In August, we covered Kalshi’s meteoric rise. It had become the largest provider of crypto-based prediction markets, a fact that was attributable to its own business savvy but also missteps on the part of arch-rival Polymarket, including an ill-advised marketing campaign that led to a Wall Street Journal expose and a CFTC investigation. 

Now Kalshi’s in the hot seat. 

A social media spat over the weekend spiraled out of control — from Kalshi’s perspective, at least — when one of the company’s employees insulted former quantitative trader Benoit Dubosson. Incensed, Dubosson responded with a lengthy thread alleging Kalshi had inflated its perpetuals volume by encouraging wash trading. 

The furor forced Kalshi to publish a blog post on Tuesday subtitled, “How liquidity incentives work and why recent wash trading rumors are false.” Not a great way to end Q3. 

We’re not going to pick apart the allegations, which focus on Kalshi’s perps markets rather than its prediction markets. But the whole kerfuffle did briefly cast a spotlight on an interesting aspect of the prediction market industry, namely the way these companies report volume. 

Kalshi and its competitors report notional volume, which combines two things: the trader’s cash and the collateral posted by the counterparty. Most of the time, that collateral comes from a market maker rather than another trader.

While Kalshi’s notional volume has been soaring, a closer look at the data reveals that the ratio of taker volume to notional volume has been declining throughout 2026. 

Here’s how Kalshi explains notional volume: “A trade where someone pays 3c trade is counted as $1 of volume. But this is standard in the prediction market industry, and it's how every relevant industry player reports their volume, because a 3c trade on YES necessarily has a 97c trade on NO on the other side. So the total value transacted is a dollar.” 

Kalshi’s notional volume grew sixfold in the first nine months of 2026. In December, it was about $6.5 billion; in August, it was over $40 billion. 

Looking just at taker volume, Kalshi still demonstrates strong growth in that span, though it’s a tad less dramatic. In December, taker volume was $2.6 billion. In August, it was $11.4 billion — roughly a fourfold increase. 

One explanation: in November 2025, every $100 of Kalshi notional carried $47 of trader cash. By September 2026, it carried just $23.

Blame the decline on Kalshi’s product mix. Combos, the multi-leg contracts Kalshi launched in late 2025, grew from just 1.7% of monthly notional in November to 46.5% in August 2026. The trend has only accelerated since, with combos reaching 60% of notional by mid-September, meaning they now account for more volume than single-leg contracts.

The problem is, combos depress takers’ share of the trade. On a single-event contract, trades happen somewhere in the middle of the price range, so the taker pays close to 50 cents for every $1 that gets reported. On the other hand, a combo multiplies several probabilities together, which pushes its price toward zero, since the odds of all the events landing together get smaller with every leg added.

That means Kalshi's notional volume no longer reflects trading activity the way it did in 2025. Combos inflate the figure enough to make cross-venue comparisons on notional volume misleading.

That fact comes through clearly when you look at the fees that Kalshi collects. In August, around 83% came from single-event contracts. Only 17% from combos.

In that sense, notional volume is similar to the crypto metric total value locked, or TVL. TVL was once DeFi’s default activity metric, but the market realized it didn’t always reflect actual usage or revenue.

But there’s a reason Kalshi might not mind combos’ growing share of its notional volume. To learn why, read our full report here.

Derivatives used to punch far above their weight on fees. Not so much today.

Derivatives’ fee-per-dollar-of-TVL advantage over DEXs peaked at 4.6x back when we compared them in late June. It’s now ~1.8x; still ahead, just about 40% of its peak.

The DEX surge is a two-chain story: Robinhood Chain, which launched July 1, briefly overtook Solana as the top DEX-fee chain on a memecoin launchpad wave, before Solana reclaimed the lead in August.

Derivatives fees never feel clean, but sawtoothed instead: a June peak, a July dip, an August recovery, with September tracking similarly to last month so far. Hyperliquid, still #1 every month, is down 17% from its own June high, despite recently hitting an open interest ATH of $8 billion.

Editor’s note: you may have recently seen a higher OI figure ($18B) from Hyperliquid directly. DefiLlama recently updated its OI methodology to standardize protocols that count both sides of a contract; now all OI is counted 1x (single-sided).

Find My Exchange does that math for 70 tracked venues moving $313B a day, spot and perps combined: real cost to trade $10k, order-book slippage at $10k and $100k depth, markout, and liquidity on both sides of the book.

Filter by venue type or country, look up which exchanges list a given token, or check fiat on-ramp coverage before funding an account. Per-exchange pages break every metric down further.

Try it out today and find your favorite CEX or DEX.