AI trading engines keep up their losing streak

We previously reported on an interesting competition being run by Alpha Arena where different AI engines are let loose on the markets with exactly the same instructions. The last time trading was confined to cryptos, and only the Chinese AI engines managed to make money(DeepSeek and Qwen).

More recently, the competition shifted focus to tech stocks and below is the outcome. Trading was closed after two weeks, but only one AI engine came out profitable – Grok 4.2 (shown in pink in the chart below).

The rest lost money.

‍

The AI trading engines were restricted to a predefined set of highly liquid assets, all tied to the AI and tech sectors, such as Nvidia, Amazon, Microsoft and Apple. Trades could be long or short, and leveraged up to 20 times. All trades required stop-losses, profit targets, and invalidation rules (such as a forced sell if a thesis like "bullish AIleg" broke).

It seems AI engines trade just like humans– they lose (mostly). A two-week sprint should not be considered conclusive proof of anything, as it seems the bots are learning from their mistakes. Where they – and humans – come unstuck is making sense of price-sensitive news feeds.As Moneyweb noted, the market moves on some news, yawns at others: “Traders have learned to treat some news as potentially significant – such as employment numbers and interest rates – but these events are often already baked into prices.”

In the race to build Artificial GeneralIntelligence -where machines would reason much like humans – can we expect machines to lose at the same rate humans do? The stats for humans are not good:95% lose in trading. We will follow this AI test with great interest. Perhaps machines can learn to eliminate the emotion, greed and fear that account for such a staggering loss rate.