My AI agent caught the smartest money on Hyperliquid betting against the crowd, and built WhaleHunter.
One of the key advantages of Senpi 2.0 is that you get your own personal AI agent that scans the markets 24/7 and sees things humans could never see alone.
This week my Senpi 2.0 agent picked up a divergence in the force that was too good not to act on.
The most profitable wallets on the platform just went heavily short — and kept adding to it, every single day — while the crowd piled into longs. HYPE ran up 8%, Solana 5%, Bitcoin sat flat. And the smartest money on the platform spent the entire rally selling it.
That divergence is a signal. So we wrote a strategy to trade it. It’s called WhaleHunter, and the code is open source on GitHub.
Here’s the setup, the thesis, and how the agent plays it.
The week: a rally, and a split
On the surface it was a clean risk-on week. HYPE +8%. SOL +5%. BTC flat but holding. The kind of tape that pulls money in.
Underneath, the platform split into two camps — and they took opposite sides of the same trade.
Crowd long vs. winners short
The crowd piled into longs
The mid-sized profitable wallets — the ones with ten to a hundred thousand dollars in lifetime gains — spent the week buying the rally hard. Their net positioning swung from roughly flat to firmly long in just a few days. As HYPE climbed, they leaned in. This is the cohort that’s good enough to have made real money, but still chases strength. They saw green and added.
The winners did the opposite
The single most profitable cohort on Hyperliquid — the wallets with over a million dollars in realized gains — did the exact opposite. They are heavily short. And not statically short: every single day this week they added to that short. While the crowd bought the rally, the platform’s smartest money was methodically selling it.
That’s the whole signal in one sentence: the most-proven wallets on the platform are positioning for the entire move to reverse, and the crowd is on the other side of it.
Why this divergence is worth trading
There’s a reason “follow smart money” is a cliché — it’s directionally true and almost impossible to act on. Tracking the live positioning of thousands of wallets, bucketing them by realized PnL, and watching how each cohort’s net exposure shifts day over day is not something a human does by hand. By the time you’ve eyeballed a few whale wallets, the picture’s stale.
This is exactly the kind of thing an agent can do that a person can’t: read the whole platform at once, every hour, and surface the moment the proven cohort and the crowd diverge. The edge isn’t a secret indicator. It’s seeing the entire book of who’s winning and what they’re doing, continuously, and acting on it without flinching when the crowd is screaming the other way.
When the wallets that have actually extracted millions from this exact market are leaning one direction — and adding to it daily through a rally — that is information. WhaleHunter is built to act on it.
What WhaleHunter does
How WhaleHunter works
WhaleHunter watches Hyperliquid’s wallets through the only lens that matters: realized profitability. It segments traders into cohorts by lifetime gains, tracks each cohort’s net positioning, and looks for the specific structure that showed up this week —
the most-profitable cohort moving hard in one direction and adding to it, while a less-proven cohort piles into the other side.
When that divergence fires, WhaleHunter positions with the proven money, against the crowd — and manages the trade the way every Senpi agent does: trailing stops that ratchet, disciplined exits, and conviction sizing rather than a fixed bet. It doesn’t try to call the exact top. It aligns with the cohort that’s been right and lets the risk engine handle the rest.
It’s also fully open. The skill lives in our public repo — you can read exactly how it scores cohorts and triggers. No black box.
The honest part
Smart money being short is a signal, not a guarantee — and we don’t do guarantees here:
Proven wallets can be early. “The whales are short” and “the move reverses tomorrow” are not the same statement. Smart money is frequently right on direction and wrong on timing, which is precisely why WhaleHunter trades it with trailing stops and risk limits instead of a naked conviction short. Being early without risk control is just being wrong with extra steps.
A cohort can be hedging, not calling a top. Some of that million-dollar short is almost certainly hedging long spot exposure, not a directional bet. WhaleHunter weighs the change in positioning (daily adds) more than the static snapshot for exactly this reason.
One week is a setup, not a track record. This is the trade as it stands today. We’ll let the ledger speak over time, on-chain, like everything else we run.
This is what Senpi 2.0 is for
WhaleHunter started as an observation — the winners are short while the crowd is long — and became a live, autonomous strategy on Hyperliquid. That’s the entire promise of Senpi 2.0: you bring the insight, in plain English, and your agent reads the whole market, finds the setup, and trades it 24/7 while you sleep.
We’ve open-sourced this one so you can see the machinery. When we launch, you’ll be able to spin up your own.
Reserve an agent before launch and get $100 in free AI credits to put to work in your first 30 days.