The Comeback Bet.

In the 2025–26 NBA Finals, Jalen Brunson and the underdog Knicks faced off against Victor Wembanyama and the young Spurs. While the Knicks pulled off the upset, it wasn't that they beat the Spurs that was surprising, it was how they beat them. Throughout all five games, the Spurs built a double-digit lead at some point. In fact, they led for 73% of all minutes played during the Finals. Yet despite controlling most of the series, the Spurs lost in just five games. As the pressure mounted, as it often does for young teams, the lights became too bright, and the Spurs blew lead after lead. The defining moment came in Game 3, when the Knicks erased a 29-point deficit, completing one of the most memorable comebacks in NBA history.

From a betting perspective, if you backed the Knicks before the series, you earned 1.6× your original stake at +160 odds. If you bet them before each game, you likely doubled your money each time at around +200 odds (except for Game 3, which they lost). That's already a solid return. But if you waited until the Knicks fell behind and bet them at their highest live odds in each game, you would have made roughly 300× your original stake.

Because sportsbooks cap live odds and build in a vig, the real return would have been closer to 150–200× your original stake. Even so, that one stretch of comeback bets could fund well over a hundred future wagers of the same size before you'd give back all of those profits.

That's the inspiration behind the comeback bet.

The math behind the comeback bet is based on the relationship between probability, expected value, and variance. When a live bet is priced at +400, the sportsbook is implying roughly a 20% chance of that team winning the game. In theory, if we observed hundreds or thousands of independent bets at those odds with no edge, we'd expect the realized win rate to converge toward 20%, about one out of every five teams coming back to win. This is a consequence of the law of large numbers, where larger samples tend to reflect the true underlying probability more closely.

The picture changes when we look at much smaller samples. If we only place 10 comeback bets instead of 100, the expected win rate is still 20%, but the realized win rate can vary much more from that expectation. The variance of the sample proportion grows as the sample shrinks, so short-term results become far more influenced by randomness.

At +400 odds, hitting 2 out of 10 bets is approximately break-even before sportsbook margin. Hit 3 out of 10 and you generate a profit; 4 or more produces increasingly larger returns. Of course, the opposite is also true, hitting only 0 or 1 out of 10 results in a loss.

The goal of the strategy is not to rely on variance alone, but to combine that natural variance with an informational edge. By identifying indicators that suggest a comeback is more likely than the sportsbook's implied probability, we're hunting for positive expected value on high-priced bets. If our analysis finds situations where the true probability is, say, 30% instead of the implied 20%, then over time we should win more often than the odds suggest. In that case, consistently hitting 3 out of 10 instead of 2 out of 10 isn't luck, it's evidence that we've found alpha by identifying situations the market has undervalued.

If I had an algorithm that could predict exactly when to bet on a comeback, I'd be retired on a beach in Malibu. The truth is, there's no way to know when a team is going to come back. Sometimes a team falls behind early and never gets back into the game. Other times, they claw all the way back only to fall short in the final minutes. There's no secret formula, and betting on a comeback is always a gamble. Still, it happens more often than people expect.

When studying for the Series 7 license to become a broker, the textbook I used mentioned a strategy that essentially translates to "fade the public." The idea is simple: when most people expect one outcome, there can be value in taking the opposite side. Most people aren't sports experts, and it's easy to assume a team won't come back once they've built a sizable lead.

There are ways to sharpen your edge through technical analysis or by noticing momentum shifts during a live game, but sometimes it really is as simple as betting against the crowd. That's often where the best odds are found.

There's no magic number for ideal odds, it's about finding value spots and staying disciplined with your approach.

Comeback bets are difficult to hit consistently, but there are situational factors that can signal when one might be more likely. The correlations are weak, treat these as context clues, not guarantees, when evaluating long-shot odds. Here's what I look for (and what the badges on the live board mean):

Quantifying momentum is one of the hardest problems in sports analytics. We can usually recognize when a team has it, but defining and measuring it is much less straightforward.

The way I approximate it is through win expectancy. When win expectancy increases, it can be read as a team gaining momentum in a practical sense. A move from 25% to 30% is relatively minor. But a jump from 20% to 40% is a meaningful shift in game state, essentially a doubling of their chances in a short window.

In betting terms, that kind of swing can correspond to a move from roughly +400 to +150. It reflects not just a score change, but a broader reassessment of control in the game.

That said, momentum is not permanent. It can fade quickly, which is why betting on a comeback too early is risky, teams frequently stall after short bursts of success. But late in the game, momentum becomes more consequential. A 10-point lead with five minutes left turning into a 2-point game with a minute to go fundamentally changes the pressure. The leading team tightens up, possessions become forced, and mistakes become more likely.

At that stage, you often want to be on the trailing side. Not only has the price moved from +400 to +150, but the game environment itself is no longer fully captured by the original number. Crowd energy, home court, and psychological pressure all become more pronounced, in some cases, that +150 can feel more like a −150 in live context.

Because of that, I often prefer waiting until a team shows real signs of life before considering a position. It can feel like you're missing the best number, but comebacks are inherently rare. Many teams go down early and never generate meaningful momentum. In those cases, patience is often more valuable than early entry.

At this point in my betting approach, I mostly focus on comeback positions. Over time they've felt like some of the most interesting and, at least in my experience, most consistently usable opportunities. They're also simply more engaging than traditional pregame bets. Here's why:

As much as people, including myself, enjoy them, parlays are generally a poor long-term strategy. The sportsbook edge is significantly higher than on straight bets or live positions, and that's where books generate a large portion of their profit. We keep using them because they're entertaining. There's nothing wrong with that, but understand what you're getting into.

Once you start focusing on comeback betting, you have to think more seriously about unit sizing and bankroll management. The appeal of a comeback bet is straightforward: you get plus-money odds similar to a parlay, but you only need one outcome, the win. Instead of multiple independent events going your way, everything reduces to a single condition. It doesn't matter how messy the game looks or how unlikely it seems at points. All that matters is whether the team finishes with the win.

A $10 parlay might return $200 at +2000, but it requires multiple legs to all hit. A single $40 comeback bet at +500 depends on one outcome. Similar payouts, fundamentally different risk profiles. From experience, I've rarely seen consistent success with high-multiplier parlays, while I've had more repeated success in shorter-range comeback scenarios.

This doesn't mean parlays should be avoided entirely, they're fine in small, controlled amounts as entertainment. The key is treating them differently from your main strategy. A small parlay is low-impact; scaling parlays up to the same unit size as live comeback bets accelerates losses quickly due to their higher inherent variance.

Since hearing that, I usually place my bet and put my phone down. Cash out is more of a psychological tool than a purely mathematical one.

When you place a comeback bet and the game moves further against you, it's easy to get frustrated. A +400 position can quickly turn into +1500, the odds feel worse, and your conviction fades. You lose confidence in the original read and feel tempted to salvage something. I've been there plenty of times, and reacting in those moments usually does more harm than good. Just because you didn't get the best possible price doesn't mean the strategy is wrong, in many cases the decision to enter is still correct even if the timing doesn't feel ideal in hindsight.

On the other side, when your team starts closing the gap, books will present a cash-out that feels increasingly attractive. It's designed to reduce variance and lock in profit, but it comes at a cost relative to true game state. There have been plenty of situations where I'd have made more by simply letting the position play out. It's hard to pass up guaranteed money, but at the end of the day, we're here to gamble, so stick with it.

If you're going to live by the comeback bet, you need to die by the comeback bet.

Now that you understand the comeback bet and how I approach it, there are a few key considerations I want to leave you with.