Seven Years, One Answer: The Conversion Gap in the U.S. Esports Betting Market

Seth Young played Counter-Strike 2 at a semi-professional level before taking an executive seat. In a recently published interview, he described a scene anyone who has walked into an esports arena recognises instantly: thousands of people queuing outside the gates, a packed house, a League of Legends match on the big screen, roaring flooding down from the stands. Then he added a line that made me read it three times. The esports betting market in the United States, where he is staking his career, is, in his own words, not there yet. And he first said the same thing roughly seven years ago.

Seven years is long enough for a young market to either mature or prove it has no intention of maturing. Over those seven years I have sat in Seoul, tracked the LCK match by match, logged the K League round by round, and cross-referenced what I saw against reports coming out of North America. The contrast bothers me in exactly the way data usually does: it does not deny anyone their enthusiasm, it simply puts that enthusiasm on a scale.

At a major North American final, attendance runs into the tens of thousands and concurrent viewership into the millions. At an ordinary LCK round in Seoul, both figures are far smaller. Chart the two side by side and the betting revenue curve does not follow the viewership curve. The gap between those two lines is what I want to dissect.

Three layers of concepts collapsed into one

Fans usually hear the phrase esports betting as a single block. In reality there are at least three distinct layers, and confusing them is the source of most bad forecasts.

The first layer is the traditional sportsbook, where players stake against fixed odds posted by the house. DraftKings, FanDuel and Fanatics sit here, operating under state gaming licences. The second is the prediction market, where participants buy and sell event contracts, price is set by supply and demand, and liquidity decides everything. Kalshi operates inside a framework of event contracts overseen by the Commodity Futures Trading Commission. The third layer is the data infrastructure behind both: real-time match feeds, latency, accuracy, and the ability to reconcile disputes.

ROLR chose the second layer. Seth Young states plainly that he is not trying to build a smaller DraftKings. That is a defensible strategic choice, but it also puts the company in the hardest possible seat: a prediction market only works when there are enough people on both sides of a contract, and that depends on an ecosystem the United States has not finished building.

The legal landmark worth remembering is May 14, 2026, when Murphy v. NCAA struck down the Professional and Amateur Sports Protection Act, opening the door for individual states to legalise sports betting. Seven years after that landmark, traditional sports betting has exploded in the United States. Esports betting has not. A CEO repeating the same sentence for seven years is not laziness in phrasing. It is a structural signal.

On the partner side, ROLR has tied itself to Spike Up Media, a lead generation firm that is also a major shareholder. This is not a one-off transaction but a long-term alignment: the partner handles user acquisition, ROLR handles product and market operations. The predecessor product, High Roller, accumulated five years of data with positive return on ad spend in markets the CEO himself rates as weaker than the United States.

Numbers do not lie, but they know when to stay silent

Five consecutive years of positive return on ad spend is a serious credential. It shows the company spends with discipline, measures properly, and does not burn money on artificial growth. But I want to separate two things the coverage tends to blend: evidence of operating capability, and evidence of market size.

A positive return on ad spend answers whether this company knows how to make money on every dollar spent. The answer is yes. It does not answer whether the pond is big enough to swim in. Those two questions need two different datasets, and the interview only supplies the first.

We do not predict the future, we only read probabilities already written. And the probability here is written by a single variable: the conversion rate from viewer to trader.

Picture it crudely. If a major event draws millions of concurrent viewers and the conversion rate into a prediction venue is one in a thousand, per-match liquidity is still enough to operate. If that rate falls to one in ten thousand, the order book thins until spreads widen, transaction costs rise, and users leave because the experience is bad rather than because they lost. This is the spiral every prediction market knows: liquidity begets liquidity, and emptiness begets emptiness.

What stands out is that Seth Young compares betting volume per esports match against major traditional leagues. The comparison sounds attractive but misleads easily. A professional basketball game has a fixed calendar, lineups announced in advance, and decades of historical data for bookmakers to price against. An esports match can have the value of every number rewritten by a single patch. In esports, a millisecond is a tactical hole, and for a prediction market, a patch is a pricing hole as well.

I have spent many seasons building models that rate players across multiple splits. The biggest lesson never came from the model but from the times the model was wrong. In 2026, when stadiums emptied, the home win rate in K League 1 fell from 47.2 percent to 38.5 percent. No team got weaker, no tactic went out of date. One environmental variable simply vanished, and every forecast built on home advantage became meaningless within weeks. The journey of data is the journey of humility.

Set that story beside the esports betting story and an uncomfortable parallel appears. The U.S. betting market does not lack viewers. It lacks a stable layer of variables to price against, something professional basketball has on hand and esports must build itself.

There is a detail financial analysts routinely overlook: in esports, the game publisher is an invisible referee with the power to change the rules mid-season. A single update can invalidate an entire tactical school, and for a prediction market it invalidates an entire pricing model along with it. The ability to adapt to a new version is often mistaken for raw strength; in truth it is a separate skill, and it never appears on a financial dashboard. Any prediction venue that builds a parameter layer that adjusts automatically with patches will hold a structural edge rivals cannot copy quickly.

Reading ROLR like reading a roster

An occupational habit: when there is no match to analyse, I still evaluate a business against the same criteria I use for a team. ROLR's paper strength rests on a predecessor product that has run for five years and a lead-generation partner holding equity. Role fit rests on the fact that Seth Young was once a professional player, meaning he understands the emotional cycle of viewers in a way a purely financial executive does not. Chemistry rests on long-term alignment with Spike Up Media rather than a short-term service contract. Bench depth, however, is thin: ROLR has limited contingency if the U.S. prediction market does not expand as hoped, beyond leaning on its partner's multi-vertical portfolio. There is no transfer deal to analyse here, but the logic is identical: strong in midfield, thin at the back.

The contrarian angle: the problem is not demand

Most analysts will read the line about the market not being there and conclude demand is insufficient. I think that conclusion misses half the picture.

Demand already has evidence: sold-out arenas, peak streaming numbers, grassroots communities running tournaments everywhere, including Vietnam, where data infrastructure is rough but player and viewer density is very high. Demand is not the missing variable. The missing variable is trust in the integrity of results.

Traditional sports bettors can tolerate refereeing error. They have a league, a disciplinary system, and a long history to trust that results are mostly real. In esports, that trust has not been built. A match can be thrown, a player can be banned, a tournament can change format at the last minute. Each such event does not affect only one match; it drains liquidity from the entire market for months.

This is where I part ways with most esports optimists. I believe the esports prediction market will mature, but along a far slower curve than traditional sports betting, and it will mature first where match data infrastructure is strongest, not where viewership is largest. The United States has viewers. South Korea has infrastructure. Vietnam has both in an untapped form. That misalignment across three factors explains most of the seven-year story.

And here is what would prove me wrong: if a major publisher, Riot or Valve for instance, released a standardised, low-latency, officially attested match data feed shared across all platforms, the cost of building trust would fall sharply. The barrier would then shift from integrity to speed of product distribution, and venues like ROLR would grow faster than I forecast. The falsification threshold is clear enough: average per-match liquidity growing above twenty percent quarter on quarter for at least three consecutive quarters.

Another hypothesis needs testing. ROLR's spending is described as surgical, focused on measurable return. For a young company that caution is a plus. But in a network market, caution sometimes means missing the window. Salary is the past, future value is what deserves to be paid, yet in this case the question is not how much to pay but when to pay. If the market booms in three years, the surgical spender of today will have to buy users at several times the price.

Signals to watch in the next cycle

Three markers will show whether the seven-year story stretches into ten. The first is how often large states such as New York, California and Florida introduce dedicated regulatory frameworks for esports betting, because each state that legalises adds a new liquidity layer. The second is user acquisition cost across prediction venues; if that figure rises more than thirty percent while revenue per user fails to follow, the surgical spending model is losing its edge. The third, and most important, is the arrival of a standardised match data feed from the publisher side.

Three major tournaments, one model, countless truths. In Seoul I am used to checking the stat sheet before the scoreline. In North America, the esports betting industry is now at the stage where it must learn that exact habit: read the data layer beneath the result before daring to price the result.

Seth Young says he wants his fair share of a large and growing pie rather than trying to swallow the whole thing. That is a far more credible statement than promises of market domination. But it also concedes something: the pie is still in the oven, and none of us knows the exact temperature.

What I am certain of is this. A market does not mature because people want it to, and it does not slow down because one CEO speaks honestly. It matures at the speed of the infrastructure beneath it. Over the past seven years that infrastructure has improved but not enough to change the market's state. If that changes, it will change fast enough that many people standing outside will not get a ticket in time. When the crowd goes quiet, the data speaks for itself, and right now that voice is saying: be patient, but do not take your eyes off the board.