Biggest Upsets In Boxing History

How to Actually Track and Verify Boxing Upsets Without Getting Fooled by the Records

Most people who write about upset fights just look at who won and who lost and call it a day. That tells you nothing useful. The real work is in understanding what the fight meant before the bell rang, and that requires looking at odds, public perception, and historical context in a way most fans never bother with. Here's the thing nobody talks about. A "big upset" isn't just about an underdog winning. It's about the gap between expectation and reality being wider than anyone reasonably could have predicted. When I started cross-referencing old fight records, I kept seeing fights labeled as massive upsets where the underdog was actually only a +250 dog. The writers were just impressed because someone managed to pull it off, but the betting markets didn't see it that way at all. I once spent two weeks trying to verify whether the Leonard-Hearns "Brawl in the Hall" counted as a true upset. The public narrative says Hearns was the dominant champion and Leonard was just a tough Game 4 survivor who got lucky after a point deduction. But when I looked at the closing lines and the expert consensus going into that fight, Hearns was only around a -130 to -150 favorite, not the overwhelming choice people remembered. The point deduction skewed the entire historical memory of how expected Leonard's win would be. That's the first rule: memories are terrible data points. Always go back to the original odds and contemporary reporting.

The second rule involves what I call era adjustment. You can't fairly compare an upset from the 1980s to one from 2015 without accounting for how the betting market and media landscape changed. In the pre-internet era, public perception was shaped by maybe three newspapers and one TV network. A fighter could be massively overvalued or undervalued because there was less information available. The Ali-Frazier I fight in 1971 is remembered as a huge shock because the public didn't fully appreciate how damaged Ali's timing had become from years away. But the oddsmakers knew. They had him as a slight underdog or pick'em in many books. The general public just didn't have that data. So here's the actual process for building a reliable list: First, grab the closing betting line for any fight you're evaluating. This isn't easy for very old fights since many books didn't track or publish lines consistently before the 1990s. For fights before then, you'll need to rely on historical newspaper archives and boxing publications like Ring Magazine or The Boxing News. I use a combination of the BoxRec odds database, old Sporting News archives, and sometimes eBay listings of vintage Las Vegas strip tickets that show opening and closing lines. Vintage fight programs from the 1970s and 80s sometimes listed the public perception of the fighters' chances on the back cover.

Second, calculate the implied probability from the odds. A -400 favorite means the book thinks they have about a 80% chance to win. If that fighter loses, that's a massive upset by definition. A +200 underdog winning is an upset, but it's far more expected than an 80% favorite losing. Most casual lists treat these the same. They don't. Third, check what the public was actually thinking versus what the books knew. This is where it gets interesting. Sometimes the public heavily backed a fighter who wasn't getting the value from oddsmakers. In those cases, the "upset" feels bigger to regular fans than the numbers suggest. The Holyfield-Lewis I fight in 1999 is a good example. Lewis was a slight favorite or even money according to most books, but a huge portion of the public money came down on Lewis because of his size advantage and prior win over Holyfield. When Holyfield won by decision, it was an upset to many casual observers but not to anyone who had actually looked at the lines or Holyfield's ring IQ going in. Fourth, account for weight class and title stakes. An upset at heavyweight carries different weight than one at welterweight because of how deep the talent pools are and how predictable certain matchups tend to be. Heavyweight upsets happen more frequently simply because the margin between the best and the next tier is often wider than in lower weight classes where depth is more uniform. When I was building my first database, I tried weighting them equally and the list looked completely wrong. Once I started adjusting for weight class depth using ELO-style ratings compiled from fight records, the results made much more sense.

There are real problems with this approach. The biggest one is that pre-1990 odds data is notoriously incomplete. Many regional books operated without filing reports, and several major Las Vegas sportsbooks lost their historical records during closures and reorganizations. You will hit dead ends. I give up on about 15% of fights I initially target because I simply cannot verify the closing line or the public betting distribution. That's a hard limitation of the sport's history, not a flaw in the method. Another problem is the re-rating effect. After a big upset, both fighters' public reputations shift dramatically, and historians tend to retroactively adjust how they view the fight. Michael Moorer defeating George Foreman in 1994 is universally remembered as one of the greatest upsets ever, and it absolutely was. But when you go back and read the actual betting lines and expert picks from that week, Moorer wasn't nearly as big an underdog as the story has made him. The public had forgotten about Moorer's knockout power and his amateur background, so the narrative grew larger over time than the actual numbers justified. If you want a practical entry point, start with these three fights and work through the exercise yourself. First, Buster Douglas knocking out Mike Tyson in 1990. Douglas was a 42-to-1 underdog. The closing line was somewhere around +500 in most major books. Tyson had not lost in seven years and was considered almost untouchable. This is about as clean an upset as you'll find in the sport. The data is solid, the narrative is well-documented, and the numbers align perfectly with the story.

Second, Joe Louis losing to Billy Conn in 1941. Louis was a massive favorite, possibly as high as 20-to-1. Conn was a popular brawler who had never fought at light-heavyweight, and the crowd was firmly behind him. Louis won by knockout in the eighth round after being ahead on all scorecards. This one is messy because modern historians argue Louis was essentially easing through the fight and didn't want to hurt Conn in front of a hostile New York crowd. The upset narrative survives because Conn covered the spread easily for most of the fight, but the actual result was a win for the heavy favorite. This is the kind of edge case where the simple win-loss framework breaks down completely. Third, David Haye upsetting Wladimir Klitschko in 2011. Haye was a significant underdog on paper due to Klitschko's size and reach advantages, but the real story is how Klitschko's defensive style had been neutralized by Haye's speed and body work. The closing line was around +180 to +200 for Haye, which is an upset but not a astronomical one. What makes this fight notable in the upset conversation is that it came right after Klitschko had knocked out Haye's training partner and seemed completely built to counter his style. The betting markets adjusted slowly, and many books opened Haye at +250 before closing closer to +200. That gap between the opening and closing line tells you something about where the smart money landed. The takeaway is that tracking upset fights properly takes real work and you'll have to accept that some data is just unavailable. But the effort pays off because you stop repeating the recycled narratives and start seeing which upsets were actually surprises versus which ones just got rebranded over time. That distinction matters if you're trying to understand the sport honestly rather than just retelling stories everyone already knows.