Crypto Crashes & World Cup Hysteria

Crypto prices fell throughout the week, falling in tandem with equities as markets reacted to the latest AI-driven shock. The sell-off left BTC down 6% for the week by Friday, while ETH was near 10% lower. HYPE was the notable outlier, highly volatile intraday but broadly resilient compared to the rest of the crypto market. The S&P held up comparatively well, with equities recovering most of Wednesday's losses before Thursday hit.

While prices fell around New York morning each day, the dynamics of the sell-offs were slightly different. Wednesday’s crash was a slow grind lower, that extended into the afternoon, with peak selling around 4pm UTC. The next day was more dramatic.

Thursday saw BTC prices plunge more than 5%, driven by heavy concentrated volumes. In a single five minute window at 1:50 UTC there was $83mn BTC sold on Binance, with over $483M in spot volume on the venue between 1pm and 4pm.

The shock also showed up in spreads. While widening was venue-dependent and Thursday was the sharper episode, there were spikes both days across the market. Binance held tight across the board, with p95 barely registering above 0bps even as price fell 5% in 15 minutes.

Coinbase and Bitstamp told a different story, as both venues spiked simultaneously at the around 13:30 UTC onset, Bitstamp hitting nearly 7bps at the peak and Coinbase close behind at 4-5bps, staying elevated well past 15:00 UTC long after price had stabilised. On Wednesday the pattern was similar but noisier, with Bitstamp already running a wide tail before the crash began. The thinner the book, the longer makers stay back.

As BTC fell on Thursday, Deribit's options market largely stopped functioning, highlighting poor hedging functionality and known issues with pricing risk in crypto. Of the 912 BTC options LO:TECH tracks on Deribit, only 8.1% had clean two-sided quotes at the bottom, in other words around 92% of the market was effectively untradeable at the exact moment hedging demand was highest.

The same measure during Wednesday's slower grind bottomed out around 13%. The difference is likely the speed of the shock, Thursday’s more dramatic move probably raised more alarms in systems. It’s important to note this isn’t an issue with market participants, it’s their systems working well. However, it comes at the cost of efficient options markets.

The implied volatility surface told the same story from a different angle. Pre-crash, ATM IV was compressed and flat across the term structure at around 41%. At the bottom of the drop, for options expiring on Friday (0.7 days to expiry), ATM IV exploded to 80.5% while the 91 DTE barely moved. The term structure inverted violently at the front end.

At 10–14 DTE, the SVI calibration hit a hard boundary constraint, meaning there weren't enough clean quotes to build a surface at those tenors at all. Within an hour the short end had pulled back to around 66% and dealers had returned.

Kalshi notched its highest ever monthly volume figure ($17bn so far in June) and crossed $100bn in cumulative volume in the process. Polymarket didn’t set any records, but its volumes still rose around 17% month-on-month from May to June. And there’s still a few days to go, and big matches to come.

The bulk of that volume has been driven by sports, accounting for 70% of volume, and the bulk of that is football (or soccer). The FIFA World Cup has been a game changer for prediction markets.

It’s not all about superstars and marquee matches either. The most traded game so far was Germany against Côte d'Ivoire, not exactly box office. And none of the other high volume games were packed with superstars or tournament favourites. For instance, none of Argentina’s or France’s group-stage matches attracted more than $60 million in volume.

This was evident in South Africa’s game against South Korea. When South Africa scored in the 64th minute, Kalshi's feed took roughly 4-5 seconds to register the first informed trade. Once it did, 3,480 trades repriced the market in a staircase over several seconds. This likely tells us there was no single dominant participant, no clean arb.

In a more efficient market we see this sort of lag isn’t there to be exploited. Take Australia vs. Turkey, where the Socceroos were the heavy underdogs, the first goal led to match probabilities repricing in under a second. This is also partially a function of a less liquid market, but not entirely.

Draws are another interesting one, they’re consistently underpriced. 14/54 group stage matches have ended in draws, 26% actual vs 21% average implied probability pre-game. That's a persistent gap, not noise. The most natural explanation is who's in these markets.

A draw is the outcome nobody wants to watch, and prediction market participants are overwhelmingly spectator/retail first. They're betting on entertainment, so a goalless 90 minutes doesn't fit the narrative, so the market systematically underprices it. That's not a market inefficiency in the traditional sense. It's the composition of the crowd showing up in the price, and you don’t need to be fast to exploit it — just boring!

As we head into the knock-out stages of the World Cup there’s bound to be even more trading activity worth following. The final group games on Friday and Saturday will decide if Scotland make the knockouts, and who teams draw in the round of 32 will have a big say in the outright winners odds, which currently looks like this.

Beyond the soccer, we’ll be monitoring volatility. With the BTC front-end term structure showing extreme sensitivity this week, any renewed market turbulence could quickly drain liquidity again. We will be tracking the 10-14 DTE tenor closely to see if SVI calibration stabilises.

Elsewhere, Rodrigo from Boole Digital joined Tim to discuss how AI is reshaping the future of trading, his journey from investment banking to crypto entrepreneurship, and what it takes to build an AI-powered trading platform.

They explored Rodrigo's path from M&A in New York and Hong Kong to founding Boole Digital, how AI is compressing years of financial work into minutes, and why intelligent agents could become the future interface for trading.

The conversation covered founder burnout, bootstrapping through the bear market, building for crypto-native traders, and why AI could fundamentally change how we manage money.