Stocks CardDashboard →

Profit

Is the business actually getting better?

Updated June 3, 2026

What it actually measures. This is the one number on the card about the company, not the stock price. It blends three things — revenue growth, margin change, and cash-flow growth — each measured year-over-year and each ranked twice: against the whole field, and against the stock's own sector. A high score means the business is improving on most of those fronts, in both contexts.

How to read it

If you see…It means…
High Profit + Low PerformThe business is improving but the price hasn't noticed. Could be an under-loved compounder — or the market sees a problem the numbers haven't caught. Worth a look; not an automatic buy.
Low Profit + High PerformThe price is ahead of the business. Either an early call on a turnaround, or a momentum story running on fumes. Scrutinize.
High Profit + High PerformA confirmed compounder — and usually fully priced. Great business, rarely a great entry. Size accordingly.
Low Profit + Low everythingA business shrinking on three fronts at once. The signal is bleaker than the single number looks.

Using it in an IRA

The card tells you about the stock. Your account tells you how much to own.

The fine print — formula, evidence, and where it fails

businessBeat ("Overall Growth") = the mean of six sub-ranks: revenue growth, operating-margin change, and operating-cash-flow growth, each computed year-over-year (trailing twelve months vs the prior twelve, to kill seasonality) and each ranked both universe-wide and within the stock's GICS sector. Rescaled to a 0–100 universe rank.

Stability. The most stable tile on the card — 71% stay-rate, 90% within ±1 decile. Fundamentals move on a slow clock, which is exactly what a business-quality descriptor should do.

Independence. Average absolute correlation +0.09; largest overlap +0.22 with Perform (good businesses tend to be recent leaders). Essentially zero correlation with Persist. Marginal information contribution ≈0.

Tier-1 evidence

A single tile's job is to describe a stable, distinct dimension of stock character — not to produce alpha on its own. We test that with three questions: does the metric stay where it puts a stock month-over-month (stability), is it independent of the other tiles on the card (orthogonality), and does it carry information the others don't (marginal contribution).

Stay in same decile
71%
90% stay within ±1 decile · 95% within ±2
Avg |correlation| (other tiles)
+0.09
Max +0.22 with Perform · window: 19y
Marginal IC contribution
≈0
IC without +0.006 → with +0.006

Correlation with the other panel headlines

Perform
+0.22
Persist
-0.01
Peril
+0.05

Diagnostics regenerated 2026-06-01

The decile evidence. Essentially flat — decile means oscillate around zero with no clean relationship to next-month return. If anything high-Profit names slightly underperform, the classic "quality is already priced in" pattern. This is a character descriptor, not a sort key.

Evidence — decile screening

Mean forward excess return vs SPY for each decile of the metric, averaged across 253 monthly anchors from 2007-01-05 to 2026-05-01. Cohort: S&P 500 (point-in-time membership via fmp PIT view).

mean fwd excess (pp/period)
value
support
D10
+0.11 pp
n=253
D9
+0.08 pp
n=253
D8
-0.10 pp
n=253
D7
+0.07 pp
n=253
D6
+0.06 pp
n=253
D5
-0.02 pp
n=253
D4
+0.10 pp
n=253
D3
+0.15 pp
n=253
D2
+0.12 pp
n=253
D1
+0.18 pp
n=253

Failure modes

  • Essentially no decile gradient — the decile-mean curve oscillates around zero with no clean relationship between businessBeat decile and forward return. High-businessBeat names slightly underperform low-businessBeat names on average — consistent with the classic 'quality is priced in' pattern. This metric is a character descriptor of the business, not a sort key for next-month returns. Use it for editorial divergences (high businessBeat + low Perform = under-loved compounder; low businessBeat + high Perform = momentum without fundamentals), not as a screen.
  • Merger accounting artifact — TTM-vs-prior-TTM comparisons across an M&A close print extreme growth that's purely definitional, not organic. EXE (Chesapeake + Southwestern, Oct 2024), SNDK (Western Digital spin-off, Feb 2025), and similar names saturate the composite at 100 in the quarters after close. The metric doesn't deduct inorganic contribution; treat extreme readings on recently-merged names skeptically.
  • Quarterly cadence creates blind spots — the v_fundamentals_ttm_q panel is PIT-correct but only refreshes when 10-Q filings arrive. For roughly 9 weeks of each quarter, the score reflects fundamentals that are 1-3 months old. Reading on a high-momentum stock between earnings is reading a stale signal.
  • Sector imbalances in narrow sectors — Communication Services has had under 25 PIT members for much of the window; sector-relative percentiles for that bucket are noisy. The composite averages six sub-ranks so the noise dilutes, but it surfaces on highly-concentrated sector cards.
  • Slow signal — the metric responds to actual business changes, which happen over 4-6 quarters. The price signal anticipates or trails it by 6-12 months in either direction. The Profit panel's editorial value is precisely in the lag — it confirms or refutes what Perform is reading, not parallel-validates it.
  • PIT membership applied — the panel uses v_sp500_constituents_pit to filter to actual historical S&P 500 members, same convention as marketBeat1Y and rankPersistence52W. Only accumDistFlow (peril) remains on a current-list cohort.

Signal = universe-rank (0-100) of the 6-sub-rank composite (rev/margin/ocf × universe/sector) at month-end Fridays. Forward = next 4-week cumulative excess return vs SPY. Top-decile basket = top 20 by composite, equal-weighted; stats are computed on the resulting per-anchor excess-return series (Sharpe = Information Ratio, CAGR = annualized excess vs SPY). PIT-correct: TTM rows only use quarters with accepted_date ≤ week_end; membership uses v_sp500_constituents_pit.

Evidence regenerated 2026-06-01

A single name across the cycle. NVDA's businessBeat percentile across the full 21-year window. The dashed line at 50 is the median; the band above 80 is the top quintile. Three distinct top-decile periods (PC gaming 2006–07, datacenter + first-crypto 2017–18, AI inflection 2024 onward), each followed by a reset to the 4th, 30th, and 42nd percentile respectively. The metric reads the settled operating cycle — it lags the operating turn by about a year because TTM-vs-prior-TTM is a slow lens by construction.

050100200520102015202020252026
NVDA · businessBeat percentile · Jun 2002May 2026 · 312 anchorslatest 79