Explore ETFs

Classification checks cover exposure and strategy, not every facet. Benchmark assignment and statistics availability are separate. Confidence scores summarize evidence strength on a 0–100 scale; they are not probabilities of correctness. The confidence filter uses the lower classification or benchmark level.

Daily split- and distribution-adjusted returns are matched on expected trading sessions. Missing prices are never filled. The selected window sets the maximum lookback, so a newer fund can use its shorter available history. Correlation, BSWA beta and both OLS betas require at least 3 calendar months of eligible matched history. Both alpha models require at least 1 calendar year; an intentional blank alpha is not a calculation failure. Eligible history must still meet paired-observation coverage checks. Actual dates, observations and shorter-window labels are shown in the table and details.

Long-only equity approach: Index funds replicate straightforward market, country, capitalization, sector or standard growth/value indices. Custom strategies, including custom rules-based quality, momentum, dividend and multifactor indices, are assessed as Active and then screened for index-like returns. Legal index tracking is recorded separately from this economic approach. Active funds receive the closet-indexer statistical label when correlation is at least 0.98 and absolute annualized excess-return alpha is at most 0.70% against an eligible independent passive index proxy. The screen uses a fixed maximum of 3 years, at least 3 calendar months of eligible history and 90% paired coverage. Qualifying funds with less than one year are marked Limited history; their alpha is tested internally but stays hidden until one year. Its result does not change with the table lookback selector. Unknown means the mandate is unverified or a required screen is incomplete, including missing eligible comparisons or insufficient history. The final Active label requires a verified active mandate, at least one year of history and a complete screen that does not meet the closet-indexer thresholds. Shorter-history nonmatches remain Unknown. A statistical label describes return behavior, not intent.

Primary beta (BSWA): Slope-winsorized, age-weighted regression on daily excess returns, adapted to the selected history window. Recent observations receive more weight. Standard BSWA uses a −2 to +4 slope band. For dated, verified fixed daily leverage targets that passed the frozen validation gate, fund excess returns are divided by the signed target before fitting and the beta is scaled back afterward. Unsupported variable targets remain flagged for review. Signed OLS beta, target evidence and clipping diagnostics remain available in details. Alpha is not recomputed using BSWA beta.

Default alpha model (OLS): ETF return − RF = daily alpha + beta × (benchmark return − RF) + residual. Displayed alpha = 252 × daily alpha. Correlation is the signed Pearson correlation of matched raw total returns.

Legacy model: ETF return = daily intercept + beta × benchmark return + residual. Legacy alpha = (1 + daily intercept)²⁵² − 1. Details include its corresponding beta. The separate Funds reports retain their own methodology.

Benchmark reviews prioritize stronger, stable return correlation over a closer mandate match, using the same dates for candidate comparisons. Mandate differences and the issuer's stated index remain visible in details. Existing assignments are reviewed incrementally. ETF proxies are labeled; leveraged and inverse funds retain their unlevered long underlying. Buffer and option-income regression alpha does not evaluate an outcome promise.

Risk-free returns use published daily risk-free returns with a recent BIL extension. If either comparison series is BIL, excess-return results end at the last published risk-free date to avoid circularity. Their calculation date can therefore be earlier.