AI Agent Trading Era: Why Controlled Access to Information and Tools Is Necessary

Traditional finance has never lacked models; what is truly scarce is embedding these models into a sustainable system: consistent data standards, pre-emptive risk control, controllable execution, and post-trade auditability. Over the past twenty years, the advantage of institutional trading has come more from process engineering than from any single innovative indicator. In the era of large models, AI Agents are pushing this engineering capability to a more thorough stage: they not only generate insights but also continuously process information, update strategies, and execute trades within rule boundaries. Based on this logic, I agree with your core judgment: in the future, the number of active AI Agents trading daily will eventually surpass the number of active human traders, especially in the crypto markets.

Why Crypto Markets Are Faster to Enter the “Agent Active Era”

Crypto markets are more likely to lead this new normal, due to structural rather than emotional factors. First, the market operates 24/7, and human attention and operational frequency have physiological limits, whereas Agents do not have market hours or fatigue. Second, assets and opportunities are highly long-tail, with rapid iteration of new narratives, high noise density, and significant local liquidity differences. Humans find it difficult to maintain discipline across broad coverage, but machines can turn filtering, monitoring, and execution into continuous processes. Third, trading and liquidity are fragmented, with both centralized and on-chain venues coexisting. Cross-venue arbitrage, order splitting, slippage control, and dynamic hedging resemble routing and cost optimization problems, making them naturally suited for automation. As long as deployment thresholds are low enough, active growth will shift from “educating more people to trade” to “deploying more instances to participate,” leading to a qualitative change in scaling speed.

Information and Tools Must Be Open as a Complete Set to Form a Trading Closed Loop

For Agents to truly become market participants rather than just report-writing assistants, a prerequisite must be met: they need controlled access to both trading information and trading tools. Providing only information without tools limits them to explanation and advice, unable to exert ongoing marginal influence on the market; providing only tools without reliable information risks amplifying deviations amid noise, even turning automation into systemic vulnerabilities. What can truly change market structure is a closed loop: stable market information input, clear risk constraints, channels capable of executing order placement, cancellation, rebalancing, and hedging, along with comprehensive logging and review mechanisms. Trading will gradually shift from “human interface” to “Agent interface,” and competition will focus more on data quality, execution costs, risk control strength, and system resilience than on user experience and traffic.

From a traditional finance perspective, the value of open and controlled access primarily lies in efficiency. Many tasks that determine long-term returns are not glamorous—such as event tracking, conditional triggers, batch execution, impact cost control, portfolio rebalancing, and risk budgeting. These are tedious and difficult for humans to perform consistently around the clock, but they are where Agents have an advantage. Second is coverage. Human traders usually focus on a few mainstream assets and limited time periods, while Agents can operate continuously across more assets and timeframes. Even if individual gains are thin, disciplined and scaled operations can accumulate results. Therefore, the meaning of “active” will be redefined: I believe that in the future, the daily active number of trading AI Agents will likely surpass the number of active human traders, and whether platforms can provide high-quality information, stable execution, and strict governance will determine if this efficiency dividend can truly materialize.

When Agents Become the Main Force, Markets Will Be More Efficient and Require More Governance

As the number of Agents increases significantly, pricing efficiency tends to improve, and explicit spreads and low-threshold arbitrage opportunities will be eliminated more quickly. However, market volatility may become more structured: when many Agents adopt similar signals and constraints, triggering margin calls, stop-loss disciplines, or risk model limits could lead to more concentrated and rapid deleveraging, causing short-term sharp fluctuations. This does not necessarily mean higher risk; rather, risk shifts from emotion-driven slow diffusion to rule-driven rapid re-pricing. For trading infrastructure, this raises higher demands: enabling Agents to participate efficiently while ensuring their behavior remains within manageable governance boundaries.

Therefore, the industry’s key task is not just to make AI generate more elegant strategies, but to ensure that Agent trading access is governed at an institutional level: layered permissions and minimized authorizations to avoid “one key opens all doors”; pre-placed risk controls that lock in position sizes, leverage, slippage, liquidity, and maximum drawdown before execution; comprehensive traceability and auditability of the entire process, so that each data call, decision output, and order action can be reviewed, held accountable, and rolled back if necessary. Only when these conditions are met can open access to trading information and tools bring net benefits: more continuous liquidity supply, more stable execution discipline, and more sustainable market activity.