Universe: 235 stocks: the top holdings (by ETF weight) of the 8 S&P 500 sectors currently most dislocated vs. their own multi-decade history on at least one lens. Weight lens (sector weight at <=10th or >=90th percentile): Technology (XLK), Financials (XLF), Consumer Staples (XLP), Utilities (XLU), Materials (XLB), Real Estate (XLRE). Performance lens (3-year total-return excess vs SPY at <=10th percentile), added 2026-09-28: Consumer Discretionary (XLY) and Health Care (XLV). Staples, Utilities, Materials and Technology are flagged by both. Source: State Street SSGA sector SPDR ETF holdings files; fundamentals from Yahoo Finance (XLY/XLV fetched 2026-09-28, the other six 2026-09-03 -- see each stock's dataAsOf).

Quality Score (0–100): 0-100 composite score, the equal-weighted average of 5 pillar scores. Each pillar score is itself the equal-weighted average of its underlying metrics' percentile ranks, computed across all 235 stocks in this universe (not within-sector) -- so a 90 means 'better than ~90% of these 235 stocks on this dimension,' comparable across sectors. Missing data for a metric excludes just that metric from its pillar's average (renormalized); a stock missing an entire pillar has that pillar excluded from its composite (renormalized) rather than penalized to zero.

• Profitability — Return on equity, net profit margin, operating margin, gross margin (all: higher percentile = better).
• Capital Efficiency — Average ROIC and worst-year ROIC across the (up to) 4 fiscal years Yahoo's free data provides (both: higher percentile = better). Excluded for banks and other financials, where EBIT/invested-capital isn't a meaningful framing -- their capital efficiency is captured by return on equity in the profitability pillar instead. See the 'roicAvgPct / roicMinPct' note below for why this isn't a true 5 or 10-year figure.
• Financial Health — Current ratio (higher better) and debt/equity (lower better).
• Valuation — Trailing P/E, price/free-cash-flow, price/book, PEG ratio, price/sales (all: lower percentile = better, i.e. cheaper scores higher). Negative or zero values (a data artifact of negative earnings/equity/FCF) are excluded rather than treated as 'infinitely cheap.'
• Growth — 1-year EPS growth (latest vs. prior fiscal year), best-available multi-year EPS CAGR (Yahoo's free data caps at 4 fiscal years, so this is typically a ~3-year CAGR and is labeled with its actual year count per stock -- NOT a true 5-year figure), revenue growth, and most recent quarterly earnings growth YoY (all: higher percentile = better).

Deliberately excluded from the score: Price/sentiment signals -- % below all-time high, 52-week range, short interest, analyst recommendation, normalized-EPS gap -- are NOT part of the quality score. They measure how out-of-favor a stock is (or how distorted its current earnings look), not how good the underlying business is. They're kept as separate sortable columns so you can combine 'high quality score' with 'deeply discounted' or 'earnings temporarily depressed' yourself -- the contrarian 'buy the grey clouds' screen this tool is built for.

Multi-year ROIC (avg / worst-year): Return on invested capital (ROIC), tracked over multiple years because a single high-ROIC year is easy to hit and means little -- a durably high ROIC sustained across an economic cycle is a much stronger quality signal, and is what this metric is trying to approximate.

ROIC per fiscal year = EBIT x (1 - effective tax rate) / invested capital, with all three inputs pulled directly from Yahoo's own pre-computed annual fields (not re-derived from raw statements). roicAvgPct is the average across the available years; roicMinPct is the worst single year -- included specifically to catch a stock whose average looks good only because of one exceptional year, versus one that cleared a high bar every year.

Yahoo's free fundamentals-timeseries endpoint caps annual data at exactly 4 fiscal years, verified empirically by querying it with a period1 as far back as 1995 and still getting only the most recent 4 years back for every ticker tested. A true 5-year or 10-year ROIC average is NOT obtainable from this data source -- roicYears on each stock states exactly how many fiscal years its figures actually span (typically 4, sometimes fewer for recent IPOs/spinoffs).

Banks, card networks, and other financials (about 15 of the 235 stocks, mostly in the Financials sector) have no meaningful EBIT/invested-capital framing -- interest income and expense ARE their core business, not a financing cost sitting outside operating profit -- so Yahoo doesn't report EBIT for them and ROIC comes back null rather than a misleading number. Return on equity (in the profitability pillar) is the appropriate capital-efficiency lens for those instead.

Normalized EPS & earnings-recovery flag: A mechanical proxy for a company's 'normal' earnings power, aimed at surfacing stocks whose reported (trailing) EPS is temporarily depressed by a one-off hit (impairment, write-down, restructuring charge) and expected to recover -- exactly the situation implied by 'buy when a temporary problem has crushed the reported number.' No line-item impairment/one-time-charge data is available from free sources, so this cannot identify the actual cause -- it's a statistical smoothing, not a real GAAP-adjusted normalized EPS a human analyst would derive from the footnotes.

normalizedEps = median of {each available annual fiscal-year EPS (up to the 4 years Yahoo's free data provides) + the current forward EPS estimate, when present}. The median absorbs a single unusual year (very low or very high) without needing to know why it happened. normalizedPE = current price / normalizedEps.

epsVsNormalizedGapPct: How far trailing (reported) EPS sits below (positive) or above (negative) the normalized level. A large positive gap is the 'earnings currently depressed vs. normal' signal; a negative gap flags the opposite -- trailing EPS currently running above its own normal level (a possible earnings-quality warning, not a buy signal).

Recovery candidate flag: True only when the gap is >20% AND the forward EPS estimate is itself above trailing EPS -- i.e. both the historical smoothing and the forward analyst consensus agree a recovery is already underway, not just a stock in secular decline.

The percentage gap can look extreme when normalizedEps itself is small (dividing by a near-zero number) -- always cross-check epsVsNormalizedGapAbs (the plain dollar gap) and normalizedPE, not the percentage alone, especially for stocks coming off a net loss.

Other notes:

• EPS growth figures use Yahoo Finance's free fundamentals-timeseries endpoint, which currently returns at most 4 fiscal years of annual EPS -- true 5-year EPS CAGR is not available from this data source. epsCagrYears on each stock states exactly how many years its CAGR actually spans.
• '1yr return %' (oneYearReturnPct) is Yahoo's 52-week price change, not exactly a trailing-12-month return.
• All fundamentals are a single snapshot fetched from Yahoo Finance; refresh by re-running the fetch + this script.