourth down. The honest answer may be that the model is a tool, not a oracle. It tells you when the odds favor keeping the ball. It doesn't tell you whether your offense can block the play, or whether your defense can stop the other team if you fail. The best coaches, I suspect, are the ones who use the spreadsheet to ask better questions, not the ones who let it answer for them. I'll keep watching the ledger. For now, the spreadsheet hasn't won. But it has earned a seat at the table. --- *Sources: FOX Sports (Greg Auman); Vertize Sports (nflverse play-by-play); PFF (Bradley Locker); Journal of Sports Analytics; Ruscio et al., TCNJ; arXiv 2311.03490; Journal of Sport Management.*