AI Crypto Trading Competition 2025: Qwen-3 Dominate as Gemini, Grok and GPT-5 Suffer Major Losses
- AI models like DeepSeek, and Qwen-3 are up big in a live, real-money crypto trading showdown.
- GPT 5 is the biggest loser of the bunch, underscoring reliability and transparency risks.
- The results deepen the divide between data-specialized and general-purpose AI approaches to finance.
The "Alpha Arena," a competition that pits prominent large language models against each other in the live cryptocurrency market, saw OpenAI's GPT-5 with staggering losses of more than 70% during the same period. (decrypt.co)
The other LLMs included in the first batch of models for the experiment, which runs until November 3, are Alibaba Cloud’s Qwen 3 Max, Anthropic’s Claude 4.5 Sonnet, Google DeepMind’s Gemini 2.5 Pro and xAI’s Grok 4. Alibaba Cloud is the AI and cloud computing unit of Alibaba Group Holding, owner of the Post.
Each AI model was given a starting capital of $10,000 to trade cryptocurrency perpetual contracts on the Hyperliquid exchange, betting on assets including Bitcoin, Dogecoin, and Solana.
Four of the most popular AI programs crashed and burned in a cryptocurrency investing competition — with most losing over 50 percent of the money they were instructed to maximize.
Only one Chinese program managed to turn even a modest profit.
The experimental Alpha Arena contest from company Nof1 gave six AI models $10,000, identical input data and prompted each to make as much money as possible while trading crypto stocks on the open market from Oct. 17 to Nov. 3.
The goal: Find which bot could best make you rich.
Jay Azhang, 34, told The New York Post that the goal of his brainchild project — and future iterations of the competition — is to help the average person get the best tools to make money.
“Our goal is to help people trade better and give them access to, hopefully, the state-of-the-art when it comes to models for trading,” Azhang said. “It’s a unique area of research.
“No one else is really training these models to look at all these numbers.”
AI programs Grok from xAI; Anthropic’s Claude Sonnet; Google’s Gemini; OpenAI’s ChatGPT and Chinese-owned bots Deepseek and Qwen, got the five-digit endowments and were directed to make all other decisions — including whether to bet long or to short their choice of cryptocoins.
The bots were able to invest in blockchain cryptocurrencies Bitcoin, XRP, Ethereum, Doge, BNB and Solana, with all trades recorded and shared publicly on the competition’s website, which charts the financial positions of each program second-by-second.
Azhang said the crypto market, rather than trading stocks and equities, proved best for the competition parameters because of the readily available blockchain data, the 24-hour trading cycle and the lack of influence and advantage individual traders have compared to hedge funds in other financial markets.
“Also, a bit more volatility,” Azhang said of the crypto market. “So a little bit more exciting.”
The results showed a shocking ineptitude.
ChatGPT, the most popular bot according to StatCounter, ended with just $3,794 — down 63%.
Next worst was Gemini, down 56% with only $4,485 left, despite making the most trades, 272.
Elon Musk’s Grok and Anthropic’s Claude Sonnet had middling results and were occasionally profitable during the 17-day trial. But Grok ended down 45% with $5,226, while Claude Sonnet pocketed $6,740, down 30%.
Chinese model Deepseek saw profits sink in the final days of trading, but ended up in the black — with $10,476 for a modest 4% return. It was at 100% around Oct. 26.
Qwen, from Chinese company Alibaba, was most volatile — operating at a loss from its initial investments for the first three days, and then successfully dumping in all but $90 of its remaining bankroll into a long position on Bitcoin. It won the competition with 20% growth and $12,287.
During the period of Oct. 17 to Nov. 3, Bitcoin was down .44 percent, Ethereum down 11 percent, XRP up .87 percent, BNB down 8 percent, Doge down 10 percent, and Solana up just over 1 percent, according to data from Coinbase.
“It’s hard to say how much we can take away from this,” Azhang said. “One thing that we do know is that there are patterns in the models and they’re clearly biased and have preferences.
“For example, Claude almost always goes long and refuses to go short. It’s like an eternal optimist whereas Gemini is happy to short,” Azhang said. “They clearly have these inductive biases when it comes to trading.”
Alpha Arena plans a next round that will add more AI models and also task the programs with trading equities along with crypto.
“We’re just getting started,” Azhang told The Post.The results could be sending a complex signal to Wall Street, as the two frontrunners represent two vastly different potential futures for artificial intelligence in finance. DeepSeek is reportedly backed by a Chinese quantitative hedge fund, suggesting its success may stem from specialized financial data and expert fine-tuning—an evolutionary step for today's data-driven firms.
Still not ready for primetime
However, the catastrophic losses of models like Gemini highlight the significant risks that make financial institutions wary. A primary concern is the "black box" nature of these systems, where the reasoning behind a trade is often opaque and unexplainable. This lack of transparency is a major hurdle for regulatory compliance and risk management, as establishing trust in a model's decisions is a critical and ongoing effort.
Beyond opacity, there are fundamental concerns about reliability. These models are known to be prone to hallucinations—fabricating convincing but false information—which could be catastrophic in a live trading environment.Furthermore, a 2024 paper exploring the implications of LLMs in financial markets warns of a novel systemic risk: if multiple, seemingly independent AI agents are built on the same underlying foundation models, they might react to market events in a correlated way, potentially "amplifying market instabilities" and creating unforeseen flash crashes.
- https://decrypt.co/345006/ai-crypto-trading-showdown-deepseek-grok-winning-gemini-implodes
- https://nypost.com/2025/11/08/business/ai-models-given-10k-to-compete-in-first-of-its-kind-crypto-trading-competition-and-most-crashed-and-burned/
Frequently Asked Questions (FAQ)
1. What is the competition being discussed and how does it work?
The competition is the Alpha Arena Crypto Trading Competition (hosted by Nof1) where six major AI models each started with equal capital (USD 10,000) and autonomously traded cryptocurrencies (such as BTC and ETH) without human intervention. South China Morning Post
Each model was given the same market data and instructions: to maximise trading returns by deploying strategies in the volatile crypto futures market. ForkLog
2. What are the main goals and significance of this competition?
The goals are multiple:
-
To test how well advanced AI can operate in real-world crypto markets — including pricing, risk, leverage, volatility. icobench.com
-
To compare different AI models’ trading strategies (long vs short, leverage, position sizing) under identical conditions.
-
From a broader perspective, the event showcases how AI is entering the autonomous finance and algorithmic trading space — which may have implications for future trading systems, hedge funds, and market structure.
3. What are the major risks or caveats investors/readers should be aware of?
While this competition is interesting, it comes with important caveats:
-
Past performance even in the contest doesn’t guarantee future results in live investing — the environment is specific and controlled.
-
Leverage is used heavily in some models; heavy losses (or full draw-downs) are possible. Some models in the competition already suffered big losses.
-
Crypto markets are extremely volatile, illiquid at times, and subject to external shocks (regulation, exchange risk, hacks). An AI doing well in a contest may still fail in real-world scalable conditions.
-
There may be differences between contest conditions and actual trading (fees, slippage, market depth, human supervision).
Hence, one should view the competition as a learning/benchmark exercise, not a ready-made “AI trading strategy you can copy”.
4. How can an investor use the insights from this competition in their own crypto strategy?
Here are some actionable take-aways:
-
Study what worked: e.g., which models favoured long vs short positions, how they sized trades, how they reacted to market moves.
-
Consider developing or selecting algorithmic strategies that emphasise risk management (draw-down protection, position limits) — since even AI models failed when the market turned.
-
Use the competition as a signal of technological trends: AI in trading may become more prominent, so staying informed may give you an edge.
-
But don’t blindly replicate: ensure you understand the underlying logic, fit it to your own risk tolerance, capital size, and market access.
5. Is this competition something that retail investors can participate in or be affected by?
Direct participation: Not really — the contest is between big AI models with large computational resources and predefined capital. Retail investors cannot join the same format.
Indirectly: Yes — outcomes may influence the broader market. For example, if AI-driven algorithms generate unusual flows, they might affect market volatility, liquidity, or the strategies of professional trading firms.
Also: If you hear claims that “AI beat the market so join us and we’ll automate your trading” — you should be cautious and do your own due diligence (see also risk of scams below).
6. Could this competition lead to more widespread automated/AI trading in crypto markets?
Quite possibly. The competition demonstrates proof-of-concept of autonomous AI trading in a high-volatility asset class. As infrastructure (cloud, exchanges, data feeds) gets cheaper and regulation advances, more firms may adopt such systems.
However: The scale and robustness required for live markets is far higher than a contest environment. Variables like slippage, market impact, liquidity crunch, regulatory risk remain challenging. So while the competition signals a trend, it should not be seen as full-scale deployment yet.
7. What do the results suggest about which strategies (e.g., long/short, leverage) are working in the contest?
From the early reports:
-
Models like DeepSeek used both long and short positions across multiple tokens with up to 10× leverage and had strong returns in favourable market conditions.
-
Qwen took a heavy long position (e.g., 25× long on ETH) when the market moved up, which gave strong gains but also increased risk.
-
Models that mis-timed the market (shorting during a rally) or used high-frequency trading without strong risk control incurred large losses. South China Morning Post
Thus: while leverage and aggressive strategies can magnify gains, they also magnify losses. Timing and strategy are still key.
8. What should I do next if I want to follow or learn from this competition?
Here are some suggestions:
-
Monitor official updates from the competition organiser (Nof1) and credible crypto news outlets for final results and strategy post-mortems.
-
Read analyses of which AI models performed best and why (e.g., strategy breakdowns, timing, risk).
-
Stay current on broader AI-in-trading developments: algorithmic systems, regulatory changes, exchange infrastructure.
-
If intrigued: explore algorithmic trading basics — strategy development, back-testing, risk control — rather than jumping into untested “AI trading bots” or services.
-
Always apply strong risk controls: maintain diversified holdings, limit exposure to speculative/trading-heavy strategies, and ensure you understand what you are doing.
- AI Market Predictions for 2025: Hits, Misses, and Ongoing Trends as of October 2025
- Top 10 Stocks and ETFs Poised to Outperform in November 2025
- Best Artificial Intelligence (AI) Stocks of 2025
Comments