Hellraiser Boxing News | Boxing By Numbers: The Use Of Stats In Boxing
The Weekend Read. Steve Fearon investigates stats in boxing and whether they're a good thing or can be misleading.
Sorry, but our website is best viewed on a device with a screen width greater than 320px. We'd still love to hear from you, feel free to contact us: Hellraiser Boxing on Facebook.
by Steve Fearon
Let’s talk about Data and Boxing, and how time and time again, the analysis gets things wrong.
Data in Boxing
Having done a bit of work in non-league football, looking at performance analysis, and which measures provide the best indication of positive performance, when I first started looking at transferring my skills to a boxing environment, my instinct was to find a source of quantitative data similar to that provided by football sites like WhoScored, that share detailed figures on pass completion, shot accuracy etc.
However, it became clear very quickly that the only source of within-fight data was Compu-box, and I quickly identified a few issues with their model of working.
First off, they were an aggressively private entity, who released very little info, and guarded their data behind a pay wall, meaning that I would have to pay fight by fight for data, and given the volume of data required, that would not be cheap.
It is also worth noting that I have contacted them previously asking for a larger data source, and a quote and never heard back.
Secondly, when I read about their methodology, I instantly became concerned about how they recorded the fight’s action, with a reliance on 2 human beings hitting a button to signify a landed or missed jab, or a landed or missed power punch.
The fact that they only measured the fight in terms of Jabs and Power Punches seemed extremely limiting, as the aspects of ring coverage and movement, Blocking and Evasion, Punch Speed and Punch Location would all provide a lot useful information, and are effectively ignored.
I also know from having done notation analysis at football matches before that subjectivity quickly becomes an issue, was a cross that veered towards the goal a shot on target? Was a tackle that sends the ball to a team-mate a pass? Is a clearance that becomes a through ball a successful assist?
All of this led me to conclude that initially at least, I would need to zoom out and work with a lower level of detail than I ideally would like.
Fortunately, Boxrec is a much more user friendly and open site, and although it lacks a ‘download raw data function’, it is still a very helpful resource, and allowed me to look at fighters in terms of their record, and the data available.
NB: The one issue to remember however, is that the shorter the boxer’s career, the less useful the data will be, as you need as many points of data as possible to be able to draw a decently strong conclusion, which is a recurring issue in analysing boxer records.
How do we use Fighter Records for anything useful?
So, I had made the decision to use the fighter’s records, as it was the only source of verified data free of bias or subjectivity, but now I had to see what conclusions could be drawn from this data, and how it may be used to indicate future performance.
The Record itself can tell you only some of the story, A win% based on a fighter’s wins, losses and draws tells you the basics about whether a fighter is rising contender, a journeyman or something between, but as we are all aware, promoters have long been savvy about manipulating records so it can’t always be taken at face value, so this is where we need to consider other aspects to mitigate against this concern.
Verifying a fighter’s record can be done by considering their past opposition, and considering aspects like their oppositions’ win% at the time of the fight, their oppositions number of fights, and their oppositions’ wins and losses by KO, which may indicate their power and their punch resistant to a slightly blunt degree.
It is also possible to map out a fighter’s career by measuring some statistics over time, such as for example, their average opposition’s win%, which I once used to compare Errol Spence’s career to Kell Brook’s, and in doing so, was able to show how much quicker they were moving Spence through the levels than Brook had been.
Stoppage Wins v Stoppage Losses
Another way that we can infer as to a fighter’s ability is to look at their wins by stoppage and losses by stoppage.
Though hardly conclusive, a fighter’s win by KO ratio gives some indication as to the relative power of a fighter, particularly when balanced against the Opposition win% data we have discussed earlier (since some fighters may fight deliberately weaker opponents which could skew their KO%).
Similarly, the losses by KO% can shed some light on a fighter’s punch resistance, and whether they tend to lose based on points or stoppages, and can factor into predictions against a fighter with a high KO%.
Trending improvement and Degradation
Another useful tool can be to compare recent fights to the rest of the fighter’s record to see if there is a different trend or pattern to the career average, thereby indicating a recent change in performance.
For example, quite often a fighter’s win by KO% will drop as their career progresses, as they face better opponents, and that is quite easy to factor in, but it can also drop very sharply indeed once they begin to degrade, meaning that you can sometimes spot a fighter that is on the decline by a sharp drop in power (aka a drop in KO%).
Similarly, you may notice fights going more rounds before stoppages, or a series of inexperienced opponents in succession that can both indicate a declining fighter, or a fighter rebuilding confidence.
What doesn’t the record cover?
The record doesn’t tell you about a fighter’s style, weakness against certain styles, tactical awareness, changes in training, punch accuracy, fitness levels etc so there is an awful lot of data missing from such a top-line analysis.
It also can’t tell you anything about their potential, exact levels of degradation or technical improvements other than in the form of direct results as already discussed, and even then not explicitly.
Examples of when the analysis gets it right
I wrote an article about Usyk v Huck recently, where I proposed that the record of Usyk suggested he would be able to defeat Huck comfortably, based upon the high standard of Usyk’s opponents so far, and Hucks seeming degradation based upon recent results.
“My measures come up with a sort of 60/40 ratio in Usyk’s favour, as despite Huck’s hardiness and tremendous competitive spirit, I feel there are signs in his record that he is starting to decline, and whilst he can still compete at the top level, I believe Usyk is probably the stronger fighter in terms of his development, his momentum, and his condition at this time.
I suspect Usyk will dominate this fight, with crisper work, and win a clear UD, but I really wouldn’t be surprised to see him get the stoppage if Huck doesn’t bring his best”
I had officially gone for an Usyk UD, because there hadn’t been enough pro-level data to draw big conclusions on Usyk’s power yet, and Huck was an experienced fighter who had only been stopped twice, but I had indicated that it was quite possible Usyk would stop him late.
With Huck we had a lot of data for a boxer, as he had 45 fights to draw data from, but with Usyk, we had a short but impressive collection of stats that we couldn’t be sure were sustainable on a career’s basis.
Ultimately what the numbers suggested was that Usyk should win comfortably, and he did, so happy days!
I had also looked at Dorticos v Kudryashov, and concluded that they were evenly matched based upon their records, but had sided with Kudrashov based upon the fact that his past opponents were of a considerably better standard than Dorticos’, and that he had been the more active fighter in recent times.
“In his most recent 10 fights, Dorticos has faced an average win% of 73%, so a bit better than his overall percentage. Kudryashov's last 10 average 85% however, so there still seems to be a large disparity in the level of opposition these fighter have faced, suggesting Kudryashov may have the edge in top tier boxing…I think I am going to stick my neck out, and go with Kudryashov, as I feel Dorticos’ record has a flimsy feel to it, whereas I think we can trust Kudryashov’s as a decent reflection of where he is”
So I couldn’t have been more wrong here…as Dorticos straight up flattened Kudryashov without much trouble.
And this is a good example where the record concealed some useful information.
For example, if two fighters both KO an opponent in the 5th round, record-wise they will show identically in the data.
However, one fighter may have been outboxed, cut and downed before getting the one punch KO, and the other may have already flattened their opponent 3 times in that fight.
Similarly, I have no doubt that Kudryashov fought opponents with a better average record than Dorticos, but it’s a lot hard to prove a fighter’s potential than it is their limitations, and although Dorticos may have been bowling over a lower class of opponents than Kudryashov, that is not to say that his overall ability is not of a better standard ultimately.
Data analysis requires several elements that come up light in boxing, in terms of the available information, and the quality of that information.
Bias is rife, as is subjectivity, as proven in both the Canelo v Golovkin and Parker v Fury fights with regards to the judges scoring.
I would like to conclude with a clumsy and pretentious metaphor if you will indulge me?
Forming a statistical, analysis driven model of a fighter is akin to painting a portrait, where you need an array of brushes, paints and techniques to achieve a realistic, detailed painting.
As it stands, performing quantitative analysis in boxing is akin to painting a portrait using one frayed brush, 3 thickening paints and a canvas made of polystyrene, meaning the end result is a simplified, unfocused version of reality.
It is the job of analysts like myself to try and find solutions and work-arounds and hopefully that is something that will improve with the greater profile of the sport at this time.
Posted: 29th Sep 2017
None at present...
Stats, Boxing, Compubox, Analysis, Steve Fearon, Hellraiser Boxing, Weekend read, hellraiser, hellraiser boxing, hellraiser fight nights, fight nights, professional boxing, London boxing, London, boxing, boxing events, London events, London boxing matches, London entertainment, London sport, London sport event, boxing matches, professional boxers, celebrities, corporate hospitality packages, corporate hospitality, hospitality packages, corporate packages, private event, corporate branding, corporate sponsorship, sponsorship, boxing sponsorship, boxing sponsor, Mickey Helliet, bespoke events, boxing revolution, accessible boxing, boxing manager, boxing promotor, Eurosport, Sky Sports, ESPN, BoxNation, fight cards, Coronet, Camden Centre, Troxy, VIP guest lists, VIP, guest list, guest list boxing, VIP boxing, alternative entertainment, London alternative entertainment, Elliott Mathews, Elliott Mathews boxing, michael helliet management, mhm, michael helliet