Statistics are subject to interpretation. But the underlying numbers should be accurate.
There’s always room for more Muhammad Ali scholarship. Muhammad Ali: By the Numbers by Bob Canobbio and Lee Groves (published by New Book Authors) is the latest addition to the fold. The book offers a statistical analysis of Ali as a fighter and is one more resource to be used in studying Ali. But to measure the book’s strengths and weaknesses, it’s first necessary to understand CompuBox.
CompuBox was created in 1985 by Bob Cannobio and Logan Hobson. One year earlier, they’d been working at a sports data company when they saw a computer program that was capable of tracking every shot in a tennis match. That gave them the idea to apply a similar technology to boxing. They designed a computer program and, on November 28, 1984, brought it to Atlantic City to test it out.
“Trevor Berbick won a ten-round decision over Walter Santemore,” Hobson later recalled. “When it was over, we looked at our results. We didn’t know the full significance of the numbers, but we knew we had come up with something good.”
Since then, CompuBox has changed the way fans and the media view fights.
In calculating statistics for live fights, the CompuBox system utilizes two computers, two keypads, and two operators. Each operator records the efforts of one fighter. Punches are divided into “jabs” and “power punches” (any punch other than a jab). CompuBox totals jabs landed, jabs missed, power punches landed, and power punch missed. The results are codified as “punch stats,” a “punch profile,” or with some other label depending on which television network has commissioned the statistics.
CompuBox now compiles statistics for approximately one hundred boxing telecasts a year. Its clients include HBO, Showtime, ESPN, FOX, and several other networks. When Matchroom USA begins televising fights on DAZN this autumn, CompuBox is expected to be onboard. There are roughly eight thousand fights in the company’s data base.
The invention of CompuBox added an important new tool to the analysis of fights. But there has always been a disconnect with regard to CompuBox technology in that the punch counts don’t register the effectiveness of punches. There’s a difference between being hit and being hit cleanly by a solid punch. And the distinction between “jabs” and “power punches” is often irrelevant because not all so-called “power punches” are damaging blows.
In that regard, Canobbio has said repeatedly that CompuBox doesn’t score fights. Rather, it’s a tool to analyze the flow of a fight and contribute to a detailed historical record of the action.
There’s a tendency on the part of the media and fans to take CompuBox numbers at face value. Many writers—including this one—cite them in post-fight reports. The numbers are also often relied upon by the media and fans in criticizing judges’ decisions.
But there’s an issue as to how accurate CompuBox really is and whether the absence of a competitor has led to a decline in the accuracy of CompuBox punch counts. Canobbio has said that there’s a two percent margin of error, which would certainly be acceptable. But is that number correct?
There’s no magic CompuBox bullet. The system might have the aura of computer science. But in reality, it consists of two operators entering punches into a data base. The data is only as accurate as the split-second judgments of these operators as they enter what they see (or think they see) into their computers. And just as ring judges sometimes get a fight horribly wrong, so does CompuBox.
This writer first examined CompuBox in the aftermath of the March 13, 1999, draw between Lennox Lewis and Evander Holyfield. The widespread impression was that Lewis dominated the fight, and CompuBox credited him with outlanding Holyfield by a 348-to-130 margin. The almost three-to-one advantage in punches landed in favor of Lewis seemed high to me. So I decided to compile my own “punch stats.”
I watched a video of the fight twice. The first time, I counted the punches that Lewis landed. Then I repeated the process for Holyfield. When I was done, I was impressed with CompuBox. There were small variations in my calculations vis-à-vis theirs. But some errors are unavoidable in the flash of the moment. And it’s possible, if not likely, that the CompuBox totals were more accurate than mine. After all, the CompuBox operators were sitting at ringside, and I conducted my punch count off a VHS tape.
Thirteen years later, I had a very different experience. On June 9, 2012, Tim Bradley won a controversial split-decision over Manny Pacquiao. In the days that followed, much of the criticism of the judges focused on round seven, which was labeled “the smoking gun.” The CompuBox “punch-stats” had Pacquiao outlanding Bradley in round seven by a 27-to-11 margin. Yet all three judges scored the round for Bradley.
Thereafter, I studied videos of round seven in Pacquiao-Bradley I from multiple camera angles in their entirety . . . Several times . . . In slow motion . . .
Bradley outlanded Pacquiao 16-to-12 in round seven of their first fight. That’s a huge difference between reality and the CompuBox numbers.
You, the reader of this article, can watch round seven of Pacquiao-Bradley I on YouTube and make your own judgment.
Cannobio bristles at the mention of round seven from Pacquiao-Bradley I. “That’s history,” he told this writer earlier this month. “Pacquiao-Bradley is old news. I’ve been doing this for thirty-three years. We wouldn’t keep getting contracts and new deals if we weren’t good.”
But if a good historian discovers an error in the historical record that he or she has made, the historian corrects it. When statistics are as far off the mark as the CompuBox “punch stats” were for round seven of Pacquiao-Bradley I (and possibly the rest of the fight), they should be corrected. I’m not talking about a discrepancy of a few punches. I’m talking about significant mistakes.
Also, in the past, CompuBox operators sat at ringside for almost every fight. Now, to cut costs, CompuBox statistics are usually compiled off television by company personnel sitting at home. This practice began in 2014.
I’ve been in the first few rows at ringside for many fights. Trust me. From that vantage point, an observer sees which punches land and which don’t much more clearly than someone watching a fight on television. That’s why ring judges sit in elevated chairs by the ring apron rather than watch a fight on a TV monitor.
Calling a fight off a TV screen comes with a loss of visual perspective. It’s sometimes difficult to know whether a jab fell just short of the mark and whether what appeared to be a grazing punch slid harmlessly by. Also, compiling punch statistics off a TV monitor leaves the person counting the punches at the mercy of decisions made by a director in a TV truck.
Canobbio defends the practice of compiling statistics off-site. “The more we do remotely,” he says, “the less difference I see between working off a monitor and working from ringside.”
But there is a difference. And when CompuBox compiles statistics remotely, its keypad operators are seeing the same thing that viewers at home see. This means that, with a little practice, fans can create “punch stats” as reliably as CompuBox.
Once again, a Manny Pacquiao fight is instructive. CompuBox calculated statistics for the July 2, 2017, fight between Pacquiao and Jeff Horn off a TV monitor and reported that Pacquiao outlanded Horn by a 182-to-92 margin with a 10–9 edge in punches landed in round one.
I won’t quarrel with the notion that Pacquiao landed ten punches in round one. But I watched round one of Pacquiao-Horn several times the same way that the CompuBox operators watched it. Off a TV screen. And I think that Horn landed 18 punches, not 9. That’s a big difference.
One day after Pacquiao-Horn, Tom Gray of Ring Online wrote, “Horn was all over Pacquiao in the opening round. Horn landed to the mid-section and had more success up top. Pacquiao evaded some shots but he was tagged a lot more than Horn was. Not only are the numbers a joke, but this is where CompuBox’s legitimacy in this contest ends.”
Once again; you, the reader, can go to YouTube and decide for yourself.
This brings us to Muhammad Ali: By the Numbers. The book contains punch totals for 47 of Ali’s 61 professional fights. Complete film footage of the other fourteen bouts is unavailable. The format for the book is to present a narrative that places each fight in context followed by an account of the action, the judges’ scorecards, and complete CompuBox statistics.
The punch totals were compiled by Lee Groves, who tracked 474 rounds from Ali’s fights. Groves began working for CompuBox in 2004 on a part-time basis and has been with the company on a fulltime basis since 2007.
“I watched each Ali fight twice,” Groves recounts. “Once for each fighter. From the start, I had a sense of the historic importance of what I was doing. I knew it would add a new dimension to how Ali is perceived as a fighter.”
Groves also put an enormous amount of time into crafting the narrative text.
CompuBox divides Ali’s career into four phases:
(1) “The Young Clay”—The six available bouts from Cassius Clay vs. Alonzo Johnson through Clay’s first fight against Henry Cooper.
(2) “Prime Ali”—The ten fights from Clay-Liston I through Ali vs. Zora Folley.
(3) “The Comeback Years”—There’s some questionable organization here. Ali’s “comeback” began with his 1970 fight against Jerry Quarry and culminated in his regaining the heavyweight throne against George Foreman four years later. But in Muhammad Ali: By the Numbers, the “comeback years” are extended to include twenty-one Ali bouts from Quarry I through Ali-Frazier III in Manila. There’s no material on Ali’s decision victory over Rudi Lubbers because footage of the complete fight is unavailable.
(4) Past-Prime Ali—The final ten fights of Ali’s ring career.
The conclusion reached by the authors of Muhammad Ali: By the Numbers is that, even in his prime (Clay-Liston I through Ali-Folley), Ali wasn’t as good as we thought he was. During this period, Canobbio and Groves posit, Ali had a defense that was “surprisingly leaky.” As proof, they cite “the high percentage of punches his opponents landed.”
“During those rare times they were in range to reach him,” the authors write, “they reached him—often. The statistics shatter the myth that the prime Ali was an expert defender. In light of this information, some experts may have to reconsider not only how they perceive the best version of Ali but also how he might fare against the best of the best.”
Groves recently elaborated on this theme, saying, “Watching Ali in the past, my eye had always been drawn to his magnetism. Here, I had to watch the opponents. And I was shocked at how often the opponents penetrated Ali’s defense. Ali fancied himself an amateur magician. And in the ring, he pulled off his greatest illusion. He was able to erase the memory of his getting hit by not showing it in his facial expression and immediately striking back. Knowing how often Ali got hit, even when he was young, was surprising to me.”
Thus, Muhammad Ali: By the Numbers states, “As great as he was to watch inside the ring and to listen to beyond the ropes, Ali was hardly the greatest in terms of statistics.”
According to CompuBox, over the course of 47 fights, Ali outlanded his opponents by a 7,953-to-7,010 margin, which was attributable to a huge 3,749-to-1,892 advantage in jabs landed. But he was on the short end of power punches landed by a 5,118-to-4,204 tally.
Moreover, Muhammad Ali: By the Numbers proclaims, “Ali rated a minus 1.7 on CompuBox’s plus-minus scale that compares the fighter’s overall connect percentage with that of his opponents. Ali’s terrible final nine fights had a lot to do with that. But even if Ali had retired after the Thrilla in Manila, his plus-minus rating would have been a puny plus 0.4 (34.9% for Ali, 34.5% for his opponents) and he still would have been underwater in power-punch accuracy (39.3% for Ali, 41.7% for his opponents).”
Further with regard to this “plus-minus” scale; CompuBox analyzed available video footage for what it calls “twelve notable heavyweight champs” beginning with Joe Louis and ending with Wladimir Klitschko. By CompuBox’s calculation, Ali ranks dead last. Joe Louis ranks sixth, while Joe Frazier ranks first.
According to CompuBox, over the course of 47 fights, Ali’s opponents landed 34.2% of their total punches and 41.8 percent of their power punches, while Joe Frazier’s opponents landed 29.7% of their total punches and 38.5% of their power punches.
In other words, according to CompuBox, Ali was easier to hit than Joe Frazier.
I don’t think so.
CompuBox also maintains that Ali was easier to hit than Rocky Marciano. Here, it’s worth noting that, on August 29, 1969, two days before being killed in a plane crash, Marciano said of Ali, “From what I’ve seen, he’s as good or better than anybody I fought. I really don’t know if I could have beaten him. I would have liked to have fought him. Even if I’d lost, I’d have known that I fought the best.”
There are a lot of small errors in Muhammad Ali: By the Numbers. For example, the authors say that Cassius Clay’s February 21, 1961, fight against Donnie Fleeman (his fifth as a pro) was “his first main event.” That’s just wrong. Ali was in the main event in his pro debut against Tunney Hunsaker. It’s a small point. But mistakes like this, by definition, raise the issue of whether the underlying CompuBox statistics themselves are reliable. More on that later.
Also, statistics in and of themselves can be misleading.
According to Muhammad Ali: By the Numbers, the “Prime Ali” was outlanded by Sonny Liston in Clay-Liston I and was even with Liston in punches landed in their second encounter. But as the book acknowledges, Liston’s 103-to-95 edge in Clay-Liston I was built largely on a 37–1 disparity in round five, when Clay was blinded and trying simply to survive. And regardless of the punch count, Liston quit on his stool after six rounds. That wasn’t an illusion.
Ali-Liston II, of course, ended in a first-round knockout for Ali.
Then there’s Ali versus Zora Folley. Muhammad Ali: By the Numbers maintains that Folley outlanded Ali by a 66-to-61 margin. But Ali was in total control of the bout that night and knocked Folley out in the seventh round.
After the fight, Folley called Ali “the greatest fighter of all time” and declared, “This guy has a style all his own. It’s far ahead of any fighter’s today. How could Dempsey, Tunney, or any of them keep up? Louis wouldn’t have a chance; he was too slow. Marciano couldn’t get to him and would never get away from Ali’s jab. There’s just no way to train yourself for what he does. The moves, the speed, the punches, and the way he changes style every time you think you got him figured. The right hands Ali hit me with just had no business landing, but they did. They came from nowhere. Many times he was in the wrong position but he hit me anyway. I’ve never seen anyone who could do that. The knockdown punch was so fast that I never saw it. He has lots of snap, and when the punches land they dizzy your head; they fuzz up your mind. He’s smart, the trickiest fighter I’ve seen. He could write the book on boxing, and anyone that fights him should be made to read it.”
That sounds like the description of an all-time great, not a fighter who came out on the short end of CompuBox numbers. And for the record; having viewed a video of Ali-Folley, I think Folley landed far fewer than 66 punches that night.
Similarly, Muhammad Ali: By the Numbers, notes that “on the negative side, Ali failed to land a single punch in a given round seven times.” But that’s like grouping apples with brussels sprouts. There’s no rational correlation between round ten of Ali vs. Larry Holmes (when a helpless Ali was being beaten to a pulp) and round one of the first fight between Ali and Floyd Patterson.
In round one of Ali-Patterson I, Ali was playing with Patterson. He feinted a lot but threw no serious punches. Blow-by-blow commentator Don Dunphy observed, “Cassius Clay seems content to just move. Clay is clowning at this point.”
Muhammad Ali: By the Numbers acknowledges that Ali “ceded round one to Patterson” and also concedes that all but one of his twelve punches were “half-speed jabs.” But even that is an exaggeration. At most, Ali appears to have thrown one punch that was intended to land against Patterson in round one.
Moreover, in terms of Ali’s overall ring career, some of his jabs were sharp, cutting, and hurting. Others were of the stay-away-from-me variety designed to dictate distance and the pace of the fight. Whether or not this latter group of jabs landed was secondary to their primary purpose.
The ultimate question, of course, is, “How accurate are CompuBox’s Ali statistics?”
Ali’s third-round knockout of Cleveland Williams on November 14, 1966, is often cited as Ali’s finest performance as a pro. CompuBox says that Williams landed ten punches that night. But is that right?
Decades ago, Bill Cayton (who owned film rights to most of Ali’s bouts) and Steve Lott (who worked for Cayton) reviewed films of several Ali fights and determined that Williams landed three punches against Ali. A recent review of Ali-Williams on YouTube seems to indicate that three is the more accurate number.
Ali’s first-round knockout of Sonny Liston on May 25, 1965, has long been shrouded in controversy. CompuBox says that Liston landed four punches that night. Cayton and Lott said it was two. A review of the fight on YouTube suggests that Cayton and Lott were right.
CompuBox says that Leon Spinks landed 419 punches in his first fight against Ali. Cayton and Lott said it was 482. That’s a significant difference. I don’t know who’s right because I haven’t counted punches for that fight.
“I’ve been doing this for a number of years,” Groves recently told this writer. “I’m confident in my ability to track punches.”
To repeat: you, the reader of this article, can go to YouTube and decide for yourself.
CompuBox has made a significant contribution to boxing. But it has the tools to be better.