Kalshi early employees: Whoever controls the traffic controls the market

Author: Adhi Rajaprabhakaran

Compiled by: Jiahua, ChainCatcher

On the evening of February 12, in an exchange that usually has a muted reaction to sports events, three NBA games suddenly ignited trading: the Dallas Mavericks vs. the Los Angeles Lakers, the Milwaukee Bucks vs. the Oklahoma City Thunder, and the Portland Trail Blazers vs. the Utah Jazz. During these three games, they generated over 13 million contracts in trading volume. ForecastEx is a prediction market operated by Interactive Brokers and regulated by the CFTC. It is a real exchange with a formal license, but prior to that night, it had never seen any substantial NBA trading volume.

I don't believe ForecastEx created any miraculous customer acquisition overnight. It did not improve its product, launch marketing campaigns, or deepen its order book with more liquidity. What happened was actually quite simple: Robinhood directed its massive order flow to another exchange specifically for that night consisting of three NBA games.

Currently, Robinhood is the dominant retail distributor of prediction market contracts. When users open the Robinhood app, click on an NBA game, and place a bet, the trade is allocated to an exchange regulated by the CFTC for execution. For most of the history of Robinhood's prediction market, that exchange has been Kalshi. But users do not know this, and they do not care. Regardless of which exchange is in the background, the interface is exactly the same: the same app, the same buttons, the same odds. The exchange has become an invisible infrastructure.

A Momentary Migration of 35% of Trading Volume

Each bar chart represents a day's NBA trading volume, stacked by exchange. Blue represents Kalshi, and red represents ForecastEx. Except for February 12, every day is all blue, and on that day, 35% of the trading volume suddenly appeared on ForecastEx. Then everything returned to an all-blue state, as if nothing had happened.

The red portion on February 12 represents those three games: Mavericks vs. Lakers, Bucks vs. Thunder, and Blazers vs. Jazz. Together, they generated 13.4 million contracts on ForecastEx. Regardless of which exchange handled the trades, the user experience on Robinhood remains the same: the same app, the same buttons, the same odds. Users cannot distinguish the difference at all. Because for them, it truly makes no difference.

This is why the number 35% is so important, as it is a relatively pure measure of Robinhood's market share of NBA spread trading volume between the two exchanges. ForecastEx essentially has no naturally accumulated sports users, so it is reasonable to assume that every contract on ForecastEx that night came from Robinhood's orders.

Moreover, since Robinhood's interface is the same in any case, these users placed bets at the same frequency as they would on Kalshi. There is reason to infer that about one-third of Kalshi's NBA spread trading volume in February came from Robinhood.

Robinhood controls the direction of trading volume, and it can flip this switch overnight.

A Similar Story in the Weather Market

The order flow of the NBA was brief and dramatic, constituting an extremely clear and compelling natural experiment for analysis. However, the rise of the weather market on ForecastEx tells a similar story on a different scale.

Both ForecastEx and Kalshi offer daily high-temperature contracts: binary options on whether the maximum temperature in a given city will exceed a specified threshold on that day. The two markets are the same product, containing the same cities and the same dates. The only real difference is the exchange that matches the trades.

Before November 18, 2025, trading activity in ForecastEx's weather market was zero. Then, trading volume exploded overnight, with no spontaneous growth transition period and no gradual adoption curve. This step function pattern is entirely consistent with the NBA's characteristics. To measure the overlap, I matched markets on ForecastEx and Kalshi with the same "city-date" pairs, excluding cities that only existed on one exchange. This resulted in 454 matched "city-date" data points.

By the way, this chart provides an interesting case illustrating that platform competition is a net positive for overall industry trading volume. Robinhood opened the weather market faucet, overall increasing the activity of both exchanges, likely due to cross-exchange arbitrage. Market makers participating in such activities effectively distributed liquidity across the entire ecosystem.

For the first five weeks, there was only Kalshi, which serves as the baseline. Then ForecastEx appeared and immediately captured 60% of the total daily temperature market trading volume. It peaked at 72% in late November and has since maintained between 53% and 67%.

The key detail is that when ForecastEx emerged, Kalshi's weather trading volume did not collapse. The blue bars remained roughly stable. Therefore, my interpretation is that ForecastEx's trading volume is layered on top of Kalshi's existing flow. It is likely that Robinhood first opened the weather market and from the start directed its flow to ForecastEx, while its users remained unaware.

This distinction is important. In the January NBA case, Robinhood briefly shifted trading volume that would have gone to Kalshi. In the weather market, however, Robinhood seems to have added ForecastEx as a parallel destination while keeping Kalshi's original flow intact. Both scenarios demonstrate the same structural point: Robinhood decides where the trading volume goes. Exchanges can only passively receive the orders that Robinhood chooses to send.

The Absolute Amplification of Distribution Channels on Product Innovation

The data from the NBA and weather markets indicate that Robinhood can direct flow. The parlay (which refers to binding two or more independent bets into a single bet, where the player can only win the prize if all bound outcomes are predicted correctly; if one is wrong, the entire bet loses) suggests it can scale up the already growing demand.

Kalshi launched multi-variable event contracts (i.e., "combinations" or "parlays") in September 2025, coinciding with the start of the NFL season. The product immediately gained attention: weekly trading volume grew from nearly zero in September to about 45 million contracts by early December. This growth was self-driven and directly pointed to Kalshi's platform. Kalshi built the product, submitted for CFTC approval, and injected initial liquidity. The market responded positively.

Then, Robinhood intervened.

On December 17, Robinhood announced it would launch preset parlays and player prop betting in its app. Within weeks, weekly trading volume exploded, jumping from the range of 45 million to 60 million to nearly 100 million, and then reaching 300 million per week by late January. The shaded area on the right marks the period after the Super Bowl when NFL parlays disappeared, supported solely by the NBA. Even without football, trading volume remained around 260 million to 290 million per week.

Kalshi did the hard work of creating a new product category. Robinhood's distribution channel elevated it to a completely different scale. Both contributions are real. The question is, which one has greater structural leverage.

Not Just Kalshi

Kalshi has achieved tremendous growth over the past year, increasing from about 7 million contracts per day at the end of 2024 to over 100 million by the end of 2025. This is not entirely due to Robinhood. Kalshi has built real direct demand: new product categories, an expanding native user base, API traders, and institutional participation. A year ago, it was widely believed that Robinhood accounted for the vast majority of Kalshi's trading volume. Today, NBA data indicates that Robinhood accounts for about 35% of spread trading volume. This de-risked business execution capability is indeed admirable.

However, Kalshi is not the only exchange that has built its growth story on distribution channels.

Nadex, as a CFTC-regulated exchange operated by Crypto.com Derivatives, tells a remarkably similar story. Before Underdog integrated with Crypto.com in September 2025, Nadex's trading volume was unremarkable. After Underdog intervened and began directing its users' sports betting flow to the exchange, weekly trading volume exploded by orders of magnitude. The same pattern, different names. Underdog to Nadex is like Robinhood to Kalshi: the distribution layer that turned a quiet exchange into a busy hub.

The most astonishing part is that both of these distribution giants have now taken action and fully owned their exchanges. Robinhood acquired its own CFTC-regulated exchange, while Underdog did the same last week. Two companies, on parallel tracks, independently reached the same conclusion.

This is no coincidence. This is game theory. If you are a distributor directing millions of trades to a third-party exchange, you have to share profits on every contract with an infrastructure that your users cannot distinguish from a white-label API. You are also handing over data, trading volume, and regulatory records to potential competitors, which are the elements that make their exchanges valuable. When you scale large enough, the rational move is to internalize that infrastructure. The exchange shifts from someone else's profit center to your cost center.

The data from the weather and NBA explain why it is so difficult to defend against this dynamic from the exchange's perspective. Even accounting for only 35% of trading volume, Robinhood can overnight add a parallel exchange for the weather market and immediately send most of the new flow to it. It can direct three NBA games to another exchange on a Tuesday night, generating the same trading volume as anywhere else. Users remain oblivious. They do not choose the exchange. They choose Robinhood or Underdog.

I Was Wrong

Last year, when rumors circulated that Robinhood was considering acquiring its own CFTC-regulated exchange, I publicly stated that it was impossible.

I was wrong with such confidence for two reasons.

First, from my experience at Kalshi, I have a firsthand understanding of how extremely difficult it is to establish and operate a regulated derivatives exchange: compliance infrastructure, monitoring systems, CFTC reporting, and so on. Robinhood made a fortune from the prediction market while doing only about 1% of the work. The exchange did the heavy lifting, while Robinhood collected distribution fees, which is one of the most perfect partnerships in fintech over the years! Why break such a good situation?

Second, I applied traditional thinking from the past fifty years of derivatives market structure. Brokers do not acquire exchanges. In the world I am familiar with, the entire significance of an exchange lies in its irreplaceable trading pipeline. The Chicago Mercantile Exchange (CME) is a $90 billion company with net profit margins second only to Visa and Mastercard, precisely because "liquidity depth" is its impenetrable moat.

An institutional trader needing to mobilize a $50 million Brent crude position would be extremely concerned about order book depth, slippage, and counterparty concentration. This depth is very difficult to establish and nearly impossible to replicate, especially in a derivatives market where contracts cannot be swapped across exchanges. In that world, exchanges earned their structural position through their own strength. Brokers are commodities that can be replaced at any time.

The prediction market has disrupted this. On Robinhood, the average sports bet is just some ordinary user clicking a button to bet $10 on the Lakers. That user does not care about order book depth. Heck, they don't even know what an order book is. When the trade size is very small and the users are not professional, liquidity depth is no longer a moat. Robinhood switched the underlying pipeline on a Tuesday night, but the same trading volume still flowed out the other end.

When the trade size is very small and the users are not professional, liquidity depth is no longer a moat.

I was wrong because I was still navigating with an old map. The structural leverage of the prediction market is not found where the past fifty years of derivatives history has pointed. It is firmly in the hands of those who ultimately own the users.

In fact, I have written an honestly somewhat ungracious article about how ForecastEx messed up sports events. This may have resonated… and there was a tiny bit of activity on ForecastEx on February 5 that I cannot explain. This could be an early test by Robinhood. It is also possible that Robinhood is distributing flow among multiple exchanges, but external analysts have no way of knowing. I think this example is still up for debate, as Kalshi's RFQ (request for quote) system and large market-making team are indeed difficult to replicate. There is an extremely deep technical moat there. Furthermore, 'how important liquidity is in prediction markets' is still inconclusive. This inevitably makes me wonder: under the evolution of game theory, are we heading towards a homogeneous competitive endgame—where all exchanges fall into a quagmire of mutual imitation, racing to launch every market available on the market?