Counting Caitlin Clark

Part One · The rates

In the 2026 WNBA season, Caitlin Clark averaged 22.3 points and 8.3 assists a game in 40 games for the Indiana Fever. Her scoring ranked third in the league and her assists second, and on October 24, 2026, the league named her to its All-WNBA First Team. Each of those sentences is a count made under a set of rules. This book follows three kinds of numbers around a modern guard: the rates that describe how she plays, the draft and rookie scale that placed her in the league, and the television audiences counted when she played. For each, it asks who counts, how, and what the count cannot show.

Start with usage. Basketball-Reference defines usage percentage as an estimate of the share of a team's plays that a player uses while on the floor. In its formula, a play is a shot attempt, a turnover, or a trip to the free-throw line, which counts as 0.44 of a play for each attempt. Five players share the floor at once, so by our arithmetic an average player sits near 20 percent, because the five shares must add up to 100. In 2026, A'ja Wilson of Las Vegas led the league among qualified players at 32.9, and Clark was second at 32.6. Her career figure through 2026 is 30.2. Usage says how often a player finishes the team's plays, and it does not say how well she finishes them.

Assist percentage asks a different question: of the baskets her teammates made while she was on the floor, what share did she assist? An assist is a pass that leads directly to a made basket, and the percentage divides her assists by those teammate baskets. In 2026 Clark's was 44.1, second to Alyssa Thomas of Phoenix at 45.8. Per game the order is the same, 8.5 for Thomas and 8.3 for Clark, and in total Thomas had 358 assists to Clark's 333. Clark's 337 in 2024 was then the most in a season, and Thomas has passed it twice, with 357 in 2025 and 358 in 2026. A rate measures what happened during her minutes on the floor, and a count measures what happened over the whole season.

A player who uses many plays also loses the ball on many. Turnover percentage estimates turnovers per 100 plays. Clark's in 2026 was 19.8, while WNBA teams averaged 14.4, although a team figure and a single player's are not the same kind of number. Her true shooting percentage, which divides points by twice the sum of her shot attempts and 0.44 of her free-throw attempts, was .600 against a league mark of .560. The pairing shows the trade in a high-usage role: more turnovers per play than the league average, and more points per scoring attempt than the league average. Her effective field goal percentage, which gives a made three one and a half times the credit of a made two, tells a similar story at .536 against .515.

Where do the points come from? Basketball-Reference's shot splits, built from play-by-play, cover 686 of Clark's attempts in 2026, regular season and playoffs together. Of those, 349 were threes, about 51 percent by our arithmetic, and she made .358 of them. Another 106 came at the rim, where she made .642. The other 231 came from the range in between, from just off the rim out to the arc. In the regular season, 88 of her 282 made baskets, 31 percent, carried an assist, so most had none. The site warns that its figures may not match official totals, and a location is not the same as a quality: the defender, the clock and the shooter's own choice shape every attempt, and the table records none of them.

Every rate rests on a sample. In 2025 Clark played 13 of the Fever's 44 games, 404 minutes. Her usage was 31.5, her true shooting .491 and her turnover percentage 23.2. In 2026 she played 1,244 minutes, and the same three figures were 32.6, .600 and 19.8. The two lines are far apart, but statisticians call the random variation in a small sample noise, and a rate computed from 404 minutes carries far more of it than one computed from 1,244. A career line, which blends all of her seasons, smooths that variation and describes none of the seasons exactly. Even 1,244 minutes is only one season, and the next season will add its own variation to the line.

Part Two · The draft and the scale

How does a college scorer become a Fever guard? Through a draft with rules. The WNBA draft has three rounds, and teams pick in the reverse order of the previous season's standings, except at the top, where a lottery decides: four picks in 2022, five in 2026. For 2027 the league has scheduled the lottery for November 15, 2026, to draw only the top two picks, and the draft for April 12, 2027. Fourteen balls go in a machine, four are drawn to make a combination, and the team holding it gets the No. 1 pick. The draw is repeated for No. 2, and an accounting firm oversees the process. Each team's odds depend on its record over the two most recent seasons, except for the two expansion teams, whose odds come from this past season alone.

A player must also be eligible. In 2024 the rule let a college player who would turn 22 that year give up her remaining college eligibility to enter, and seniors who had used theirs did not need to. Clark's college career ended on April 7, 2024, with a loss in the national final, but she had turned 22 on January 22, 2024. On February 29 she announced she would enter the draft rather than use her last season, and on April 15, 2024, the Fever, who held the first pick, chose her. ESPN said 2.45 million people watched, the most ever for a WNBA draft. The old mark, 601,000, was set in 2004.

The contract that follows is not negotiated pick by pick. A rookie scale sets a salary for each draft slot and each year. The scale for 2026 came from a seven-year agreement between the league and its players' union, announced on March 20, 2026, and running through 2032. The league's release projected $500,000 for the No. 1 pick that year, and a published table of the full scale, which CBS Sports credited to Front Office Sports, listed $466,913 for the No. 2 pick in year one. That table runs four years for a first-round pick, with the figure rising each year. A pick's number, not her college record, sets the starting salary, so the order of the draft, which the standings and the lottery decide, also sets the order of the first contracts.

Part Three · The audience

How is a television audience counted, such as the one for Clark's 2024 title game? Nielsen, whose estimates the networks quote, builds its numbers from two sources. A national panel of more than 42,000 homes and 100,000 people, with meters that capture each person's viewing, supplies who is watching. Cable and satellite set-top boxes and smart TVs in 45 million households supply what is on the screen, but only at the device level, so they cannot say who is in the room. Nielsen uses the panel as its source of truth to calibrate the device data and project it to the whole country. It calls the blend Big Data + Panel and, from September 1, 2025, asks networks to base press claims on it.

Then the definitions. A rating is the percentage of the television-owning population that watched. An average audience is the audience in an average minute, found by averaging minute one, minute two and so on, and a peak is the best single minute. Each can count live viewing only, live plus same day, or live plus days of playback. Each can include viewing away from home: Nielsen's out-of-home panel, which uses wearable meters in places such as bars, hotels and airports, covered 66 percent of the country's television population in 2024 and 100 percent from February 2025. A count of unique viewers, anyone who watched at least once, is another measure again.

Apply that to two title games. On April 7, 2024, Nielsen estimated 18.9 million for Iowa and South Carolina, live plus same day, with a peak of 24.1 million. On April 5, 2026, UCLA and South Carolina averaged 9.88 million on ABC, ESPN and ESPNU, and a 3.9 rating, according to Sports Media Watch, against 8.6 million and a 4.3 rating for the 2025 game. That site noted that the 2026 rating fell 9 percent while viewers rose 15 percent, and wrote that Nielsen's out-of-home expansion and its new method had probably lifted sports audiences across the board. It is one analyst's reading, not Nielsen's, but it points at a real problem: the 2024 and 2026 figures came from different measuring systems.

What the counts cannot show is cause. NCAA.com wrote in February 2024 that Clark's impact had amplified an upward trend already under way: the 2023 final, Iowa against LSU, drew 9.9 million, and college women's attendance set a record in 2022-23. The WNBA's report of October 14, 2026 counts 3.8 million fans over 330 games and more than 90 million unique viewers, which counts anyone who watched any game once, unlike the average audience of one game. Totals rise when teams and games are added, 12 teams in 2024 and 15 in 2026. The numbers show that attention grew while she played. They cannot divide it among a player, a league and a measuring system.

The last numbers of her 2026 season came in a first-round series against Las Vegas. In Game 2, on September 29, Clark scored 27 points with 15 assists in a 99-89 Fever win. In Game 3, on October 1, she scored 18 on 6-of-18 shooting, and Las Vegas won 94-83 to end the Fever's season. Across the three games of the series the Fever scored 85, 99 and 83 points, and they lost two of them. The same formulas apply to every one of those games, and they produce very different answers. Every figure here is through early November 2026. The next ones will come from a season that has not yet been played.