The Empty File in Combat Sports: When Data Is Missing, Every Verdict Is Just Noise

The file sits on my desk in Guangzhou. I printed it out because I still read paper before screens. Fourteen fields: date of birth, height, competition weight, off-season weight, injury history, number of bouts in the last eighteen months, average weekly training load, rest days between the two most recent bouts, surface type, head-contact count across the last three bouts, pre-weigh-in weight cut, joint function test results, team doctor's notes, and one blank line for video description. All fourteen fields read N/A.

The sender was a promoter. They needed a safety assessment on a fighter ten days before a semifinal. I read the file twice and answered in four minutes. No data, no assessment. That was the only honest answer I could give.

Half an hour later, the phone rang. "Just watch the video, you'll see it right away." I watched. Three and a half minutes of highlights, all beautiful moments, not a single frame showing how that fighter walked in the third minute of the final round. A highlight reel is advertising, cut to sell tickets. In my profession, an empty file is one of the loudest warning signals there is, louder than an abnormal test result, because an abnormal result is at least information.

In July 2026, in Kazan, I sat in a small studio and said into a microphone that Neymar had lost 12 percent of his change-of-direction ability in the second half, that his left thigh responded 0.3 seconds slower than his own first-half baseline. The stadium believed the opposite. Brazil lost 1-2, and my listener count rose 300 percent overnight. That night in Kazan taught me that public opinion is noise and numbers are signal. It taught me a second thing too, less often repeated: I could only say those sentences because I had twelve matches to measure. Had the organizers handed me a sheet with fourteen N/A fields, I would have stayed silent for the whole match.

Two sports sharing one label

In almost every data system I have touched, "martial arts" is the worst label available. It is enough to schedule a broadcast and sell a ticket. It is not enough to assess a body.

Sanda runs three rounds of two minutes, with throws, kicks to the thigh and body. Amateur boxing runs three rounds of three minutes, hands only, no kicks, no grappling, but a far higher count of direct head impacts. Muay Thai runs five rounds of three minutes, permits elbows and knees at close range, and a well-placed elbow splitting an eyebrow is routine. MMA runs three five-minute rounds, or five for a main event, with takedowns, chokes, and strikes landed on a downed opponent. Wushu taolu hits nobody at all: a form lasts roughly one minute twenty seconds, and the entire risk sits in the landing, in the ankle, in the patellar tendon, in the lower back.

Four sports, four different injury maps, folded into one box. When the sport field on a file says only "martial arts," my search area is close to zero. I do not know where to look first.

This is a technical question with a concrete answer. In taolu, what I need to count is landings and the height of each. Over one six-day training week for a taolu athlete I was tracking, I logged 312 landings, 84 of them single-leg from above one metre. That means every week, that athlete's Achilles tendon and patellar tendon absorbed 84 controlled compressions, plus hundreds of smaller ones nobody counts. Nobody sees that number on television, because it happens off camera, at six in the morning.

In boxing, what I need to count is head impacts, including the ones that knock nobody down. A fighter can win three straight bouts looking flawless while quietly accumulating a debt nobody is recording.

Vietnam has both worlds and is strong in both. Vietnamese wushu taolu has collected SEA Games medals for years, with names like Duong Thuy Vi. Boxing has produced athletes who reached continental and Olympic stages, such as Nguyen Thi Tam at 51 kilograms. Domestic professional MMA has LION Championship, launched in 2026. These three streams flow through three different medical systems, three different measurement habits, and they barely speak to each other in the same data language.

I once sat in a four-person meeting: a coach, a doctor, a videographer, and me. We discussed the same fighter for forty minutes. The coach talked about spirit. The doctor mentioned a clicking shoulder. The videographer mentioned that the fighter no longer rotated his hip through full range on high kicks. Three people, three languages, and nobody wrote anything down. Had I asked fourteen questions matching the fourteen fields of the file, we might have produced data. But nobody asked, because nobody was responsible for filling the blanks.

An unclassified domain label does more than inconvenience a reader. It breaks the entire processing chain behind it. When the input simply reads "martial arts," every downstream step points the wrong way: load thresholds from one sport get applied to another, recovery windows are calculated from combat-round structures while the athlete actually performs a single form lasting over a minute, and head-impact metrics enter a file in which no head impact ever occurred. The chain still runs. It still prints a report. It still draws a beautiful chart. Only the conclusion is meaningless.

Why an empty file is a risk, not a neutral fact

There is an assumption I run into everywhere: if no bad information exists, everything is fine. In matchmaking, emptiness gets read as "probably fine." That is a systematic bias, and it operates in the most dangerous direction possible.

In 2026, while working as a commentator for a television station in Guangzhou, a club asked me to review the injury record of a Brazilian striker ahead of a long-term deal. That file was not empty. I had 47 matches across 18 months, GPS data from training sessions, and I found that his sprint output dropped 15 percent on artificial turf compared with natural grass. I advised the club against a long-term contract. Six weeks later he tore a hamstring in a match. The point is not that I guessed right. The point is that I could only guess at all because the file contained something to read. Same person, same body, but had I received a sheet with fourteen N/A fields that day, the only honest conclusion would have been: unknown.

The distance between those two situations is the distance between medicine and fortune-telling. Injury data never lies; only readers lack patience.

Four kinds of missing data

Missing baseline

In December 2026, a 21-year-old ONE Championship fighter named Yang Jian Bing died after a weight cut. The incident forced the industry to re-examine how it measures fighters. From 2026, ONE Championship introduced urine specific gravity testing alongside multiple weigh-ins instead of a single one, to block extreme dehydration before competition.

Here is the key point: with only one weigh-in, a fighter who drops 8 kilograms in ten days looks identical to a fighter cutting weight healthily. You need a series. Three weigh-ins across two weeks, plus an off-season weight, reveal who is dehydrating and who is losing fat. A missing baseline does not make a file look worse. It makes a file look clean.

For Vietnamese combat sports, the baseline gap sits elsewhere: there is no injury register that follows an athlete across a career. A child starts taolu at ten, reaches twenty having passed through four coaches and three training centres, and nobody has kept a record of the ankle pain episodes. When that athlete reaches a major event, the file starts from zero.

Missing time series

In 2026, when the pandemic postponed domestic competition and the stands held nobody, all my commentary contracts were cancelled. I contacted 23 young athletes, collected sensor data from their home training sessions by phone, and spent eight months building a load-and-recovery model, centred on the ratio between accumulated training volume and actual rest days. I tested it on my own body before applying it to anyone else, because that is the only way to know whether I was fooling myself. When competition resumed in June 2026, the squad recorded four injuries in its first ten matches, roughly 30 percent below the average of the two previous seasons.

I tell that story to show the other side too. Because I am poor at long-term planning, the model ended up scattered across twelve spreadsheet files and was never widely adopted. Technically, the data existed. Institutionally, it did not. A time series fragmented across twelve files is, to a decision-maker, the same as a blank sheet.

A single data point is a photograph: it says how today looks, not where things are heading. A series is film. Career-ending injuries almost always appear as film, never as a photograph. Anyone who reads bodies the way I do knows that every ache is an answer. The problem is that the question must have been asked months earlier.

Missing rules and procedure context

In many American states, fighters receive medical suspensions of 45, 60, 90, or 180 days after a bout, or an indefinite suspension, depending on documented injury. The gap between a 45-day and a 180-day suspension is the gap between a laceration that needs rest and a fracture that needs intervention.

That context matters because rules apply locally. A fighter suspended in one jurisdiction can be booked in another if the two systems do not share data. This is the most dangerous category of missing information, because it does not live in the fighter's body. It lives in the administrative structure.

In Vietnam, LION Championship has professionalised weigh-ins and medical checks since 2026, and I credit that as real progress. But most grassroots, youth, and provincial events still operate in isolation. A fighter can compete three times in two months in three different provinces, and no single body sees the whole series. What is missing is not the organisers' enthusiasm. What is missing is a shared register.

Missing comparison group

Every percentage needs a denominator. I once heard a coach state that his athlete had a patellar tendon injury rate "12 percent higher than normal." I asked three questions: higher than whom, across how many people, measured over how long. No answer came. The figure had been copied from an article about a different athlete group, in a different sport, at a different age.

A 19-year-old taolu athlete and a 31-year-old MMA fighter can both show a 12 percent decline in some metric. For the 19-year-old, it may signal a growth phase. For the 31-year-old, it may signal a degenerating tendon. Same number, two meanings, because the comparison base differs.

In sports medicine, most injuries occur in people carrying no risk flags at all. That is why a model without a comparison base is no different from a horoscope presented as a chart.

2026 taught me one more thing. An empty stadium does not make a fight cleaner; it only exposes the truth more starkly. With no crowd noise, every breath at the fourth minute became audible, and I began to hear signals that the roar had previously covered.

What the data does not see

This is the part I have to remind myself about, because I belong to the spreadsheet-believing camp.

The industry's problem is not a shortage of data. The problem is the demand for a verdict at any cost. The matchmaker needs an answer to sell tickets. The writer needs an answer to publish. The bettor needs a number to act on. Inside that chain of demand, the person saying "not enough data" is treated as useless. I have lost contracts for exactly that reason, and I consider it the price of staying in the profession.

Some things sit beyond the reach of data, and I list them here so that every conclusion I write knows where its blind spot is. First, fear of losing a contract: a fighter nearing expiry will hide pain, and pain-hiding is common enough that every model must assume it. Second, family pressure: for many young fighters, one bout's purse equals half a year of household income. Third, the coach's incentive, which may be tied to results rather than to rest days. Fourth, a culture of enduring pain, taught as a virtue and paid for in tendon and cartilage. Fifth, the growth phase of young athletes, whose bodies change faster than any table. Sixth, and this is what I think about most, post-injury psychology: someone who has torn a ligament moves differently afterwards, and no sensor measures that.

There is a trap on the opposite side too, among people who trust data too much. Twenty-three athletes, eight months, twelve spreadsheets can produce a very beautiful curve. A beautiful curve is not a correct one. With a sample that small, one athlete changing a training schedule for personal reasons bends the entire curve. I have drawn such curves and had to erase them myself, and I consider the erasing the most important part of the work.

My contrarian point sits here, and I rarely hear it raised at sports medicine conferences. The largest determinant of injury in combat sports is not technique, not surface, not protective gear, but competition calendar density. No medical team on earth saves a fighter competing twice a week. The 2026 spreadsheet taught me this: the body does not rest, only the algorithm needs to be patient enough.

I once reviewed a sequence of three bouts in forty-five days by a young fighter. In the first, his strikes were fast. In the second, he still won but closed distance less often. In the third, he lost on a sequence he himself could not explain, wondering why he was slow. Nobody on the coaching staff treated it as injury, because nothing tore and nothing broke. But sensor data showed his hip rotation range narrowing week by week, until it vanished entirely in the third bout. His body had been answering since bout one. Nobody asked the question.

One more point about the limits of this trade. When I say "not enough data," that is not an attitude, it is a finding. It resembles a doctor holding a blurry scan: the honest one says the film is unreadable, the one seeking to please the family reads it anyway. In both cases the patient's body is still there, indifferent to what anyone wants to hear.

Takeaway

If I could set rules for everyone with the authority to sign a fighter onto a card, I would set three.

One, a file with blank fields must not be read as "fine." A blank must be read as "unverified," and "unverified" must mean not cleared for high-density scheduling.

Two, every percentage must carry a denominator, a measurement window, and a comparison group. Without those three, the number should be struck from the record.

Three, the person signing must be the person who read the data series, not the person who read someone else's conclusion. The distance between those two people is the distance between a medical examination and an administrative stamp.

I still keep that sheet of fourteen N/A fields in a drawer. I do not keep it to blame anyone. I keep it because it is the clearest illustration of what I believe: the greatest value of a blank field is that it forces a slow answer.

Ten days later, the fighter in that file may walk out and win beautifully. That would not prove the empty file was safe. It would only prove that this time we were lucky. And in my profession, luck is the worst thing to enter into a spreadsheet.