APG

An ABSOLUTE intrinsic-value signal, distinct from the relative multiples (PE, EV/EBITDA) which only compare a stock to its industry peers. Intrinsic value comes from the Residual Income Model: book value plus the present value of the residual income (earnings above a charge for the cost of equity) the firm is forecast to generate. This is the best-VALIDATED absolute value signal in the academic literature (Frankel & Lee 1998, replicated through 2026): the value-to-price ratio robustly predicts 1-3 year cross-sectional returns, survives every major factor model, and subsumes book-to-market. Because it is scored across the WHOLE universe (not within-industry), it complements the relative multiples rather than double-counting them. Pair it with a quality screen (Piotroski) to avoid value traps, and read the EPV decomposition to see whether a discount is unpriced earnings power or just speculative growth coming out of the price. One caveat: the model anchors on book equity, so for low-book, buyback-heavy, or asset-light firms (think Apple or Coca-Cola) book is a poor floor and the model can read durable franchises as wildly overvalued. Readings are clamped, and the metric is currently shadow-scored — shown for context, with zero weight in the fundamentals grade — pending validation.

Productivity of the asset base — how efficiently the firm converts assets into sales. Combined with net margin, this is the core of the DuPont decomposition: ROA = net margin × asset turnover. Two businesses can land at the same ROA via opposite paths: thin-margin / high-turnover (Costco, supermarkets) vs. fat-margin / low-turnover (luxury, software).

Sanity-check on earnings quality — a ratio of 1.0 means every dollar of GAAP earnings shows up as free cash flow. Persistently below 1.0 means working-capital build or capex is consuming earnings; persistently above 1.0 is unusual and may flag accounting conservatism (heavy non-cash charges) or a one-time release. Trend matters more than absolute level.

Joel Greenblatt's reformulation of return-on-capital, designed to filter for businesses that earn high pre-tax returns on the tangible operating capital that actually generates EBIT. Excludes goodwill and acquired intangibles so that two firms with the same operating economics rank similarly regardless of acquisition history. Pairs with Greenblatt's Earnings Yield (EBIT/EV) to produce the famous 'Magic Formula' rank — quality stocks at value prices.

Operating margin minus interest, tax, and below-the-line items. Useful but noisier than operating margin: tax rate changes, debt-service shifts, and one-time items can move net margin year-over-year even when the operating business is unchanged. Always pair with operating margin and FCF margin — divergence between them is a tell about leverage, tax structure, or earnings quality.

Pre-capex version of cash conversion: compares operating cash flow directly to net income, before subtracting capex. Numbers persistently below 1.0 suggest accruals are doing the work (revenue recognised before cash collected, or expenses deferred). Numbers above 1.0 are typically driven by D&A exceeding true economic depreciation — a mild positive sign but not always meaningful.

Captures the leverage between revenue and operating profit — the core test of whether a business converts sales into income at the operating level. Independent of capital structure (excludes interest) and tax-jurisdiction noise. Trend matters as much as level: an improving operating margin indicates pricing power or operational efficiency gains; a falling margin signals input-cost pressure, wage inflation, or competitive erosion.

Capital-efficiency measure that ignores how the assets are financed. Useful for comparing within a sector but heavily biased by industry — banks and utilities run low ROA by design (huge balance sheets), software and franchise businesses run high ROA (asset-light). The DuPont decomposition links it to ROE: ROE = ROA × (assets / equity), so leverage inflates the gap between ROA and ROE.

Pre-tax cousin of ROIC using operating capital employed (long-term debt plus equity, approximately equal to total assets minus current liabilities). Damodaran publishes ROCE by industry annually, so it benchmarks cleanly against sector norms. Particularly useful for capital-intensive businesses where you want to see pre-tax returns on the long-term capital base.

The headline profitability metric. Net income divided by average book equity tells you how productive each dollar of shareholders' capital is. High sustained ROE compounds book value when the business can reinvest at a similar return. Two warnings: ROE rises mechanically with leverage (high D/E inflates ROE without making the business better), and aggressive buybacks can shrink the equity denominator faster than they shrink earnings, artificially boosting ROE even as fundamental returns are flat.

Measures how productive the actual operating capital base is, independent of capital structure. Strips out the effects of cash, leverage, and below-the-line items so you see the underlying economic engine. The benchmark is the firm's weighted-average cost of capital (WACC) — sustained ROIC above WACC creates value, sustained ROIC below WACC destroys it. The single most important profitability metric for long-term compounding.

The single best operations-quality metric. Negative CCC means the firm is funded by suppliers — the customer pays before the supplier needs to be paid (Apple, Costco, Amazon at scale). Long CCC ties up working capital and forces operating-loan dependence. Combines three sub-metrics: days inventory outstanding, days sales outstanding, days payable outstanding.

How long the average sale sits as a receivable before becoming cash. Earnings-quality flag: receivables growing faster than revenue is one of the cleanest tells for revenue recognition stretching — Beneish flagged exactly this pattern as a primary fraud indicator. Industry context matters: B2B / project businesses run longer DSO than retail, but the trend is more important than the level.

CCI measures how far the current typical price ((H+L+C)/3) sits from its rolling SMA, normalized by the mean absolute deviation. Originally built for commodity cycles by Donald Lambert; widely used as a generic momentum / overbought-oversold gauge on any asset. CCI is unbounded but ±100 is the conventional 'normal range' edge — readings beyond that are interpreted as strong directional thrusts or stretched extremes depending on context.

+1 while CCI is below -100 (oversold); -1 while CCI is above +100 (overbought); 0 inside the ±100 band.

+1 = oversold (stretched below the typical-price mean, mean-reversion buy bias), -1 = overbought (stretched above), 0 = inside the ±100 band.

Designed to buy oversold pullbacks within an established uptrend (price above the regime moving average). Uses a 2-period RSI rather than the standard 14 to make it hyper-responsive to short-term swings. The default exit_method is RSI-threshold based: hold while RSI(2) is below the configured exit threshold and exit when RSI(2) crosses above exit. The book-style SMA exit is also supported by setting exit_method=sma with exit_ma.

+1 when price is above regime_ma and RSI(2) drops below oversold; held while inside the trade; back to 0 when the configured exit_method fires.

+1 = pullback-buy state active, -1 = pullback-short state active (symmetric rule in a down regime), 0 = flat.

MFI is essentially RSI computed on (typical price × volume) instead of close alone, so high-volume bars push the oscillator harder than low-volume bars. It's 0-100 bounded and uses the same 20/80 OB-OS convention as a slower RSI. Because it folds in volume, MFI can flag distribution / accumulation underneath a flat price tape that RSI would miss.

+1 while MFI is below 20 (oversold); -1 while MFI is above 80 (overbought); 0 in between.

+1 = oversold (bullish, volume-weighted), -1 = overbought (bearish), 0 = mid-range.

Stochastic RSI applies the stochastic oscillator formula to RSI values rather than price, producing a faster, more sensitive 0-100 momentum read. Readings below 20 are oversold and above 80 overbought. The signal latches bullish when FastK crosses above FastD while in the oversold zone and stays on until FastK drops back below FastD; the bearish mirror latches on a cross down in the overbought zone. More reactive than plain RSI, so best filtered by trend context.

+1 latches when FastK crosses above FastD while FastK is below the oversold threshold (default 20), and holds until FastK falls back below FastD; -1 latches when FastK crosses below FastD while FastK is above the overbought threshold (default 80), and holds until FastK rises back above FastD; 0 otherwise.

+1 = bullish oversold cross still latched, -1 = bearish overbought cross still latched, 0 = no latched cross.

TRIX is the one-bar rate of change of a triple-exponentially-smoothed closing price. The triple smoothing strips out short-term noise, so TRIX oscillates slowly around zero. A cross up through zero marks momentum turning positive (bullish); a cross down through zero marks it turning negative (bearish). Smoother and lower-frequency than most momentum oscillators, which trades responsiveness for fewer whipsaws.

+1 while TRIX is above zero; -1 while TRIX is below zero; 0 only at exactly zero.

+1 = TRIX turned positive (bullish momentum), -1 = TRIX turned negative (bearish momentum).

Williams %R maps close-to-range position onto a -100 to 0 scale: -100 means the close is right at the lookback period's low (oversold), 0 means right at the period's high (overbought). The oscillator's interpretation mirrors RSI but the scale is inverted. Like other OB/OS oscillators, it works best as a tactical timing aid inside an already-established trend, not as a standalone reversal trigger in choppy markets.

+1 while %R is below -80 (oversold); -1 while %R is above -20 (overbought); 0 in between.

+1 = oversold (bullish), -1 = overbought (bearish), 0 = mid-range.