Why is a 2-goal lead the worst lead in hockey?
The psychology of attack and defense.
Yes, it was a good team we were playing, but we should not have been down by three. I looked up at the clock. Four minutes left. It was going to be an uphill battle; even my brothers, who are never tired, looked tired.
Suddenly the thought occurred to me: what do you have to lose? The game was already lost, in a sense. No harm in taking a few risks. So, even though I was playing defense, I moved forward on the next faceoff.
For the next four minutes, I played some of the best hockey of my life, scoring two goals and assisting on a third, which my brother tapped in with about 30 seconds remaining. Somehow, we’d sent this championship game into overtime.
During the break, we agreed that we should stay aggressive. “Keep pressing,” I told everyone. But on the opening faceoff in overtime, I hesitated. For the final four minutes of regulation, I had been playing forward as a defenseman, leaving nobody behind me. But now I worried. It was sudden death—one mistake and the game was over. So I put myself in no-man’s-land, somewhere between the aggressive and safe option.
Sure enough, the puck squirted out to my side. I went forward—a tad cautiously, if I’m being honest. The opposing player beat me to the puck, chipped it past me, and scored on the breakaway. After clawing back from a 3-goal deficit, we’d lost in overtime in less than 5 seconds.
The Score State Trap
There’s an old hockey adage that the worst lead in the game is a two-goal lead.
The idea is simple: teams that go up 2–0, 3–1, or 4–2 tend to surrender their lead more than would be expected in a world untouched by psychology. Something about a cushion of two (or more) goals lulls a team into a false sense of security—after which they get outplayed, outshot, and, not uncommonly, outscored.
To the skeptical reader, this might smell like availability bias: we remember the blown leads because they’re dramatic. But as it turns out, one of the most predictive variables in hockey is something called the score state—whether a team is leading, tied, or trailing. This single factor tells you more about a team’s expected shot share in the next period than almost anything else, including roster strength, coaching decisions, or home-ice advantage. Teams that are trailing take more shots; teams that are leading take fewer. Score effects have been documented across thousands of National Hockey League (NHL) games and apply to many score states, including but not limited to two-goal leads.
As Micah Blake McCurdy puts it:
One constant of NHL hockey is that teams that are losing, on average, dominate play. They have the puck more, they take more shots, they score more goals, even though they usually still lose.
Score effects make some intuitive sense. A losing team has every reason to take risks, whereas a winning team wants to minimize danger, and may be willing to sacrifice their own offense for better defense. That being said, if Team A is better than Team B and has taken a lead as a result, why change anything?
As a multi-sport athlete growing up, I can attest to a wide gap between theory and practice when it comes to protecting a lead. My coaches would routinely say “Pretend you don’t know the score” or “Don’t take your foot off the gas.” NHL coaches echo the same sentiment in intermission interviews: “We need to stay aggressive” or “We can’t give them life.” Even fans will pound the table and insist: “The best defense is a good offense.”
Yet again and again, we see winning teams “sit back,” play defensively, and pray that the clock runs out before their lead does.
Why?
Ecologically Speaking
You don’t have to be a professional athlete to recognize the conservatism that creeps in when you’re ahead—whether in poker, a romantic triangle, or Settlers of Catan. What’s striking is that even world-class athletes, backed by billion-dollar organizations, fall into the same trap. To be fair, some situations call for caution. Maybe the golfer with a comfortable lead lays up instead of going for the green over water. But in many cases, it’s the conservatism itself—playing scared—that squanders the lead.1
And yet, even when everyone knows not to sit back, they sit. Whoever could solve this problem—systematically and repeatably—would be worth more than any free agent on the market.
The problem begins with shallow descriptions of the problem. Broad terms like “complacency” or “momentum” are invoked to explain comebacks, but as Rob has pointed out, those aren’t explanations; they’re just words. To really understand what’s going on, we have to investigate our evolved psychology, which was shaped for the savannah and survival (and certainly not hockey rinks or modern sports).
The key system at play here is the brain’s evolved risk calculator. If the purpose of risk is reward, it makes sense that as a reward is secured, our appetite for risk declines. A team leading 2–0 has less to gain and more to lose by pushing for a third goal, which is why offense feels less urgent than defense. The same logic applies off the ice. Suppose you believe you're worth $120k but earn $100k, so you agitate for a raise and get an extra $10k. Now at $110k, the urgency to go after the remaining $10k diminishes because you have less to gain and more to lose.
For the trailing team, the opposite is true. Down 2–0, the risk of falling to 3–0 barely registers compared to the reward of making it 2–1.2 This was my realization in the opening story—that risk had become cheap.3
Given that this effect is bidirectional—that is, both teams participate in it—why does so much commentary focus on the team that’s ahead? Why are they seen as primarily responsible for the outcome? Probably because we assume they have more control. After all, they’ve already demonstrated control by taking the lead, which their opponent wanted too. This assumption can cause us to underestimate the trailing team’s role in the reversal—an oversight Paul Maurice, the excellent coach of the Stanley Cup–winning Florida Panthers, managed to avoid. Speaking about the opposing Edmonton Oilers, he noted: “No team is more dangerous when their risk profile changes.”
Let’s return to score effects. Once it became clear that sitting back with a lead was counterproductive—or, conversely, that playing with the desperation of a trailing team was productive—why didn’t everyone adapt? Why do teams still let the score dictate their behavior? Economists tend to label this kind of thing “bias” or “irrationality.” But as Gerd Gigerenzer would say, our decision-making is ecologically rational—well suited to the environments in which our brains evolved. It’s just that what worked in the Pleistocene doesn’t always carry over to the modern world, whether in a hockey rink, econ lab, or food court. Not even when millions of dollars are on the line.
In the opening story, for example, I knew where I should have positioned myself on the overtime faceoff. I just couldn’t do it. It felt like some distant puppeteer had taken over, pulling strings my conscious brain couldn’t override. That distant puppeteer was evolution by natural selection.
In a sense, every animal exhibits ecological rationality, because all animals employ decision-making heuristics that are adaptive in the environments they evolved in. The interplay of risk and reward is central to many of these heuristics. For example, the !Kung of Southern Africa leverage the fact that lions have it, too. They’ll only chase a lion off a kill if it’s already partly full—because a well-fed lion has little to gain and plenty to lose by fighting back. Even lions understand cost-benefit.4
What’s Artificial about Sports?
When humans consistently behave suboptimally—even in ways that run counter to their stated or actual goals—a common culprit is artificiality: an environment, task, or combination of the two that lies outside the bounds of human evolutionary experience.
Sports might be artificial in at least three ways.
First, there’s an arbitrary timer. Across the animal kingdom—and throughout human evolution—conflicts didn’t end when a horn sounded; they ended when someone had clearly won. Victory meant establishing dominance or securing resources, not simply having the upper hand when time ran out. (Sports are a bit like musical chairs in this sense.) It’s possible that this alternate definition of victory creates a kind of confusion in the leading team, especially the more they’re winning. If the game feels over, why keep pushing? This might be part of a deeper explanation for what we casually call complacency.
Second, sports violence is fake in that you’re not actually killing enemies. (What a waste of time, right?)
Finally, the defender’s advantage that is observed across the animal kingdom may pertain to space much more than score. Let me explain.
Territoriality: The Defender’s Advantage Applies to Space, not Score
By this point, readers familiar with animal literature might be wondering: What about all the studies showing a defender’s advantage? Shouldn’t that translate into teams protecting a lead better than chance would suggest?
The defender’s advantage is a well-documented and evolutionarily significant pattern: in territorial conflicts, the animal that already occupies a territory or resource is more likely to win, even if smaller or weaker than the intruder. This has been observed across multiple species, including speckled wood butterflies, red-winged blackbirds, and lizards.
Why does this happen? Many explanations center around an asymmetry of knowledge. The defender is more familiar with their abode, including escape routes, defending angles, and any fortifications they may have erected. They also know the value of their territory, including how much they have invested in it, which helps them determine how hard they should fight for it. The attacker knows less about all of these things. It seems like a quality spot, but is that really enough information to risk life and limb?
So shouldn’t defenders—in this case, teams defending a lead—get an additional boost?
Not really. The key is that the defender’s advantage applies to space, but not necessarily score. It’s more of a “we must protect this house” than “we must protect this lead.”
For example, the defender’s advantage might explain the consistent finding, from hockey to tennis to soccer, that home-field matters.5 But it may offer an even better explanation for why teams give up leads. Once the winning team has a nice nest-egg, you see, their instinct is to sit back and defend it. It makes no sense to our evolved logic to go out and acquire more—to spend time in the attacking zone, i.e., someone else’s territory—when we already have a good situation in ours. It would be ecologically irrational to leave home unguarded. However, this often backfires. In hockey, for instance, this brooding instinct gives the trailing team more time on the attack, which is exactly what the trailing team wants. You can’t defend a lead without defending, but that also means the other team is attacking.
Funny enough, there is even a term for this in hockey: a “stay-at-home defenseman” is one who does just that. With a two-goal lead, such homebodying can pervade an entire team, turning the nest into a cage.
I mentioned earlier the tendency to underestimate the trailing team’s tactical adjustments. As it turns out, that assumption might be justified. Some data suggests that blown leads really are the fault of the leading team—specifically, their tendency to hang back. McCurdy’s research shows that expected goal production by leading teams declines around 10 to 15 percent, while, on average, the trailing team continues to push at a pace similar to when the score was tied. He writes:
I have spent a great deal of time trying to understand precisely what causes [score effects]. This article summarizes my most-recent attempt, and my conclusion is: Leading teams sitting back drive score effects on shot rates more than trailing teams pushing.
So let’s get ultra precise for a moment. The natural instinct to protect a valued territory or resource leads to “sitting back” behavior for obvious ecological reasons. The whole point of aggression is to secure a valued thing and then protect it, not secure it and then leave it unguarded to chase another valued thing. However, in the artificial realm of sports, this instinct often backfires, because one isn’t protecting a territory as much as a score.
In other words, leading teams sit back and defend the home because that is the ecologically logical thing to do once one has acquired something worth protecting, but in many sports scenarios, that is the best way to forfeit an advantage.
The Strength of a Deficit Mindset
Meanwhile, the solution to the problem of maintaining a lead is as clear in theory as it is difficult in practice: always play as if the game is tied—or even better, as if your team is losing.
Another way to understand the power of a deficit mindset—playing as if you’re behind even when you aren’t—is to look at the psychology of many of the world’s most successful (and often most miserable) people.
Plenty of those who acquire the most in life—whether status, money, or championships—operate as if they’re always down two goals. For example, Michael Jordan pretty much always felt slighted. More broadly, it’s practically a meme at this point that great players and teams invoke as their motivation someone—anyone—who didn’t believe in them for a split second. Even the Florida Panthers, fresh off last year’s Stanley Cup and returning with an arguably stronger roster, leaned into this mindset. After winning the Cup again, forward Brad Marchand said, “Everybody wrote us off.” I literally laughed at my TV.
The deficit mindset can be remarkably adaptive in certain circumstances because it offers a workaround to the ecologically-rational tethers of risk and reward. In ancestral environments marked by scarcity and danger, Michael Jordan would’ve been taking an unnecessary risk chasing the next win. In the modern world, he’s known as the greatest basketball player of all time.
Optics as a Final Overlay
In the artificial context of sports, a perfectly viable way to protect a lead is to extend it—to keep pressing the advantage that earned the lead in the first place. Yet even when coaches say the right things—“stay aggressive,” “don’t sit back”—it can be hard for them to really mean it, and even harder for players to act on it. Part of the reason is that our evolved risk systems are old, strong, and stubborn. But the other part is social: we know we’re being watched and judged.
So even if a player or team manages to override their instincts, that’s not enough. They also need social buy-in.
Fans don’t forgive teams that blow two-goal leads. Owners don’t forgive coaches who can’t “seal the deal.” The media feasts on collapse. Even if a coach is convinced that the right move is to stay aggressive, then, the safer bet might still be to play it safe: protect the lead, avoid blame. An owner might forgive a coach for doing the expected and losing—that’s life—but will they forgive one who gambled, stayed bold, and lost anyway?
This social pressure helps explain why 4th down attempts in football were initially less successful than the analysts predicted. When teams first began adopting aggressive 4th down strategies, the actual success rates were often lower than forecast by models. This wasn’t because the models were wrong in principle, but because their recommendations went against conventional wisdom, causing coaches to hesitate, players to tighten up, and fans to pounce on failure. As it so happens, coaches and players want to win a bit less than they want to keep their jobs, and the leash is shorter for those who defy convention.
Some other fun examples of the social judgment tax are shooting free-throws overhand and kicking penalty shots to the side of the goalie. Research has shown that more free-throws would be made if released underhand and more goals would be scored if kicked along the ground at the center of the net. But, of course, no professional athlete wants to be seen shooting like a granny or kicking like a five-year-old.
Speaking of professionals, let’s not forget the referees—often endearingly, as well as insultingly, called “zebras” in hockey.
Although referees are trained to be neutral, their base model is still that of an evolved ape. As Chris Boehm argues in Hierarchy in the Forest—and as Rob explores in his upcoming article on “boosting”—humans possess a deep-seated leveling instinct. Referees have it too, and it can be trained out of them about as effectively as the territorial instinct can be trained out of the athletes they adjudicate. This explains the finding that, at least in hockey, referees call more penalties on the winning team and fewer penalties on the trailing team. Referees unconsciously nudge the game back to even. This is such a given that when my friends and I bet on hockey, we factor in a power-play or two for the trailing team, depending on time left and score deficit.
This leveling instinct, though, might not even be the strongest social force acting on referees. It may be their aversion to hearing boos—and often worse—from the home crowd. That is the conclusion of Moskowitz and Wertheim, authors of Scorecasting, who argue that the home team’s biggest edge isn’t reduced travel or turf familiarity, but a loud, boisterous, usually drunk posse that intimidates the refs into making favorable calls, especially in key moments of the game. The instinct not to piss off a mob is, understandably, deeply wired.
Can it be fixed?
Even when one is aware of these things, it is hard to change them. The instinct to protect what you have—and do nothing that might risk losing it—is powerful and persistent, even after being shown that this very behavior undermines your protection. Likewise, even after viewing the data, and instructed not to boost the losing team, referees still find it hard to call games impartially.
In his paper on going for it on 4th down in football, economist David Romer concludes:
…the behavior of National Football League teams on fourth downs departs systematically from the behavior that would maximize their chances of winning…This is true even though the decisions are comparatively simple, the possibilities for learning and imitation are unusually large, the compensation for the coaches who make the decisions is extremely high, and the market for their services is intensively competitive. Despite these forces, the coaches who fail to make maximizing choices are not fired and replaced by ones who do. How can this be?
In my view, teams do not make maximizing choices for the same reason that individual people can’t stop checking their phones or eating more than they should. Humans are ecologically rational—not strictly, formally, or economically so. Our ecological rationality often makes us look stupid in the artificial environments we’ve created, against the artificial tasks to which we put ourselves, and—it must be said—in comparison to the artificial self-portraits we’ve erected. As Romer puts it: “Much economic analysis is built on the idea that the assumption that agents maximize simple objective functions leads to reasonably accurate descriptions of their behavior. This paper demonstrates that in a case where this hypothesis can be tested directly, it fails.”
It fails because humans are ancient organisms with risk systems calibrated to avoid death, injury, and social punishment. These instincts can override vast amounts of training, coaching, and incentives, and rise to the surface most forcefully under stress—like in the final minutes of a big game.
However, this framing finally offers a few ways forward. One is the power of socializing best practices. When an entire organization aligns around an aggressive mindset—as Dan Campbell and the Detroit Lions have in recent years—it can shift conventional wisdom to match actual data. This reduces social backlash and gives coaches and players permission to go out and win, rather than merely avoid losing. The Philadelphia Eagles embraced this too, going for it on 4th down early and often with their (in)famous “tush push.” Their approach was simple and consistent: keep doing what works, even when you're ahead.
The second insight is that humans can learn to override their ecological rationale—but only through intensive training and experience. Experience, in particular, matters. My sense is that part of the reason the Panthers successfully defended their Game 6 lead—which won them the Stanley Cup—was because they had blown a three-game lead to the same Edmonton Oilers just last season, not to mention coughing up a three-goal lead earlier in this year’s series. Heading into Game 6, they knew firsthand what sitting back could cost them.
This lends credence to the often-derided practice of general managers seeking out players with playoff experience. Drills, training, and culture can reinforce an aggressive mindset, but these pale in comparison to actual playoff experience—just as actual playoff experience pales in comparison to millions of years of evolutionary fine-tuning.
The bottom line is that deep instincts don’t yield easily. I should know. Returning to the opening story a final time, I learned a valuable lesson that day—but then again, did I? I think it’d be hard, even today, to act differently.
Endgame
The two-goal curse is a perfect case study in how deep-seated, evolutionary behavior can override conscious intent—even among highly paid professionals in high-stakes environments. It is another example of evolutionary mismatch, in which evolved instincts clash with the artificial environment of modern sports.
If teams want to override protective, counterproductive instincts, they’ll need more than analytics. They’ll need to leverage—or else circumvent—a few million years of evolutionary wiring.
Until then, the two-goal lead will remain the worst lead in hockey.
Coda: Why Hockey Especially?
While the phenomenon of blowing a lead happens in many sports, it may be especially acute in hockey for a few reasons.
First, the fluidity of play. In hockey, unlike in football or baseball, there are few set plays and lots of improvisation. Players make dozens of real-time decisions per one-minute shift. Because there are fewer interventions for a more formal or economic rationality to insert itself—not to mention fewer scenarios that mimic previous scenarios—hockey players rely more on instincts and reactions than coaching decisions or training. In football, for example, coaches can think for a moment about how aggressive they want to be, then call a play that their team has run dozens of times in practice. This rarely happens in hockey.
Another important difference is that possession is much less distinct than in other sports. In football, basketball, baseball, and so on, it is clear who is on offense and who is on defense. In hockey, it’s much less certain. Possession changes more often and much faster, largely because it’s much harder to control a puck—with a stick—while on skates—than it is to stand there and hold a ball. This makes play even more fluid, and reduces the role of coaching and play-calling further. On faceoffs, for example, coaches must stipulate: “If we win the draw,” “if we lose the draw.” In football, the coach knows their team will start the play with or without the ball.
Third, hockey has the right scoring distribution. Basketball has too many points for a two-point lead to matter. The phrase “a 20 point lead is the worst in basketball” just doesn’t have the same ring. Soccer has too few. In order for a team to blow a two-goal lead, at least four goals total need to be scored, which is well above the average for a soccer game (2.85 for English premier league) but well below the average for an NHL game (~6). In hockey, scoring is rare enough to feel precious, but not so rare that comebacks are impossible.
All this said, the phrase could be tweaked to fit other sports. A two-score lead in football—say, 14–0—often leads to conservative play-calling and predictable offense. In golf, leaders notoriously tighten up while ahead. “A two-stroke lead is the worst in golf” wouldn’t be a bad shout.
As the football saying goes, “The only thing the prevent defense prevents is winning.”
Or as McCurdy puts it: “the marginal benefit to the leading team of a goal, in standings points, is so much smaller than the marginal value of a goal to the trailing team.”
This is also behind the recommendation from the analytics community that hockey teams pull their goalies (netting them an extra skater) earlier and oftener. They’ve got nothing to lose. Patrick Roy is probably most responsible for implementing this strategy at the NHL level.
See The World Until Yesterday by Jared Diamond, p. 336.
Then again, the defender’s advantage may have little to do with it. In Scorecasting, Tobias Moskowitz and L. Jon Wertheim argue that the biggest factor in home-field advantage is referee bias.