PB and ROE: comparing valuation and profitability

Chinese version

This page preserves the public summary of historical PB/ROE experiments and their method notes. The unified research entry now lives in the PB/ROE study at Quant Factor Observatory.

Plain-language takeaway: In this historical comparison using the same candidate stocks, adding static ROE produced lower returns and information ratios than ranking by PB alone. This finding applies to this sample and these rules; it does not show that ROE is generally ineffective. The “PB-only” group still required ROE data to be available.

Annualized excess return is the annualized return above the stated benchmark. Maximum drawdown is the largest fall from a portfolio high to a later low. The information ratio summarizes excess return relative to its variability; none of these measures alone proves that a result is reliable or tradable.

The published numbers are historical derived results through 2026-08-31. They were checked against source CSV files, run manifests, and audit records. This site does not independently rerun the complete history and does not present the comparison as a tradable strategy.

Narrative and published figures were checked on 2026-10-30 without rerunning the backtest or adding observations. The shared-pool PB return of 12.05% versus 9.06% with static ROE is a 2.99 percentage point annualized gap. Maximum drawdowns were 46.74% and 45.40%. This provides a baseline for testing quality information today. Current tradability still needs verified filing availability, historical financial versions and a matched execution test.

Interpretation must distinguish valuation, profitability, portfolio construction, and execution assumptions. Any conclusion is limited by the source data’s point-in-time availability and the documented sample definition.

For PE, nonrecurring PE and asset-deducted PB definitions, see the separate current-snapshot illustration. It does not extend this historical PB/ROE return experiment.