Lotus Holdings' Compute Pivot: A Classic Case of Factor Backtest Overfitting

Comparing a MSG company suddenly pivoting to computing power to the most common trap in quantitative strategies—factor backtest overfitting—is as apt as it gets. In historical data, the core MSG business was stable with decent cash flow. When management decided to "cross tracks," they often only saw the high average returns of the computing power industry, ignoring sample variance and tail risks. It's like using three months of data in 2015 to backtest a "sure-win" factor, then going all-in with real money, only to hit a black swan event like the 2020 pandemic, causing the curve to collapse instantly. Lianhua Holdings' computing power landmine is essentially a decision that wasn't validated by sufficient out-of-sample testing, yet was executed by management as if it were a stable strategy.