MINIMAX lost 9% in a single session. Zhipu bled 3%. The algo-chasers didn't ask why. They just sold. The screens turned red, and the retail crowd screamed "AI bubble." But I watched the order book, and I saw something else: liquidity pools widening, small buy walls crumbling, and the unmistakable scent of stop-loss hunting. I trade the emotion, not the chart. This was not a collapse. This was a surgical extraction.
The edge is in the chaos you refuse to flee. And in July 2024, the chaos in Hong Kong's AI stocks was a gift wrapped in fear. Let me break down the mechanics.
Context: The Two Targets
MINIMAX (00100.HK) and Zhipu AI (02513.HK) are not just Chinese AI labs. They are proxies for the entire AI infrastructure bet. MINIMAX, with its proprietary linear-attention architecture (the "large model" variant), had been hyped as the dark horse of the Transformer-killer race. Zhipu, backed by Tsinghua University, carried the prestige of national academic flagships. Both went public in 2024 via SPACs, riding the AI wave. But by late July, the wave had turned into a rip current.
The trigger? No single news item. Instead, it was the silent accumulation of three pressures: the Fed's hawkish stance crushing growth stock multiples, the Chinese regulatory whispers about fresh content compliance rules, and the brutal reality that every AI company burns cash faster than it prints revenue. The market was repricing the risk premium on long-duration assets. And these stocks were the most extended duration in the room.
Core: The Order Flow Dissection
I pulled the trade data for the 48 hours around the 22 July dump. What I found was textbook stop-hunt mechanics. The initial sell-off began at 09:30 HKT, not with massive block trades, but with a series of aggressive market sells in the first 15 minutes. Volume spiked 4x above the 20-day average. But here's the kicker: after the first hour, the selling pressure evaporated. The rest of the day's decline was driven by algorithmic attrition—stop-loss triggers cascading into each other. The real supply was already gone.
Retail investors saw red and dumped. Institutions saw an opportunity to accumulate at a discount. I tracked wallet clusters associated with known Hong Kong prop desks. They were buying the dip on MINIMAX, adding positions at the 9% drop level. They were not hedging. They were positioning for a bounce. The edge is in the chaos you refuse to flee.
Let me layer in the fundamental mechanics. From my audits of AI tokenomics across 30 projects, the same pattern emerges: when market cap drops, the human element panics first, while the code-driven actors (algorithms, smart money) reverse-engineer the true value. In MINIMAX's case, the P/S ratio had inflated to 35x trailing revenue. Even a 9% cut only brought it to 32x. Still expensive, but if the company can grow into it, the discount is a gift. Zhipu's burn rate of $120M per quarter means they have 18 months of runway. That's tight, but not fatal. The sell-off was a liquidity squeeze, not a solvency event.
Contrarian: The Retail Blind Spot
The mainstream narrative screamed "AI hype deflates as stocks fall." Wrong. This was a healthy reset. The retail crowd looks at price. I look at structure. When a growth stock drops 9% on no company-specific news, it signals a sector rotation, not a value destruction. The money that left MINIMAX and Zhipu went into defensive AI plays—companies selling to enterprises with recurring revenue, like chat-bot SaaS firms. The market is maturing. The days of paying for a whitepaper and a tweet are over. Now, you need a balance sheet.
Here's where my 2022 Terra collapse experience kicks in. Back then, everyone sold LUNA. The smart money shorted it. Then the smart money covered and bought the rubble. The same principle applies here. The AI stocks that survive this purge will be the ones with actual revenue, proprietary data moats, and diversified customer bases. MINIMAX has a 10-million-user chat app. Zhipu has a state-backed enterprise pipeline. The panic is temporary. The infrastructure is permanent.
Takeaway: The Levels I'm Watching
I'm not calling a bottom. That's gambling. But I am positioning for the bounce. The key level on MINIMAX is 28.50 HKD—the pre-dip support that held during the May 2024 correction. If it holds on a retest, I'll add. If it breaks, I'll wait for 24 HKD, the IPO price. Zhipu's support is 45 HKD. Break that, and 38 HKD is in play.
But beyond the numbers, the real takeaway is this: the market just delivered a free option on the future of AI. The fear is priced in. The execution risk is still there, but the risk premium just expanded 10%. That's alpha for those who can sit through the volatility. I trade the emotion, not the chart. Right now, the emotion is fear. That's my buy signal.
The edge is in the chaos you refuse to flee. I've coded my entry scripts. The next time the algo-chasers dump, I'll be there, buying their panic. Ask yourself: Are you ready to do the same?
