Frequently Asked Questions

On QW Research, enter any stock ticker in the search bar to see which institutions hold it, how many shares they own, and how positions have changed over recent quarters. You can view the top holders table, ownership trend charts, and Sankey flow diagrams showing capital movement between positions.

Crowding analysis measures how concentrated institutional positioning is in a stock relative to its peers. A highly crowded stock has an unusually large share of its float held by institutional investors. While this can indicate strong conviction, extreme crowding can also signal risk — if many institutions try to exit simultaneously, selling pressure can be amplified.

Alpha is the excess return a fund's reported holdings generated beyond what its risk exposures explain — our attempt to separate genuine stock-picking skill from luck or style. QW Research shows a “Skill Alpha (FF)” for each institutional filer: we reconstruct each fund's quarterly return from its 13F holdings, then run a Fama-French 4-factor regression. The return left over after removing overall-market, size (small vs large cap), value, and momentum exposure is the skill alpha. Browse and sort funds by it on the Filer Screener.

For each quarter we take the portfolio a fund reported at the end of the previous quarter — so we never use hindsight about how the quarter turned out — and measure what that exact portfolio returned over the quarter, weighted by each position's size. We then regress those quarterly returns on the four Fama-French factors; the regression intercept is the annualized skill alpha. We also show a simpler 'vs S&P' figure: the raw return minus the S&P 500, with no risk adjustment. Everything is computed only from public 13F holdings, so it covers long U.S. equity positions only and cannot see intra-quarter trades.

'vs S&P' is the raw return of a fund's holdings minus the S&P 500 — it is not risk-adjusted, so a fund that simply tilts toward small-cap or value stocks will show up with alpha (or, since the S&P 500 was a high bar, often negative). 'Skill Alpha (FF)' strips out those market, size, value, and momentum tilts and keeps only what is left — a much better estimate of genuine stock-picking skill. We lead with Skill Alpha and show 'vs S&P' as context.

They tell you how much to trust an alpha number, since it is estimated from a limited number of quarters. The t-stat is the alpha divided by its standard error — roughly how many standard errors it sits from zero; a t-stat above about 2 means the alpha is large relative to its noise. The p-value is the chance of seeing an alpha that big by luck if the fund truly had no skill; below 0.05 we call it 'statistically significant'. Funds whose alpha is not significant are marked 'ns' — treat those as noise rather than skill.

Probably not — and we tested this rather than assume it. We ran an out-of-sample persistence test: we split each fund's history in half, measured skill alpha in the earlier half, and checked whether it predicted alpha in the later half. It barely did. Funds in the best-performing and worst-performing quartiles had almost identical future alpha, and of the funds that were significantly skilled early, only about 43% stayed even positive later and under 1% stayed significant. This matches decades of academic research: past outperformance rarely persists. So we present alpha as an honest historical, risk-adjusted record — useful for understanding what a fund has done, not as a buy signal.

Yes, QW Research is completely free. There is no paywall, no account required, and no usage limits. All data is sourced from public SEC EDGAR filings.

All data comes from SEC EDGAR, the official electronic filing system of the U.S. Securities and Exchange Commission. We process 13F-HR filings for institutional holdings and XBRL financial facts (from 10-K and 10-Q filings) for shares outstanding and other company metrics. This is the same primary data source used by Bloomberg, Refinitiv, and other financial data providers.

CUSIPs are 9-character identifiers assigned to securities. Some 13F filings contain CUSIPs for securities that have been delisted, merged, or reorganized — making them difficult to map to current tickers. Others may reference private placements, foreign securities, or convertible instruments that don't have standard stock tickers. Our enrichment pipeline resolves the majority of CUSIPs to tickers, but some remain unmapped due to these data quality challenges inherent in the 13F filing process.