Quant Trading Is Becoming Agentic

Talos Head of Quant Execution Services Eliad Hoch sat down with Toby Babb of Harrington Starr at Talos’s New York headquarters for an episode of FinTech Focus TV. They discussed how agentic AI is reshaping quant work, why Bitcoin’s price is the wrong health barometer for digital assets, the rise of prediction markets, and the shift to 24/7 markets.

Watch the interview

Highlights

Bitcoin is not the health barometer for digital assets

Eliad argued that judging digital assets by Bitcoin’s price misses the point. Real-world assets are moving onto digital rails in 2026, from tokenized stocks to pre-IPO perpetuals, gold, and weekend oil trading.

“We don’t think about the equities world as Apple price. We should not think about the digital world as BTC price.”

Prediction markets are now institutional

Institutional adoption of prediction markets (or event contracts) has moved faster than Eliad expected, drawing brokers, hedge funds and systematic traders. Their value is precision: a direct yes-or-no question to hedge or trade, instead of a proxy asset. Talos is integrated with Kalshi, with Polymarket coming next, and brings the full trade lifecycle to these markets: calibrated execution algos, RFQ, FCM settlement, and tick-by-tick market data.

Money should never sleep

Eliad sees no reason capital should sit idle between market close and open. The obstacles are regulation and back-end infrastructure, and Talos is working with DTCC on seamless on- and off-ramps. Adoption, he said, is the easy part, and institutions that were hesitant two years ago are now in.

AI shifts the focus from writing code to stating intent

Eliad’s global quant team went through a full transformation this year as models became capable enough to change how the team works day to day. The team now works agentic-first: less time on syntax, more on intent and on reviewing what the models produce.

“We moved so much from the nitty-gritty of the syntax of the coding into intent-based. … We’re never gonna go back.”

Agentic execution will include a human in the loop

The next phase is bringing those agentic experiences into the Talos platform. Eliad described a trader asking a Talos agent to work a large BTC perpetuals order across their usual venues over a set window. The agent would run scenarios across algos and return a few options with its reasoning. The trader picks one, and it executes. The same model extends to portfolio managers, quant researchers and operations teams.

“No one likes clicking buttons on a UI. … The vision could be one day that an entire shop … can be done with a thin layer of supervisors and agents.”

Young professionals should build

Eliad’s message to early-career quants is optimistic: build, publish and be able to explain your work.

“There are no excuses anymore. Do you have an interesting idea? Go and vibe code it.”

Reprinted from Harrington Starr.

^{Disclaimer: This communication is for informational purposes only and does not constitute an offer, solicitation, recommendation or investment advice regarding any digital asset, security, derivative, commodity, event contract, prediction market product or trading strategy. All images are for illustrative purposes only and do not represent actual data. Talos Global, Inc. and its affiliates ("Talos") make no representation as to the accuracy or completeness of this information, which is subject to change without notice, and accept no liability for its use. Event contracts referenced are financial instruments traded on regulated venues. They are not gambling, wagering or betting, and Talos does not offer, operate or facilitate gambling services. Reproduction or redistribution without Talos's prior written permission is prohibited.}