Return-based style analysis, introduced by William Sharpe, evaluates a fund's performance by regressing its historical returns against a set of passive benchmark indices.

• Option C is correct: In style analysis, the fund's returns are regressed on relevant passive benchmark indices (which represent different asset classes or styles). The $R^2$ (coefficient of determination) of this regression measures the percentage of the fund's return variability that is explained by its style (its asset allocation). The remaining unexplained variance ($1 - R^2$) is attributed to the manager's active security selection.
• Option A is incorrect: Historical studies (such as Brinson, Hood, and Beebower) and typical style analysis applications show the exact opposite: asset allocation (style) usually explains the vast majority (often over 90%) of the variation in a fund's returns, whereas individual security selection contributes much less.
• Option B is incorrect: While stable exposures make a single period regression more representative of the future, modern style analysis can accommodate changing risk factor exposures (style drift) by using rolling window regressions. Constant exposures are not a strict requirement for the methodology itself to function.
• Option D is incorrect: A key feature of Sharpe's style analysis is that it relies on regressing returns against tradable and investable indices (like the S&P 500, Russell 2000, etc.), not non-tradable macroeconomic factors, so that the resulting style benchmark could theoretically be replicated by an investor.