Esports Betting in the US: ROLR, Seth Young and the Not-There-Yet Problem, Read Through Data

Seven years, five years, and an unexplained gap

Seven years. That is how long Seth Young, a former professional CS2 player and now CEO of ROLR, says he has repeated the same assessment of the US esports betting market: it is not there yet. On the other side of the same picture, the High Roller product operated by his company recorded positive return on ad spend (ROAS) for five consecutive years, and all of those results came from markets he himself ranks as weaker than the United States.

The two data points sit side by side and produce a noticeable gap. When data speaks, the whole stadium goes quiet: the problem with US esports betting is most likely the distance between watching and wagering, not the demand for watching broadcasts. That distance is measurable, and like anything measurable, it can be mispriced.

Context: a platform standing between two regulatory systems

The source material for this analysis is a business interview with Seth Young. Across the entire text, no specific match, team or tournament appears. That detail has value of its own: ROLR is not selling the story of one game title, it is selling a mechanism for trading on the outcomes of many titles. In my tracking sheets, this is the product category most often misread, because it wears the shape of betting while operating on the logic of a financial market.

Structurally, the US market has three large groups. DraftKings, FanDuel and Fanatics run traditional sportsbooks under state licences. Kalshi runs event contracts under federal CFTC oversight. ROLR picks a position in between, where prediction markets sit under a framework distinct from ordinary sports betting. That is a sensible defensive choice: it avoids head-on competition with names holding far deeper pockets, while placing the company in a regulatory zone that still has plenty of grey area.

Another key piece is Spike Up Media, both a large shareholder and a lead generation partner. That relationship has had enough time to prove itself: five years of positive ROAS is not a small sample. As an analyst, I always separate two questions. Whether the number is real, and whether it can travel to the US market.

My method uses three metric groups. First, broadcast tracking metrics: audience size and viewer retention by event. Second, a proxy for trading volume per match, built from match counts, active users and the number of open markets. Third, user acquisition cost placed next to ROAS. These three groups cannot replace a company's internal data, but they are enough to test whether the narrative matches the structure of the market.

The evidence chain: watching does not automatically become trading

The first and clearest piece of evidence is the audience paradox. The CEO describes people packing an arena to watch a League of Legends match as an illustration of the scale of US esports. That scale is real. But when the same yardstick is applied to two different products, a ticket and a prediction contract, conversion rates diverge far enough that culture alone cannot explain it. Trading volume per esports match remains substantially lower than a single match in the major traditional sports leagues, even though online viewership is not inferior.

The second piece is spending discipline. ROLR describes its strategy as surgical: measured spend, tied to ROAS, without burning cash to buy share. The statement that it does not need the whole pie, only a fair slice of it, is a claim about cost structure rather than ambition. In an immature market that is a plus, because early-stage user acquisition is usually cheap. The condition attached is equally clear: it stays cheap only until someone bids higher.

The third piece is the historical data base. Five years of positive ROAS in markets weaker than the US suggests the unit economics do not depend on the market being large. This is the kind of evidence I rate highly, because it comes from operations rather than forecasts. Behind every shot that hits the crossbar are thousands of data points whispering that nobody has the patience to hear, and behind a positive ROAS figure are thousands of small spending decisions that were declined.

The fourth piece, less noticed, is the state-by-state legal barrier. US sports betting expands state by state with its own rulebooks; prediction markets sit under a different layer of federal oversight. The result is that a platform like ROLR must operate inside several legal frameworks at once, with a product catalogue that can be open in one state and closed in another. Compliance cost therefore becomes a growth variable rather than a fixed cost.

Finally, there is data quality. Prediction contracts require real-time feeds accurate enough to settle. In esports, where schedules shift, events run on multiple patch versions and fixtures sometimes sit in hard-to-track time zones, this is an infrastructure problem rather than a marketing one. I do not commentate football. I read football through charts. The same reading applies here: a trading product is only as good as the data pipeline running behind it.

The same metric set, two markets

Born in South Korea and working in the United States, I have the chance to place the same yardstick on two different markets. In South Korea, esports is institutionalised to the point where finals fill arenas and are broadcast like national events. But esports betting channels are essentially closed by regulation, so even with enormous viewing demand, the conversion rate into transactions stays low.

In the US, the picture is half reversed. The legal framework is more open in many states, but the cultural penetration of esports into sports consumption habits is thinner. Placing the two markets side by side, the conclusion is fairly clear: audience behaviour is measurable, and it is not governed by fan culture the way many people assume. It is governed by product design and the regulatory framework.

This is why I do not believe the explanation that Americans are not yet used to esports betting. The evidence shows they are used to watching. There simply has not been a product easy enough to turn watching into a transaction.

Absence is also data

Across the entire interview there is no league name, no team name, and no passage discussing event integrity. That absence is worth recording. A title-agnostic platform has an expansion advantage, but it also means data quality depends on third parties at every event it serves.

The larger risk lies in the dimension that goes unmentioned: match-fixing in esports. For a prediction market, trust in the integrity of results is the only asset that cannot be bought with an ad budget. A single exposed fixing scandal can collapse liquidity across an entire product line, and no ROAS model protects against that.

Limits of the data

Every conclusion above has a ceiling. I have no access to ROLR's active user counts, average contract value or actual user acquisition cost. The five-year positive ROAS figure is company-disclosed information, not independently audited data. The observation sample is limited to one company, so it cannot be generalised to the whole market. A large part of the US prediction market is closed to outside observers, which makes any cross-country comparison only relative. That is why I mark a confidence level for each inference instead of merging everything into a single conclusion.

The contrarian angle: positive ROAS in a weak market can be a sign of weakness

The comparison looks simple: if the model works in weak markets, it should work better in strong ones. I think this is precisely the biggest blind spot in the story.

A weak market usually means few rivals bidding for advertising. Acquisition costs are low, so positive ROAS is easy to reach even with a product that is not yet optimised. The US is a market where DraftKings, FanDuel and Fanatics have spent billions of dollars educating users and occupying mindshare. Entering it, user acquisition cost can rise by orders of magnitude while lifetime value may not rise in step. In other words, efficiency in a low-competition setting does not measure efficiency in a high-competition one. A correlation between positive ROAS and scalability is not the same as causation.

The second contrarian point concerns the length of those seven years itself. An assessment that was correct in 2026 can become an anchor by 2026. When a leader repeats the same sentence for seven years, the data cannot tell whether that is caution confirmed by events or an assumption never re-tested. For investors, those two possibilities lead to opposite conclusions.

The third blind spot sits in the partnership structure. Spike Up Media operates across multiple verticals, meaning its resources can shift elsewhere if US esports slows. That is a cushion against risk, but it is also a signal that the depth of commitment focused specifically on esports has limits.

Signals to track next quarter

Three variables will speak louder than any statement. First, user acquisition cost: if it rises more than 30% over a few quarters, assumptions about sustainable ROAS need rewriting. Second, licensing progress in large states, since every state opening prediction markets for esports widens the addressable customer base, but also attracts large competitors. Third, whether the incumbents genuinely step into the esports vertical; a DraftKings pilot carries far more weight than a strategy press release.

The most notable thing in this whole story is not the ROAS number, but a CEO volunteering that his own market is immature. In an industry where everyone wants to draw a bigger pie, caution is a rare data type. The task now is to check whether that caution comes with real scalability, or is simply another way of saying wait. The numbers will answer. The only question is who has the patience to keep reading until then.