Low correlation alone does not make a trading bot a good diversifier
A developer running a paper-trading fleet of crypto bots discovered that adding trend-following bots to decorrelate his portfolio did not improve overall performance. While the breakout bot achieved near-zero correlation (r ≈ 0.05) with his mean-reversion bots, it posted a loss of 445 units and carried a negative Sharpe ratio across backtests. Meanwhile, two bots he assumed were diversified turned out to correlate at 0.93, meaning he was effectively doubling down on the same bet across two exchanges. The key insight from two months of daily P&L data is that low correlation is a necessary but insufficient condition for diversification — a bot with negative expected returns only adds a separate stream of losses. The author concluded that decorrelation only benefits a portfolio when each strategy also carries a non-negative edge.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in