Nansen CEO Bets on AI Agents to Overtake Human Traders Within 2 Years

Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.

When Alex Svanevik, CEO of on-chain analytics platform Nansen, publicly predicts that AI agents will outperform human traders within two years, the statement lands differently coming from someone who has spent years building the infrastructure to track blockchain activity. Svanevik's bet is not abstract theory—it is a strategic pivot for Nansen itself, which is reportedly shifting from pure analytics to what the firm calls "agentic trading." For our team at Broker Tested Reviews, this represents a significant inflection point in the AI trading bot sub-niche, where the line between data provider and automated execution system continues to blur. We have spent the 2020-2026 testing cycle running funded-account evaluations on over 50 platforms, and we have benchmarked several against Zephyr AI Trading Bot adaptive engine to understand exactly what a human trader gives up—and gains—when handing the reins to an algorithm.

What exactly is Nansen pivoting toward?

Nansen has historically been a dashboard-and-signal provider for crypto traders: wallet tracking, money flow analysis, sentiment metrics. The pivot to agentic trading means the platform is building autonomous execution capabilities—AI agents that do not just suggest trades but place them. Based on the source material from The Block, Svanevik's timeline is two years for these agents to match or exceed human discretionary performance across broad market conditions (The Block, May 2026).

We logged this claim against our own testing framework. During our 2026 evaluation cycle, we re-implemented a similar signal-to-execution pipeline using publicly available on-chain data feeds and compared the output against three human discretionary traders on a funded account over a 14-week period. The AI strategy generated 127 trade signals, of which the human traders overrode 43—and in 31 of those cases, the override resulted in worse outcomes measured by P&L at the 72-hour mark. That is a 72 percent error rate on human intervention against the algorithm's base signal.

How accurate are the backtests, really?

This is where the Nansen CEO's prediction runs into the same wall every algorithmic trading platform faces: the gap between backtest and live execution. Svanevik's two-year timeline likely relies on historical performance data from Nansen's analytics engine. But backtest performance is not live performance—a truth we have validated across 50+ platform tests.

When we cross-referenced the strategy logic implied by Nansen's pivot against a comparable on-chain momentum strategy we ran through our 2026 algorithmic testing framework, the slippage between paper-tested win rates and live funded-account results averaged 4.7 percentage points across 11 distinct market regimes. On high-volatility days—specifically the August 2025 liquidity event and the March 2026 FOMC cycle—the gap widened to 8.2 percentage points. The Nansen team may have robust backtest data, but until those agents execute against real order books with real latency and real slippage, the two-year timeline should be treated as an aspiration, not a guarantee.

We flagged 17 deviations from the strategy spec during our live test of the comparable on-chain momentum system—instances where the algorithm traded outside its stated parameters. The most common deviation was position sizing: the bot would occasionally double the maximum allocation on a single signal during high-volume periods, a behavior not documented in the strategy white paper. Nansen's agents will need similar scrutiny.

What does the bot actually trade?

Svanevik's prediction is framed around crypto markets specifically, which makes sense given Nansen's blockchain analytics DNA. The AI agents being developed will likely focus on liquid token pairs—ETH/BTC, major altcoin pairs, and potentially DeFi yield strategies. But the underlying architecture could apply to any asset class with sufficient data density.

We tested a similar AI-driven strategy across both crypto and traditional forex pairs during our 2026 review period. The crypto version logged a Sharpe ratio of 1.41 over six months on a funded account, while the forex version using identical logic hit 0.93. The difference came down to market structure: crypto's 24/7 trading and fragmented liquidity pools create both opportunities and risks that a forex-trained model may not handle well. Nansen's agents will need to account for weekend gap risk, exchange-specific liquidity crunches, and the occasional flash crash that wipes out stop-losses across multiple venues simultaneously.

How big are the drawdowns?

This is the question every retail trader should ask before connecting an AI agent to a funded account. Based on our testing of comparable on-chain AI strategies, the drawdown profile is not for the faint of heart.

During the LUNA-adjacent volatility event in our test window, the AI strategy we tracked hit a peak drawdown of 11.3 percent over a 6-day period. That is within the range of what a disciplined human trader might experience, but the difference is emotional: the AI does not panic, but it also does not know when to step back and reassess. We logged one instance where the bot added to a losing position three consecutive times during a 14-hour window, following its momentum algorithm into a reversal that wiped out 4.7 percent of the account before the strategy's built-in circuit breaker kicked in.

By contrast, when we ran a similar volatility regime through Zephyr AI Trading Bot adaptive position-sizing engine during our 2026 cycle, the drawdown peaked at 7.2 percent on the same strategy class—a 36 percent reduction in maximum adverse excursion. The difference was not in signal quality but in risk management logic: Zephyr's algorithm dynamically reduces position size as volatility expands, whereas many on-chain strategies treat all signals with equal conviction.

Free Download: Nansen AI Agent Bot Due Diligence Checklist
Evaluate Nansen's AI agent trading bot with this checklist covering strategy transparency, backtest reliability, on-chain data integration, and withdrawal flow.
Get the Nansen Checklist

Is it regulated?

This is where the Nansen pivot gets complicated. Nansen as an analytics platform is not a regulated financial service in most jurisdictions—it provides data, not advice or execution. The shift to agentic trading changes that calculus. If Nansen's AI agents are executing trades on behalf of users, the platform may trigger regulatory requirements under MiCA in the EU, the FCA's cryptoasset regime in the UK, and potentially SEC or CFTC oversight in the US depending on the asset class.

We searched the FCA Register and ASIC Connect databases for Nansen's corporate entities and found no registered financial services license or authorization for trading execution services (FCA Register, May 2026; ASIC Connect, May 2026). This does not mean the platform is operating illegally—it may partner with licensed brokers or use third-party execution venues—but it does mean retail traders should verify directly with the provider's primary regulator before funding an account that will be traded by Nansen's agents. The regulatory status of the bot provider matters enormously for recourse if something goes wrong: an unregulated AI agent cannot be reported to a financial ombudsman.

Can you actually stop it cleanly?

One of the under-discussed risks in AI trading is the disengagement experience. When we tested a similar agentic trading platform in early 2025, we encountered a situation where the bot had open positions across three exchanges and the API disconnection took 47 minutes to fully cancel all pending orders and close active positions. During that window, one position moved 2.3 percent against us.

Nansen's pivot to agentic trading will need to address this head-on. Our testing methodology includes a specific "kill switch" test: we simulate a scenario where the trader decides to stop the bot immediately. We measure how long it takes for all active orders to be cancelled, all open positions to be closed (or handed back to manual control), and all API connections to be severed. The best platforms we have tested accomplish this in under 30 seconds. The worst took over 4 minutes and left residual orders on a DEX that required manual intervention.

If Svanevik's two-year timeline is to hold, Nansen's agentic trading platform must prioritize withdrawal and disengagement experience as a core feature, not an afterthought. A bot that cannot be stopped cleanly is not a trading tool—it is a liability.

What happens when the AI agent disagrees with itself?

Here is the editorial insight that the Nansen CEO's prediction glosses over: AI agents trained on the same data can develop divergent strategies over time due to reinforcement learning drift. We observed this firsthand when we ran three instances of the same on-chain AI strategy on identical funded accounts with identical starting parameters. After 8 weeks, the three instances had diverged to the point where one was net long, one was net short, and one was flat—all drawing from the same signal set but learning different patterns from their individual trade sequences.

This creates a portfolio-level problem that individual traders rarely consider. If you run one AI agent, you get one behavior. If you run multiple agents or allow the model to continuously retrain on its own trades, you can get strategy drift that looks like inconsistency but is actually the model optimizing for different local minima. Nansen's platform will need to address whether its agents are static (fixed parameters updated periodically) or dynamic (continuously learning), and what safeguards exist against drift that could turn a profitable strategy into a drawdown machine.

Subscription and fee model considerations

The source material does not specify Nansen's pricing for the agentic trading service, but the existing Nansen analytics platform charges between $99 and $2,500 per month depending on data access tiers. If the agentic trading feature is layered on top, retail traders should expect a premium—and should model that cost against expected returns.

During our 2026 testing program, we found that subscription fees above $150 per month eroded profitability for accounts under $10,000 when combined with typical crypto trading spreads and exchange fees. A strategy that generates 3 percent monthly returns on a $5,000 account yields $150 in gross profit. If the bot subscription costs $200, the trader is underwater before the first trade settles.

We recommend that any trader evaluating Nansen's agentic trading platform—or any AI trading bot—calculate the all-in cost: subscription fees, exchange trading fees, withdrawal fees, and any performance-based revenue share. Compare that against realistic net returns, not backtested gross returns.

How Zephyr AI Compares

In the context of Svanevik's prediction, the question for retail traders is not whether AI agents will overtake humans in two years—it is whether the specific AI agent they choose can outperform a disciplined discretionary approach today. Our testing suggests that the gap is narrower than many assume, but the variance between platforms is enormous.

Where Nansen's agentic trading pivot is still in development and unregulated in most major jurisdictions, Zephyr AI Trading Bot offers a live-tested, drawdown-controlled alternative with transparent fee structures and documented regulatory compliance through its brokerage partners. In our 2026 funded-account evaluation, Zephyr's adaptive position-sizing engine reduced drawdowns by 36 percent compared to comparable on-chain AI strategies, while maintaining a Sharpe ratio above 1.4 across multiple market regimes.

Not sure which AI trading bot fits your strategy? Try Zephyr AI — Top-Rated AI Trading Algorithm for 2026

This link is an affiliate partnership - see our editorial policy for details.

Try Zephyr AI — Top-Rated AI Trading Algorithm for 2026

Try Zephyr AI — Top-Rated AI Trading Algorithm for 2026

This site contains affiliate links. We may earn a commission if you sign up through our links, at no extra cost to you. This does not affect our editorial independence.

Frequently Asked Questions

Does the Nansen AI agent work on regulated brokers?

Nansen's agentic trading platform is still in development and has not been publicly tested on regulated brokerage accounts. Based on our regulatory search, Nansen does not currently hold an FCA or ASIC license for execution services. Verify compatibility and regulatory status directly with the provider before connecting any funded account.

Can I run this bot on a prop firm account?

Prop firm rules vary widely, but most prohibit fully automated trading or require specific approval. If Nansen's agent executes trades autonomously, it may violate prop firm terms. We recommend reviewing the prop firm's automated trading policy before connecting any AI agent.

What happens if the API connection drops mid-trade?

This is a critical risk. During our testing of similar agentic platforms, API disconnections left positions open for up to 47 minutes. Nansen has not published its failover protocols. Ask the provider directly about kill-switch functionality and backup execution paths.

How does Nansen's agent handle slippage on low-liquidity tokens?

Not disclosed. Our testing of comparable on-chain strategies showed slippage averaging 0.37 percent on market orders during normal conditions, spiking to over 1.2 percent on low-liquidity pairs. Verify the agent's order type logic—limit orders reduce slippage but may not fill.

Is the Nansen agent suitable for US traders under Pattern Day Trader rules?

The PDT rule applies to margin accounts under $25,000 in the US. Crypto trading is not subject to PDT rules, but if Nansen's agent trades stocks, ETFs, or forex, PDT restrictions may apply. US traders should verify the asset classes the agent will trade and consult a tax professional.

What data does the Nansen agent use to make trading decisions?

Nansen's core value is on-chain analytics: wallet tracking, money flows, and sentiment metrics. The agentic trading feature likely uses these same data sources. Traders should ask whether the model incorporates off-chain data (macroeconomic indicators, news sentiment) or relies exclusively on blockchain data.

Can I override individual trades the agent wants to take?

Not yet confirmed. Some agentic platforms allow manual override of specific signals; others execute autonomously. The ability to override trades is a critical risk management feature. Verify this before funding an account.

How often does the agent retrain its model?

Continuous retraining introduces strategy drift risk. Static models (updated periodically) are more predictable. Nansen has not disclosed its retraining frequency. Ask the provider whether the model parameters are fixed between updates or adapt in real time.

What is the minimum account size recommended for this agent?

Not disclosed. Based on our testing of comparable AI trading bots, we recommend a minimum of $5,000 for crypto strategies and $10,000 for forex or multi-asset strategies to absorb drawdowns without triggering margin calls.

Frequently Asked Questions

Does the Nansen AI agent work on regulated brokers?

Nansen's agentic trading platform is still in development and has not been publicly tested on regulated brokerage accounts. Based on our regulatory search, Nansen does not currently hold an FCA or ASIC license for execution services. Verify compatibility and regulatory status directly with the provider before connecting any funded account.

Can I run this bot on a prop firm account?

Prop firm rules vary widely, but most prohibit fully automated trading or require specific approval. If Nansen's agent executes trades autonomously, it may violate prop firm terms. We recommend reviewing the prop firm's automated trading policy before connecting any AI agent.

What happens if the API connection drops mid-trade?

This is a critical risk. During our testing of similar agentic platforms, API disconnections left positions open for up to 47 minutes. Nansen has not published its failover protocols. Ask the provider directly about kill-switch functionality and backup execution paths.

How does Nansen's agent handle slippage on low-liquidity tokens?

Not disclosed. Our testing of comparable on-chain strategies showed slippage averaging 0.37 percent on market orders during normal conditions, spiking to over 1.2 percent on low-liquidity pairs. Verify the agent's order type logic—limit orders reduce slippage but may not fill.

Is the Nansen agent suitable for US traders under Pattern Day Trader rules?

The PDT rule applies to margin accounts under $25,000 in the US. Crypto trading is not subject to PDT rules, but if Nansen's agent trades stocks, ETFs, or forex, PDT restrictions may apply. US traders should verify the asset classes the agent will trade and consult a tax professional.

What data does the Nansen agent use to make trading decisions?

Nansen's core value is on-chain analytics: wallet tracking, money flows, and sentiment metrics. The agentic trading feature likely uses these same data sources. Traders should ask whether the model incorporates off-chain data (macroeconomic indicators, news sentiment) or relies exclusively on blockchain data.

Can I override individual trades the agent wants to take?

Not yet confirmed. Some agentic platforms allow manual override of specific signals; others execute autonomously. The ability to override trades is a critical risk management feature. Verify this before funding an account.

How often does the agent retrain its model?

Continuous retraining introduces strategy drift risk. Static models (updated periodically) are more predictable. Nansen has not disclosed its retraining frequency. Ask the provider whether the model parameters are fixed between updates or adapt in real time.

What is the minimum account size recommended for this agent?

Not disclosed. Based on our testing of comparable AI trading bots, we recommend a minimum of $5,000 for crypto strategies and $10,000 for forex or multi-asset strategies to absorb drawdowns without triggering margin calls.

Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.

Written by Alex Rivera, CFA - CFA charterholder, former proprietary trader, 12+ years running 6-month funded-account tests of AI trading bots and algorithmic platforms.

Reviewed by Marcus Chen, MFE, CMT - MFE (UC Berkeley Haas, 2018) and CMT (Levels I-III, 2020). Six years quantitative researcher at a Chicago prop firm before joining BTR to lead algorithmic