DeepSeek outpaces competitors in AI trading and Binance’s CZ wants to know how

Binance co-founder Changpeng Zhao says he has been receiving a lot of positive opinions on artificial intelligence AI trading on his X feed, but questioned how shared trading algorithms could outperform the market when many traders are using the same strategies simultaneously.

Responding to a post on X showing AI trading models competing in real time early Monday, CZ said he thought “AI strategies work if traders have their own plan that is better than others, and no one else has it.” 

“Saw this a lot in my feed. DeepSeek out performing the rest in AI trading. How does this work? … You are just buying and selling at the same time as others,” he surmised.

Saw this a lot in my feed.

DeepSeek out performing the rest in AI trading. How does this work?

I thought trading strategies work best if you have your own unique strategy that is better than others, AND no one else has it. Otherwise, you are just buying and selling at the same… https://t.co/ExXZeAwx8p

— CZ 🔶 BNB (@cz_binance) October 20, 2025

CZ was reacting to a trading leaderboard from Alpha Arena by Nof1 showing various large language model (LLM)-powered trading bots competing over the course of 72 hours. 

DeepSeek tops leaderboard as AI bots compete

According to the Alpha Arena chart, DeepSeek Chat V3.1 led rankings with a total account value of $11,995.57, a 19.96% gain in the three-day trading session. The bot virtually executed several leveraged long positions, including an ETH 15x long worth $19,375 with an unrealized profit of $1,648.53, together with SOL and BNB positions showing profits of $696.63 and $327.30, respectively.

In second place, Claude Sonnet 4.5 posted a total account value of $10,584.58, up 5.84% from its starting balance. The model held an XRP 8x long valued at $12,335 with $441.78 in unrealized profit and a BTC 20x long worth $67,357 generating $380.99 in profit.

At the lower end of the leaderboard, GPT-5 saw 36.82% loss on a total account value of $6,318.04. The OpenAI LLM made short bets in XRP and DOGE, both posting losses of $378.52 and $169.89, and a small gain from a BTC 10x long worth $9,647 with $29.22 in profit.

The post that caught CZ’s attention originated from an advisor at global digital asset management firm Amber Group, who shared the Alpha Arena chart lauding DeepSeek’s lead.

“Right now, DeepSeek is way ahead! To be honest, this feels so much more impactful than the AI memes from last year’s Goat/Ai16Z era,” the advisor wrote. “Has AI x Crypto finally found the right way to take off?”

The advisor contrasted the LLM’s trading choices with human traders who neglect to define exit parameters before opening positions.

“I’ve never been a fan of trading futures, but if you’re going to do it, shouldn’t you learn from these AIs and set proper take-profit and stop-loss conditions before opening a position?” they asked.

Is AI really good for trading? Reddit approves

CZ’s remarks and the discussion on X  is similar to the forum chatter on other social media platforms like Reddit’s trading communities, where users have been sharing anecdotes of using ChatGPT and other LLMs for market analysis.

On Reddit’s Wall Street Bets, one user claimed to have built a neural network predicting stock movements using daily news topics. “I tested it for one month and it outperformed my HYSA account by about 0.2%,” they claimed. 

However, co-founder of blockchain infrastructure company XYO Markus Levin believes many investors overestimate how much insight AI tools can provide. 

“In my experience, LLMs often pull from a small, self-reinforcing pool of sources when analyzing early-stage or niche projects. That usually means company press releases, founder posts on X, Reddit threads, and carefully managed media appearances,” Levin argued.

According to Levin, proprietary AI systems used by trading firms can achieve higher accuracy rates, but publicly accessible models are prone to bias and lack real market data feeds. 

“They’re asking public LLMs questions like, ‘what project will 5–10x this year?’ and treating the responses as investment insight. That’s where the real risk lies,” he added.

Some Wall Street and business enthusiasts like Eric Croak, president of Croak Capital, coin retail use of generative AI in crypto markets “algorithm-assisted gambling.”

“I believe what makes AI dangerous with crypto and options specifically is its inability to explain asymmetric risk in real terms. These bots can easily omit factors like tax consequences or liquidity concerns. If it reads like a marketing blog instead of an investment memo, that is your warning,” he concluded.

Join a premium crypto trading community free for 30 days - normally $100/mo.