SwimSwam's 2028 Recruiting Database: What the Spreadsheet Measures and What It Misses in a Swimming Talent

In a recruiting file I reopened last week, a 15-year-old girl held the fastest 100-meter butterfly time in her state's age group. Her peak form landed exactly in the month when college programs open contact. Just below, the coach's note ran a single line: "She swims the 200 far better than the 100." Two data points sit side by side on the same page and tell two different stories about the same person. I have sat beside the pool deck for 32 years, from swimming features for Thanh Nien newspaper in 2026 to tracking spreadsheets in the V-League, and the lesson has not changed: a dataset is honest about what it measures, and never honest about what it leaves out.

Context: a product inside a recruiting ecosystem

SwimSwam is one of the most widely read specialist swimming outlets in the United States and is generally a credible source in the field. I am focused on one specific product here: the 2028 Recruiting Database, run by Anne Lepesant. Lepesant is a veteran SwimSwam writer who has tracked the American college recruiting market closely for years.

One thing needs stating up front: this product introduction blends two functions. It describes what the product does, and it sells the product. I do not read it as a neutral document but as a piece of data that needs cross-checking before use. That is why I place medium confidence on the description, even though the source itself is reputable.

To understand the product, you have to understand the system it serves. American scholastic swimming runs on two kinds of pool. The high school and college seasons swim in short-course yards, measured in yards. Summer and international competition swims in a 50-meter long-course pool, measured in meters. A single athlete can carry two parallel sets of marks that do not convert cleanly between each other. Colleges recruit mainly on short-course results, because that is their competitive surface.

The class of 2028 consists of students graduating high school in 2028 and entering college in autumn 2028. The database therefore begins tracking them around freshman or sophomore year of high school, from roughly 2026 and 2026, and will keep running for several more years. The recruiting system has two official signing periods, an early one in November and a regular one in April. Before that sits the verbal commitment stage, when coaches and athletes agree without a binding document.

Within that ecosystem, data platforms of this kind also use a standardized index developed by USA Swimming, commonly known as Power Points, which converts marks across events onto a common scale for comparison. It is a useful tool, and also a place where misunderstanding creeps in: a standardized score looks more objective than it is, because it still rests on raw marks that do not account for age or the conditions in which they were set.

An aggregated database has real value. It pulls marks scattered across hundreds of meets into one place, tying names to schools, coaches, and commitment status. On pure convenience, there is nothing to criticize. The problem lies elsewhere.

What the data does measure

I am spending most of this piece on what the spreadsheet cannot measure, because that is the hard part. But first, fairness to what it measures well.

A personal best is the basic currency. A recruiting database records each athlete's best time in each event, usually with the date and the meet name. This data is verifiable, because meet results are public. Technically, it is the cleanest data in the entire ecosystem: a number either exists or does not, with no grey zone.

The second layer is commitment status. Who has accepted which school, who remains open, who is drawing interest from multiple programs. This carries high news value and is the main reason readers return to the tracker every week. In nature, it is event data, not capability data.

Both layers are useful. But a time on its own says nothing about where that number is heading. And that is the question every college program actually needs answered: not how fast this athlete swims today, but how fast she will swim four years from now.

That is where the spreadsheet is blindest.

The compressed age variable

Take two athletes who both swim the short-course 100 butterfly in 54 seconds. One is 15, the other 17. On the sheet, the two rows are identical. But the 15-year-old's developmental headroom is far larger. The 17-year-old is close to the natural ceiling of her age group; the 15-year-old has two more years of physical growth, two more base-training seasons, two more taper cycles. One number, two entirely different values. No column prints that difference.

I call it the compressed age variable. It sits inside the data but is never encoded as its own metric. An experienced reader separates it out; a new reader sees two 54-second rows and assumes they are equivalent.

In the same family of problems sits the time-drop metric, the improvement between two bests. It is an attractive metric because it tells a story of progress. But time drops depend heavily on the starting point. An athlete who posted a weak mark at 14 will show a more dramatic drop than one who swam near her optimum early. On the numbers, the first looks more promising. On real potential, not necessarily.

Pool and conditions

A 25-yard short-course pool and a 50-meter long-course pool create two different ecosystems of results. The same athlete, the same event, will show meaningfully different times because the number of turns differs. Converting between the two systems is an estimate, not an exact calculation. When a database blends marks from both systems without labelling sources, it creates a silent error: readers compare two numbers that are not in the same unit.

I have made exactly this kind of error at another scale. My 2026 mistake reminded me that data is a mirror, not a lamp. A mirror reflects what I bring to it; a lamp reveals what I have not yet seen. A good database is a very sharp mirror. It is not a lamp.

Conditions matter just as much. Not every number is born in the same setting. A mark set at a championship, after weeks of reduced training volume to peak, is qualitatively different from a mark set at a mid-season dual meet while the athlete is under heavy load. People inside the sport distinguish the two. A spreadsheet usually places them side by side without distinction.

Individual marks also fail to capture speed in relay events. An athlete can be a critical link on a relay team through her starts and turns, while individual metrics show nothing of it. Relays are where technical skill — the start, the underwater, the turn — shows most clearly, and also where individual data is blindest.

Numbers and sample size

There is one more variable I always watch: how often an athlete races. A peak mark set in a season where the athlete raced three times means something different from an equivalent mark set in a season with fifteen races. More races mean clearer consistency, and consistency is what converts into scoring at a college championship. A single breakout says nothing beyond the fact that the athlete can break out once.

I learned this lesson from a different dataset in a different sport. In 2026, I dissected Hanoi FC against Thanh Hoa in round 18 of the V-League. Hanoi FC completed 612 passes and held 58 percent possession, yet registered only three shots on target. On the aggregate numbers, the team played well. On structure, the defensive line pushed so high that the centre-back often stood 30 meters from the goalkeeper, and that was the gap opponents exploited. Data only recounts; tactics begin with the mistake. A swimming recruiting sheet faces the same trap: attractive aggregate numbers can hide a structural flaw located somewhere else.

So what does the database do best? It aggregates and tracks status. Who committed to which school, when, with what marks. That is real informational value, useful for fans, parents, and programs watching rivals. Its value lies in the flow of information, not in its predictive power.

The execution blind spot

The counterintuitive part sits here: we tend to believe that the more complete recruiting data becomes, the more accurate the forecast. That holds for data with a causal structure, such as training-load calculation or energy-metabolism analysis. It does not hold for recruiting data, because what determines an athlete's college success largely lies outside the columns.

Immature physical development, the speed of adaptation to a new training load, the ability to handle competitive pressure, the quality of coaching at the destination school, the stability of the living environment, a history of shoulder and knee injury — none of it is in the spreadsheet. I do not believe in hunches. I believe in how many variables the hunch had loaded into it. Here, the number of variables loaded into a recruiting database is far smaller than its appearance suggests.

There is another layer readers should recognize: this product is introduced by the very article that discusses it. That is a common and legitimate media model, but it raises a question of frequency: are the athletes mentioned most often in the database genuinely the best athletes, or merely the most closely tracked? In the data industry, those two questions are routinely conflated, and that is where belief gets steered without the reader noticing.

The transfer market is a vast error map. The wise look for the blind spot, not the treasure. Recruiting swimming is the same. The blind spot here is the distance between a mark at 17 and potential at 21, a distance no column of numbers can close.

Thinking forward

I am not writing this to diminish a data product. I am writing to place it correctly on the desk: it is a lookup tool, not a forecasting tool. The best user is one who knows that every number arrives with an unanswered question attached.

For Vietnam, the lesson is not distant. We are entering the stage where youth swimming results are being digitized, and the greatest temptation is always to look at a number and draw a conclusion about a person. Stepping into the world of data, I learned to stay silent in front of the numbers. That may be the earliest skill worth teaching anyone who wants to work with sports data, whether in swimming or football.

When the 2028 recruiting sheet closes in a few years, people will know how many names on it truly reached the top. By then, the interesting question will not be who guessed right, but what the data left out along the way.