The Data Void and the Art of Saying 'Not Enough Evidence' in Golf

Last dry season, I opened an analysis file for a match-play golf event in Southeast Asia. The Strokes Gained column was empty. The driving-distance column was empty. The greens-in-regulation column was empty. Three spreadsheet pages, over two hundred rows, and not a single cell thick enough to say anything. A former colleague called me after the meeting: 'So what did you conclude?' I answered: not enough evidence. The line went quiet for a few seconds, then he laughed. I understood that laugh, because I had heard it many times. In this profession, an empty answer is treated as a sign of someone who cannot do the job, not as a conclusion. Three years of tracking domestic golf taught me something no classroom ever did: the absence of data is itself data, and often the most important kind. The problem is that the golf-analysis world has never been trained to read that absence properly. People are taught how to calculate, how to chart, how to spin a story out of the numbers already on hand. They are not taught how to say 'I don't know' without feeling like they are confessing failure. Golf has gone through more than a decade of data explosion. The turning point came around 2026, when the Strokes Gained method was broadly introduced and later systematized in Mark Broadie's Every Shot Counts, published in 2026. Before that marker, people viewed a round through crude numbers: total strokes, fairway rate, average putts. After it, they began splitting a round into four segments — off the tee, approach, around the green, putting — and assigning every shot an expected value based on ball position and distance to the hole. That revolution is real and valuable. It allows an individual to be separated from a collective result — something that does not exist in individual golf but does exist in match-play formats like the Ryder Cup or Presidents Cup. It allows a golfer who plays a links course on the Scottish coast to be compared with a golfer who plays parkland in the middle of a continent. It allows the claim that a golfer is scoring well because of a hot putter, not because of improved approach play. But precisely as data became abundant, people began forgetting its limits. A common mistake: assuming everything important can be measured, and everything measured is important. Meanwhile, most of what decides a round of golf lies outside the spreadsheet. Pressure on the 18th hole of a major. The feel of the ball when the temperature drops below ten degrees. Confidence lost after three straight missed putts. A practice session cut short by rain. None of these appear in any Strokes Gained column. Start with the small-sample problem, the one I encounter most. A golfer holes putts on twelve consecutive holes in a tournament. The media calls it 'form'. Predictive models start raising his win probability. But if you look at the sample size, twelve holes amount to roughly three rounds of putting — a nearly meaningless number when you know the variance of putting skill at tour level is very large. Many long-run studies of golf data have shown that putting is the least stable of the four segments, and also the hardest to forecast. A hot streak lasting three rounds mostly speaks of luck, not ability. What is striking is that both fans and part of the analytics world do not want to hear that. They want a story. They want to believe the golfer 'found his feel'. And when I write that the putting streak is not enough evidence to conclude anything, I am treated as a spoilsport. The rule I set for myself: a variable is only called a signal when it reappears in at least three independent samples, or across a minimum of three different tournament cycles. Below that threshold, it is noise, and any conclusion drawn from noise is a fallacy. The second problem is course fit — the match between a golfer and a course. This is a data zone where regional golf, Vietnam included, is almost entirely empty. A golfer can dominate a parkland course with wide fairways and slow greens, then collapse on a coastal links course where wind blows sideways and the ball runs long on dry grass. To assess course fit seriously, you need data on fairway width, green speed, grass type, seasonal average wind direction, elevation above sea level, and bunker placement. At many major events, this data is collected rigorously. But at most courses in Vietnam and Southeast Asia, it does not exist in retrievable form. That means when I assess a golfer preparing to compete in Nha Trang, I am working with an empty data column. I know the golfer has good numbers on his own tour. I do not know how he will react to the Cam Ranh sea breeze in the afternoon, when direction shifts and humidity spikes. No data answers that question. And the honest answer is: not enough evidence. That emptiness is not the golfer's fault, nor mine. It is the fault of a data-collection system never designed for regional golf. And the only way to handle a systemic fault is to name it, not to fill it with guesswork. The third problem is the weather variable, which standard data usually ignores entirely. In a small study I once took part in, I compared results from rounds played in strong wind with rounds on the same course in calm conditions. The results showed GIR rates falling sharply once wind speed crossed a certain threshold, and the drop was uneven across golfers. Some lost very little performance in wind; others lost nearly half. But to know this, you need hourly weather data matched to each shot. Most public stat sheets do not have it. This is an example showing that empty data does not scatter randomly; it clusters where measurement is hardest, and those are precisely the places that decide outcomes. People measure what is easy to measure, then assume the easy measurement is the important one. That is the fallacy of the drunk under the streetlight, searching where the light is instead of where the keys fell. The fourth problem, and the largest structurally, is the asymmetry of data across tour systems. Since LIV Golf appeared, the golf world has split into two competing systems with two different data infrastructures. The PGA Tour has detailed, standardized, relatively openly published statistics. LIV Golf, with its match-play format and different schedule, offers a dataset not directly compatible. When a golfer moves from one system to the other — as Jon Rahm did when he joined LIV Golf in December 2026 — his numbers cannot be compared one-to-one with the past. You cannot place side by side a Strokes Gained figure calculated over a long schedule and a similar figure calculated over only a few events. The consequence is that the world rankings and every model built on them carry a hole that no algorithm can patch. This is not an arithmetic problem. It is a philosophical one: you are comparing two things never measured with the same ruler. And this is where I must state my view clearly. Talent-valuation models in golf — as in football — tend to overrate youth potential and underrate unmeasurable factors, such as harmony within a support team or the ability to handle pressure on the final hole. In golf, a 'locker room' does not exist in the team sense, but an equivalent does: the caddie, the coach, and the sports psychologist. A young golfer with impressive numbers but no suitable team often fails to convert potential into titles. No metric measures that. And because it cannot be measured, the model treats it as zero. Over three years of tracking domestic golf, I have seen this repeat often enough to believe it is a rule, not an exception. The golfers who go furthest are not the ones with the highest numbers at twenty-one. They are the ones with good-enough numbers plus a support system that appears in no spreadsheet at all. Tiger Woods, with 15 major titles and 82 PGA Tour wins, is a historical outlier, not a model to copy. Scottie Scheffler rose on approach play and driving stability at an almost unimaginable level — but even he has admitted the deciding factor is in the head, not the hands. Here I must say plainly what few in the profession dare to: the sports-analytics industry rewards noise and punishes silence. An expert who makes a bold, wrong prediction, then makes another bold one, will draw more engagement than one who says 'not enough evidence' ten times in a row. The media needs stories, and stories need conclusions. An empty conclusion does not sell ads. So structural pressure pushes the analyst toward saying something, anything, as long as it sounds decisive. The result is a paradox: more data, more unfounded conclusions. Because abundant data creates the feeling that everything can be known, admitting not-knowing becomes more uncomfortable. People fill empty cells with speculation that wears a scientific look. They call it a model. I call it decorated fabrication. Audiences clap to emotion, but data hears a different rhythm. And that rhythm is often a silence. My job is not to fill that silence. My job is to name it, measure it, and state clearly that it exists. I do not need recognition in the newsroom; the numbers know their own way to tell the story. A report saying 'not enough evidence' today may become the most correct report next month, when the late data finally arrives and confirms the gap was real. Data is never in a hurry; it only waits for those who know how to read it. In the coming season, I expect the golf world to face an increasingly clear paradox: the more numbers there are, the more unmeasured empty zones there are. Whoever grasps those empty zones before others will hold an advantage that comes not from algorithms, but from honesty with themselves. I write the report, close the file, and the market reopens on its own. And in the time between those two moments, what I do is wait for data, not wait for inspiration.