The Line Between Analysis and Guesswork in Swimming

Brisbane, 3:12 a.m. On the screen is an analysis table for a swimming final, and every cell is empty. No 50-meter splits, no reaction time, no underwater propulsion figures, no turn notes. Just one phrase repeating: N/A.

I sat there for fifteen minutes. Fifteen minutes is exactly enough time for an analyst to start filling in the blanks. The most sophisticated form of filling-in always looks the same: a plausible number dropped into the empty cell, a "probably" dressed up as reasoning, an industry average nobody verified. An empty table is harmless in itself. The confidence that fills it is what causes the damage.

I know this because I paid for it. "Kazan was the day I learned that a 99 percent probability can still die on the betting table." But Kazan comes later. First, let us talk about the empty table.

Swimming Is a Sport Whose Most Important Data Sits Where Nobody Looks

A 200-meter race can be captured by dozens of numbers within two minutes. The automatic timing system returns reaction time, six 50-meter splits, per-lap times, and sometimes stroke rate. On paper, it is a feast of data.

But the number that decides victory often is not in the results sheet. It lives in the underwater segment after the start and after each turn — the part television cameras cut away from because "there is nothing to see." A swimmer can lose 0.4 seconds over the first 50 meters on a slower stroke rate, then take it all back in fifteen meters of dolphin kicking beneath the surface, with no split ever recording it. Viewers see the result. Analysts must see the cause.

The problem: the cause only surfaces when enough raw data exists. When it does not, an empty cell stays empty. No interpolation turns a table of N/A into an honest conclusion.

In recent months I have watched how swimming analytics platforms handle season data. Most do the presentation well: clean charts, tidy comparison tables, attractive colors. But a troubling habit is spreading: when data is missing, people fill it with assumptions, and assumptions quickly get read as facts. That is the moment an analysis table becomes a false indictment.

Raw Data First, Conclusions Second

My first rule, set in 2026 after a night at Suncorp Stadium, is simple: data first, emotion second. I was the only female analyst in the press room then, publishing a prediction that Melbourne Victory would win despite trailing 1-0 at halftime, based on xG of 2.4 versus 0.6 and distance covered of 112 km versus 98 km. A male commentator smirked: "Sweetheart, football is not mathematics." Final score: Melbourne won 2-1.

I retell that not to boast. I retell it to say: "Numbers have no gender, but the people who read them do." The same xG table, read by a man in a press room, produced luck; read by me, it produced a broken pressing model. The data did not change. The interpreter did.

That principle applies intact to swimming. A 50-meter split means nothing on its own. It means something only beside reaction time, stroke rate, distance per stroke, and which lane it happened in. Pulling one number out of its sequence is the fastest way to tell a false story that sounds persuasive.

And here is what I must say plainly, even though it irritates some people: an analysis table full of numbers but short on sourcing is more dangerous than an honest empty table. An empty table forces humility. A full but rotten table gives you the feeling that you already understand, when in fact you are guessing.

The Nine Layers of Decent Swimming Analysis

For a swimming analysis to stand, it must pass through a series of layers, and any layer short on data must be labeled as such. I call it the map of limits.

The first layer is technique. Here analysts read the start, the underwater segment, the turn, and the finish. A good turn in a 50-meter pool can save 0.3 to 0.5 seconds over an average turn — enough to change a final's ranking. But to assert that, I need per-lap timing, not a feeling that "she turned beautifully." Without turn data, I can only say: this is a hypothesis, not a conclusion.

The second layer is performance and positioning. A time only means something inside a coordinate system: the world record, the all-time list, and the current season ranking. The gap to the world record decides whether a result sits at world-class level or merely decent. Here is a familiar trap: comparing short-course and long-course results while forgetting that more turns in short course create a systematic time advantage. Mixing the two reference systems into one table is a basic error, yet it is everywhere online.

The third layer is the competition system. Is a meet a selection meet? Does a result meet the A-cut or B-cut standard? Which year of the Olympic cycle is this — Olympic year, adjustment year, buildup year, or sprint year? The same result, read in an Olympic year versus an adjustment year, yields opposite conclusions. A swimmer going 1.5 seconds slower than a personal best in an adjustment year is normal. The same margin in an Olympic year is an alarm signal.

The fourth layer is the world map and resources. Who dominates each event? How deep is a nation's talent pipeline? A nation can have a lone star, but if the youth ranks behind are empty, that star is an exception, not a system. Here I must repeat a view I have pursued for years: scouting networks in developing countries can find genius, but they can also manufacture sports lottery tickets and broken families. Swimming is no exception.

The fifth layer is rules and governance, including anti-doping. This is the layer where the absence of information is most easily misread. An article that does not mention doping does not mean the swimmer is clean. It only means the topic was not raised. Silence is not a certificate. I learned this from cross-checking official data: eligibility cases, equipment cases, false-start faults usually surface only after results are published.

The sixth layer is the athlete's career. Where does age sit on the performance curve? For young female swimmers, this is the most sensitive layer, because puberty can stall or reverse results even when the training plan is unchanged. Ignoring this variable is a common cause of hasty conclusions like "she is finished." Then there is injury history — swimmer's shoulder and breaststroker's knee — and the capacity to handle final-night pressure.

The seventh layer is the risk profile. Here I separate two kinds: competitive risk and process risk. The second is rarely discussed but lethal: when the analysis input is empty yet conclusions still emerge, that is data risk, and it is more serious than any professional risk.

The eighth layer is the media narrative. Is a result being inflated by crowd emotion? Can a story survive a sample-size test? German fans attacked me after Kazan, and I know that the heat of a story has nothing to do with its accuracy.

The ninth layer is industry ripple. A good result can lift commercial value, attract investment into facilities, and expand the coaching market. A bad result can freeze a youth development program. Swimming runs along a chain: youth ranks, athletes, broadcasting, sponsorship. A shock at any link spreads to the others; nobody just records it.

When the Score Sheet Is Empty and the Trap of Confidence

The nine layers above sound complete. But they only operate when the input material exists.

In a recent process check, I received a swimming analysis report running through all nine dimensions — technique, performance, competition system, world map, rules and anti-doping, athlete career, risk profile, media narrative, and industry ripple. It sounded formidable. But when I opened the core data section, every cell was N/A. No athlete name, no time, no meet, no source. Nine layers of analysis standing on an empty foundation.

What is notable is that the report still looked "correct" in form. It listed every frame, every heading, every term. But no conclusion could be verified, because there was no data to verify it against. That kind of report, if skimmed, can make people believe someone analyzed very deeply. In reality, nobody analyzed anything.

This is when I think of "The Day Germany Collapsed in Kazan." In 2026, at the World Cup in Russia, Germany lost 0-2 to South Korea and were eliminated despite 74 percent possession. In my piece for a betting site, I pointed out Germany managed only 11 passes into the box and an xG of 0.7 — lower than South Korea's 0.9. I called it the arrogance of the rich who refuse to press. German fans attacked me, demanding I delete the article. A week later, FIFA published official data confirming every number. An Australian broadcaster put me on air.

The lesson from Kazan is not that "data is always right." The lesson is: data is only right when it exists and can be verified. When data is absent, confidence cannot replace it. A 99 percent probability can still die on the betting table, and an analysis table stuffed with words but empty of data can collapse just the same.

The Contrarian Angle: The Enemy Is Not Bad Data, but Certainty

Most people think the biggest risk in sports analysis is wrong data. I believe the real enemy is certainty — the feeling of understanding, when there is nothing yet to understand.

A wrong number can still be caught when cross-checked against a source. But a conclusion built on unverified assumptions is hard to catch, because it does not present itself as an assumption. It presents itself as fact. And once it spreads far enough, people start citing it as a data point.

In swimming, this trap appears where correlation and causation get mixed. A swimmer changes coaches and swims faster, and people immediately conclude the new coach is better. But perhaps she just emerged from a puberty plateau, or recovered from a shoulder injury, or simply matured another year. Correlation is not causation. Numbers do not lie, but they do not explain themselves either.

I remember the Daniel Arzani case in 2026, when a Brisbane betting firm hired me to assess him. I presented the data: average distance covered of 8.2 km per match, below the 10.1 km typical of a Celtic forward; dribble frequency of 2.1 per match; and a history of two ACL ruptures. I concluded the transfer would fail. The sporting director objected, saying I saw people as machines. Two seasons later, Arzani played a mere 20 minutes at Celtic.

"Valuing a player is not a calculation; it is a war between belief and the numbers." I still hold that view. But I must also admit: in Arzani's case, the data was thick enough for me to conclude. Had I only had a few scattered metrics, my conclusion would have been guesswork in analysis clothing. And guesswork in analysis clothing is more dangerous than naked guesswork.

The Limits of Data

Since 2026, after being criticized as mechanical and dismissive of national spirit when I predicted Italy would win the EURO penalty shootout, I have added a fixed section to every analysis: the limits of data.

That section acknowledges what cannot be measured in milliseconds: final-night psychology, pool conditions, a coach's bad decision one morning, luck, and fear. "I do not believe in emotion. I believe in the data series longer than your emotion." But I believe emotion is data too — we simply lack the tools to measure it.

In swimming, the limits are many. We cannot measure the anxiety of a 15-year-old girl stepping onto the blocks before a packed stand. We cannot measure that she has just passed through puberty and stalled, though this is common among female swimmers. We cannot measure a coach changing the training plan mid-season for off-field reasons.

None of this renders data-driven analysis meaningless. It renders analysis demanding of humility. And that humility must be written into the analysis itself, not offered as an apology tacked on at the end for cover.

The Signal for the Next Round

The empty table I stared at in Brisbane that morning stayed empty. I did not fill it. I added a single line: "Data insufficient for a conclusion."

With the swimming season ongoing, what is worth watching is not a meet result but the speed at which analytics platforms admit their own limits. Whoever dares to write "insufficient data" when data truly is insufficient is the one worth trusting next round. As for those who fill every empty cell with confidence, wait a week for the official numbers, as I once waited after Kazan.

Numbers have no gender. But readers full of certainty do — and history has shown they are usually wrong at the very moment they feel most sure.