Key insights
- The post explores a hypothetical high-frequency trading strategy using AI on SGOV, a short-term Treasury bill ETF. It questions the feasibility and profitability of exploiting minuscule price movements at scale. While theoretically appealing, transaction costs and existing sophisticated algorithms likely negate the profitability for retail-level implementation. The post raises concerns about the potential impact of large-scale AI-driven trading on market liquidity and stability, but the immediate impact on US equities is low.
How close are we to having AI turn the USD into Monopoly money?
Take something very predictable like SGOV, which today will bounce from $100.64 to $100.65.
I have $10M and have trained an AI to buy at .64 and sell at .65, hundreds, if not thousands of times per day.
10,000,000*(1.0001^1000)=11,051,653.926
(Take .01% profit and reinvest, 1000 times in a day)
30 days of that is $221M.
Would the cost of these trades make this not profitable? Is this already happening? What’s this look like large-scale? Where does this money come from? Are we fucked? Am I crazy?