Break the Wheel: Things that Don’t Matter #3

The industry spends vast resources analyzing fund manager performance through increasingly sophisticated metrics, yet the selection process relies on the same flawed logic that makes results nearly unpredictable. Allocators know this doesn't work. They do it anyway, then blame managers for "deviating from their process" when things go wrong. The question isn't whether better metrics exist. It's whether the entire approach to picking managers has become an elaborate performance of science rather than science itself.

• The metrics everyone uses to pick managers don't actually predict future performance. Sortino ratios, alpha calculations, downside capture rates: they all fail at their stated purpose, yet allocators continue applying them as if they're scientifically rigorous.

• This isn't a manager problem. It's an allocator problem. The industry has embraced data mining dressed up in academic language, starting with returns and working backward to find reasons why they look good, rather than starting with economic theory and testing it.

• When a chosen manager underperforms, the excuse is always the same: they deviated from their process. The allocator's own process is never questioned, only the manager's, creating a self-protecting logic that can never actually fail.

• Most multi-manager portfolios generate tracking error of only 2-3%, meaning active manager bets account for less than 5% of actual portfolio risk. The enormous time spent on selection has almost no bearing on what actually happens to returns.

• The real barrier isn't finding better screening tools. It's that there may not be enough data to scientifically determine which managers will outperform at all. The question becomes whether the effort is worth the negligible impact.