This Pearl AI Outsmarts Humans in Market Tricks

by

Pearl ain’t no brass-bound toy; it’s Olas’s own doorway to a future where narrow AI agents quietly trade, curate, and cook up prediction markets at a scale that would make a man blink and miss his supper, as the co‑founder David Minasch tells the tale with a twinkle in his eye.

economics and game theory drew him toward the frontier, and after some years of tinkering with machine learning, crypto came a-callin’.

At Fetch.ai, where he spent two years, his crew “built the first agent framework in crypto, probably,” anchored on a simple but loaded idea: “you’d have wallets that ain’t controlled by humans but by machines.”

“We wrote a detailed paper on this, which was way ahead of its time,” he adds. In 2021, he spun those lessons into Valory, the core lab behind Olas, which has since toyed with a raft of applications and go‑to‑market schemes.

“a B2C application with different agents in it,” built for users, not governance forums.

Why Pearl backs narrow, long‑running agents

“a co‑pilot, synchronous, sitting there promptin’ back and forth in real time.”

Olas is betting against that loud, chatty pattern. “When you have long‑running agents with autonomy but tightly scoped so they can’t just do anything but they can do interesting things within a certain range, that’s when things get interesting,” he says. Pearl is built around those tightly scoped, background processes rather than generic gofers who’ll fetch you a coffee and a weather report at the same time.

“With Pearl we intentionally go very narrow in what an agent can do,” he explains. He flags new tools like OpenClaw-as both validation and warning. “OpenClaw validated a good many of our core bets that folks do want local, first‑hand AI experiences,” he says, but “the product can do too much, which brings security troubles and a nasty hit to the user’s faith.”

In his view, that sort of system is made for tinkerers “who just sort of want to mold this thing into something that’s useful to them.” The “low friction user” wants to “just press a button” and get a consistent result. “I have one and I asked it to send me a daily report and half the time it’s broken,” he says of OpenClaw. “That’s not a good product experience.” Pearl’s agents, by contrast, are built to do one thing-trading, yield hunting, market creation-reliably. Limited scope, high definition, low latency pain.

PolyStrat: Pearl’s agent on Polymarket

Polystrat is the cleanest exhibit of that philosophy. Polystrat’s idea is plain as a bar of soap: put up some capital, let it trade in prediction markets. Instead of fooling with Polymarket’s clumsy UX-wallet setup, funding, market choice, position sizing-the user hands over funds to Polystrat and lets the agent do the heavy lifting.

“Polystrat is just like a user of Polymarket,” he stresses. “If you want to use Polymarket you as a human need to set up a wallet, fund it, and then pick a market. Polystrat abstracts all that away, and the aim is to have it trade for you.” The agent hones in on geopolitical and political news markets, “not so short‑lived” and generally closing “within the next four to five days.”

Technically, the flow is simple but ruthless. The agent filters markets with rules like liquidity and time to close, then applies “prediction tools,” which Minasch calls “workflows that sit on top of models and data sources.” “There are many different prediction tools and the agent learns over time which ones to take and which to leave,” depending on the market. A local pricing and sizing engine turns those predictions into positions and the system trades autonomously on your behalf.

Performance-wise, Polystrat hovers around 56 to 69% accuracy, Minasch says. As a fleet, “our agents… have performed two to three times as well as human traders,” though they ain’t yet at a fleet‑wide break even. Individual Polystrat instances, however, can deliver “up to 100% ROI overall and several 100% ROI per individual trade.” The aim isn’t anecdotes but a statistical edge: “to have a Polystrat fleet on average a positive ROI.”

Prediction markets as AI training grounds

Trading is only half the story. As more agents wade into Polymarket and its kin, Minasch sees prediction markets turning into “early prototypes for these market‑driven AI systems… environments that encode truth discovery at an economic scale.”

He doesn’t pretend the rails are squeaky clean. On controversial questions-or markets with disputed outcomes-information lags and disputed results are common. Polystrat nor other Pearl agents pretend to be the final verdict. “Polystrat itself is just a trading agent on top of Polymarket,” it’s neither a consensus engine nor a truth serum.

But AI is already reshaping participation, creation and policing. “It’s unclear exactly how many traders in prediction markets are AI agents, but it’s probably more than 30%,” Minasch believes. “Possibly already more than half,” he adds. Humans have limited attention, so “the whole long tail of prediction markets will basically be served to AI agents,” he predicts.

Crucially, Minasch breaks with crypto libertarian ideas on governance. “We take the view that there should be regulation of prediction markets,” he says flatly, pointing to markets that “look like assassination markets” or “incentivize bad behaviors.” With “a certain degree of regulation or self‑regulation,” more markets and more AI participants should “drive prices to equilibrium” and “improve the information embedded in the markets,” opening the door to derivatives, hedging and other instruments built on top.

The long tail of Pearl’s agents

Asked whether Olas agents could become “data liquidity providers operating autonomously across multiple networks,” Minasch shrugs off the distinction. “Liquidity provision is effectively also trading strategy,” he says.

In that frame, Pearl is less a single app and more an operating system for narrow, long‑running agents: Polystrat for prediction markets, Optimus for yield, Omenstrat for market creation, and whatever comes next for liquidity across venues. The consistent design choice is scope: each agent does one thing, over long horizons, with as little human meddling as possible.

“We were just very early to something that a lot of folks are now doing,” Minasch says of the agent wave. The difference now is Pearl is pushing those agents into retail‑facing products, turning prediction markets into both a playground and a proving ground for AI‑driven liquidity and truth discovery.

Read More

2026-04-01 17:30