We Ran 19 Valuation Models on 3,091 US Stocks -- Here Is What the Data Shows

Almost exactly half of the US stock market trades below its own multi-model fair value estimate. As of July 1, 2026, we scored 3,091 US-listed stocks with Equity Rank's 19 valuation methodologies and blended each stock's estimates into a single model consensus fair value. The result: 1,553 stocks (50.2%) trade below their model consensus fair value and 1,521 (49.2%) trade above it. The median stock trades within a quarter of one percent of its model estimate.

That headline sounds like textbook market efficiency. The distribution underneath it does not.

Key findings at a glance

• 50.2% of covered US stocks (1,553 of 3,091) trade below model consensus fair value; 49.2% trade above it. The median margin of safety across the whole universe is +0.2% -- essentially zero.
• Only 27.9% of stocks (863 of 3,091) trade within 10% of model fair value in either direction. The typical stock is not near its model estimate; it is far from it, on one side or the other.
• The tails are heavy. 651 stocks (21.1%) trade at least 30% below model consensus fair value, and 363 (11.7%) trade at least 50% below it. On the other side, 376 stocks (12.2%) trade at least 30% above model fair value, and 112 (3.6%) trade at more than double it.
• Size is the strongest pattern in the data. Only 7 of the 68 mega-cap companies (market cap of 200 billion dollars or more) trade below model consensus fair value -- 10.3%. Among small caps (300 million to 2 billion dollars), 61.5% do. The median mega-cap trades 25.3% above its model estimate; the median small cap trades 6.9% below it.
• The models disagree with Wall Street on most stocks. Across 2,980 stocks with a published sell-side consensus analyst estimate, the model consensus fair value sits below the analyst figure for 66.9% of names, with a median gap of -14.5%. Only 20.6% of stocks have the two numbers within 10% of each other.
• The models also disagree with each other -- a lot. For the median stock, the most generous of its individual model estimates is 9.8x the most conservative one, before any consensus weighting. Model disagreement is widest in Automotive (15.8x median spread) and narrowest in Banks (4.4x).

Everything below is sector-level and distribution-level data. This study names no individual companies and offers no view on what any reader should do -- it is a snapshot of what 19 valuation lenses see when pointed at the same market on the same day.

The distribution: a market of extremes, not averages

Margin of safety here means the gap between a stock's model consensus fair value and its current price, expressed as a percentage of price. Positive means the stock trades below model fair value; negative means it trades above it.

Margin of safety band	Stocks	Share of universe
Price more than 50% above model fair value	112	3.6%
30% to 50% above model fair value	264	8.5%
10% to 30% above model fair value	711	23.0%
0% to 10% above model fair value	434	14.0%
0% to 10% below model fair value	429	13.9%
10% to 30% below model fair value	490	15.9%
30% to 50% below model fair value	288	9.3%
More than 50% below model fair value	363	11.7%

The interquartile range runs from -17.3% to +24.8%. In plain terms: a quarter of the market trades more than 17% above its model estimate, and another quarter trades nearly 25% below it. The "efficient" median is an artifact of two large, opposing tails canceling out.

One more distribution fact: 783 of the 3,091 covered stocks (25.3%) are currently loss-making. Their median margin of safety is +3.0%, versus -1.0% for profitable companies -- a small gap, which surprised us. The models do not systematically flag unprofitable companies as trading above fair value, in part because earnings-based methods drop out where earnings are negative, leaving revenue-anchored and asset-anchored estimates to set the consensus.

Sector breakdown: where prices sit relative to the models

The table below covers every sector with at least 40 scored stocks (internal sub-sector labels such as semiconductor sub-industries are folded into their parent sector). "Below model FV" is the share of that sector's stocks trading below model consensus fair value. Median MoS is the sector's median margin of safety.

Sector	Stocks	Below model FV	Median MoS
Communication Services	45	64.4%	+22.1%
Asset Management	95	65.3%	+20.4%
Software	177	66.1%	+10.9%
Consumer Discretionary	96	60.4%	+8.6%
Specialty Chemicals	40	60.0%	+7.5%
Materials	65	64.6%	+7.2%
Consumer Staples	78	61.5%	+6.9%
Medical Devices	88	58.0%	+6.2%
Financials (diversified)	107	56.1%	+4.2%
Real Estate	150	58.0%	+3.3%
Biotechnology	210	58.1%	+2.9%
Energy	177	54.8%	+2.2%
Mining	48	52.1%	+1.3%
Banks	261	53.3%	+0.9%
Industrials	274	48.5%	-0.5%
Insurance	81	45.7%	-1.6%
IT Services	42	40.5%	-5.8%
Life Sciences Tools	49	32.7%	-6.5%
Utilities	88	30.7%	-14.8%
Aerospace & Defense	52	23.1%	-15.6%
Semiconductors	79	17.7%	-19.0%
Specialty Retail	46	23.9%	-21.6%
Hardware	48	12.5%	-24.7%
Automotive	42	4.8%	-31.2%

Three observations, stated as observations:

1. Automotive is the most stretched sector in the study. Only 2 of 42 automotive stocks (4.8%) trade below model consensus fair value, and the sector's median stock trades 31.2% above its model estimate. Hardware (12.5% below model FV) and Semiconductors (17.7%) are close behind.
2. The widest positive readings cluster in Communication Services, Asset Management, and Software, where roughly two-thirds of stocks trade below model fair value. Under these assumptions, those readings correspond to potential undervaluation at the sector level -- though sector-level medians can hide very different company-level stories.
3. The banking sector is the market's most "fairly priced" large group. With 261 banks scored, the median margin of safety is +0.9% -- the closest to zero of any large sector -- and, as shown below, banks are also where the 19 models agree with each other most tightly. Book-value-anchored businesses are simply easier for valuation models to triangulate.

The size effect: the bigger the company, the bigger the premium to the models

Grouping the same 3,091 stocks by market capitalization produces the cleanest monotonic pattern in the entire study:

Market cap group	Stocks	Below model FV	Median MoS
Mega cap (200B+)	68	10.3%	-25.3%
Large cap (10B-200B)	847	37.2%	-9.5%
Mid cap (2B-10B)	1,067	51.5%	+1.0%
Small cap (300M-2B)	1,081	61.5%	+6.9%
Micro cap (under 300M)	28	57.1%	+3.7%

The median mega-cap stock trades 25.3% above its model consensus fair value; the median small cap trades 6.9% below it. Sixty-one of the 68 largest companies in America trade above what the blended models estimate they are worth.

Two honest readings of the same fact. One: the market pays a persistent premium for scale, liquidity, index membership, and perceived moat quality that fundamentals-anchored models do not fully capture. Two: valuation models built on reported financials may systematically underestimate the durability of mega-cap economics -- the models have been "wrong" about this group before. This study measures the gap; it does not adjudicate which explanation is right, and a gap between price and a model estimate is not a prediction about future prices.

The models vs. Wall Street

For 2,980 of the covered stocks, a published sell-side consensus analyst estimate is available alongside the model consensus fair value. The two disagree far more than they agree:

• The model consensus fair value sits below the published analyst consensus figure for 66.9% of stocks (1,995 of 2,980), with a median gap of -14.5%.
• Only 20.6% of stocks have the model estimate and the analyst estimate within 10% of each other. The median absolute disagreement is 25.3%.
• The published analyst consensus figure sits above the current market price for 79.9% of stocks. The model consensus sits above the current price for 50.2%.

That last contrast is the study's most striking external comparison: sell-side consensus numbers imply that four out of five US stocks are worth more than their current price, while a mechanical 19-model blend lands at almost exactly half. Academic literature has long documented optimism bias in sell-side estimates; this dataset is consistent with that finding. It does not establish that either number is more accurate -- only that they disagree, and by how much.

The models vs. each other: a 9.8x spread on the median stock

A fact rarely shown by platforms that publish a single fair value number: the individual models disagree enormously before they are blended. Across the 2,938 stocks with at least three positive model estimates (the median stock has 13), we computed the ratio of each stock's highest individual model estimate to its lowest:

• The median spread is 9.8x -- the most generous model typically values a stock at nearly ten times the most conservative model's figure.
• The interquartile range runs from 5.9x to 17.9x. 94.8% of stocks have at least a 3x spread.
• Disagreement is widest in Automotive (15.8x), Hardware (13.8x), Specialty Retail (13.6x), Software (13.2x), and Real Estate (13.1x).
• It is narrowest in Banks (4.4x), diversified Financials (6.2x), Insurance (6.8x), Utilities (7.2x), and Consumer Staples (7.9x).

This is why single-model valuations -- a lone DCF, a lone P/E comparison -- should always be treated with suspicion, and why the platform blends methods, trims outliers, and weights by data quality before publishing a consensus estimate. The raw spread above is measured before that trimming, so it overstates the disagreement that survives into the final blended number. But the direction of the lesson stands: any single-model fair value is one draw from a very wide distribution.

Methodology

Universe and date. The study covers the 3,139 US-listed stocks in Equity Rank's screener universe as of July 1, 2026. 3,091 of them (98.5%) had both a live market price and a model consensus fair value and form the analysis set. Every covered stock was revalued between June 26 and July 1, 2026. Sub-universe sizes vary slightly by statistic: 2,980 stocks had a published sell-side consensus estimate, and 2,938 had at least three positive individual model estimates for the dispersion analysis.

The 19 valuation methodologies. Each stock is scored by up to 19 methodologies spanning five families: intrinsic-value models (discounted cash flow, dividend discount, residual income, economic value added, earnings power value); trailing multiples (P/E, P/B, justified P/B, P/S, EV/EBITDA, PEG, Graham Number, CAPE-adjusted earnings); forward multiples (forward P/E, forward P/S, forward EV/EBITDA, forward price-to-free-cash-flow); and sector-specialist models (FFO and AFFO-based models for REITs, net asset value for miners, EV/EBITDAX for energy producers). Including sector-specific variants, 22 distinct model output columns were present in the dataset; not every method applies to every company (a dividend discount model needs dividends; a Graham Number needs positive earnings and book value), so the median stock carries 13 populated estimates.

Model consensus fair value is the platform's blended output: individual model estimates are sanity-checked, outlier-trimmed, and weighted by sector fit and data quality to produce one number per stock. Margin of safety is computed as (model consensus fair value minus current price) divided by current price, times 100. A margin of safety of +25% means the model consensus fair value is 25% higher than the price; -25% means the price is 25% above the model estimate.

Computation. All statistics were computed directly in SQL against the production valuation database on July 1, 2026 -- counts, medians (percentile_cont), and distribution buckets were aggregated in-database over the full covered universe, never from a sample. Sector cuts use the platform's internal sector classification, with sub-sector labels folded into parent sectors; sector tables include only sectors with at least 40 covered stocks. Headline figures were re-computed through two independently written queries as a consistency check.

Limitations -- read these before quoting the numbers.

1. Model estimates are not predictions. A model fair value is the output of explicit assumptions applied to reported financials. It says nothing about where a price goes next, or when, or whether the gap ever closes. Stocks can trade below model estimates for years for good reasons the models cannot see (litigation, dying end markets, governance) -- the classic value-trap problem.
2. A snapshot, not a time series. Every figure describes one date. Distributions move with prices and with quarterly fundamental updates.
3. Models inherit their inputs. Reported financials, consensus earnings estimates, and sector multiple baselines all carry error, and several methods share inputs, so their errors are correlated.
4. The size gradient has competing explanations (scale premium the models miss vs. genuine broad-market concentration in the largest names), and this data cannot distinguish between them.
5. Coverage skews to liquid US names. The universe excludes most OTC and very small securities, so universe-wide statements describe roughly the investable US market, not every listed share.
6. The analyst comparison uses third-party published consensus figures as reported by our data providers; it says nothing about any individual analyst or firm.

FAQ

How many stocks were analyzed, and when? 3,091 US-listed stocks with complete model coverage, out of a 3,139-stock screener universe, as of July 1, 2026. Every stock in the analysis set was revalued by the platform within the preceding five days.

What does it mean when a stock trades below model fair value? It means the blended output of up to 19 valuation methodologies, under their stated assumptions, is higher than the current market price. That reading corresponds to potential undervaluation under those assumptions -- and nothing more. It is a description of a model output, not a prediction of future prices and not a suggestion to act.

Does a large margin of safety mean a stock's price is going to rise? No. A margin of safety measures the distance between a price and a model estimate today. The gap can close by the price moving, by the fundamentals (and therefore the estimate) moving, or it can persist indefinitely. Nothing in this study measures forward returns.

Why do the largest companies trade so far above model fair value? The data shows the pattern (median mega-cap: 25.3% above model consensus fair value; only 7 of 68 below it) but cannot explain it. Candidate explanations include a market premium for scale, liquidity, and index inclusion that fundamentals-based models do not capture, and the possibility that models anchored to reported financials underestimate mega-cap economics. Both could be partly true.

Why do the 19 models disagree with each other so much? Because they measure different things: a DCF prices projected cash flows, a P/B multiple prices the balance sheet, a Graham Number prices a conservative earnings-and-assets floor. For companies whose value lives mostly in the future (growth software, pre-profit biotech), asset-anchored and cash-flow-anchored models diverge most -- which is exactly what the sector dispersion table shows. The blended consensus exists precisely because no single lens is reliable on its own.

---

This content is for educational and informational purposes only and does not constitute investment advice. Equity Rank is not a registered investment adviser. All figures are model outputs computed under stated assumptions as of July 1, 2026; model fair value estimates are not predictions of future prices and may differ materially from realized outcomes. Statistics describe sector-level and universe-level aggregates, not any individual security. Always conduct independent research and consider consulting a qualified financial professional before making investment decisions.