All About AI reviews three autonomous trading experiments that ran on VPS serversA trading agent is an automated system that observes markets and places or manages trades according to a strategy. for roughly a month. Strategies that showed no useful activity were shut down early, leaving a one-cent Bitcoin market lottery, a weather-market system and a machine-learning model for five-minute Bitcoin prediction markets.
The one-cent strategy was the weakest. It earned about $13, or 3.5%, but experienced a maximum drawdown near $63Maximum drawdown is the largest recorded decline from a strategy's previous value peak to a later low point.. The approach places very low bids on both outcomes and relies on rare fills, so its modest positive result came with an unattractive risk profile.
The weather-market system produced the strongest reported curve, with about $121 in net profitNet profit is the money remaining after applicable costs and losses are subtracted from trading gains or revenue., a 28.9% returnInvestment return measures the gain or loss from an investment relative to the capital used over a stated period. and a maximum drawdown near $10. The video also demonstrates another autonomous weather strategy placing early resting orders, showing how pricing gaps and sparse fills affect execution.
A low-frequency machine-learning model for five-minute Bitcoin markets reported about $110 in net profit, a 19% return and a 72% win rate across roughly 32 tradesWin rate is the percentage of completed trades or predictions classified as successful under a stated rule.. The presenter treats all three as continuing experiments rather than settled systems, noting small samples, uneven drawdowns and ongoing strategy changes.
Watch on YouTube
