undamentalytics Newsletter
Sign in

Priced in

A share price is a forecast. We run our valuation models backwards to read it, then set it against what the same models expect from the company's own filings. Where the two disagree, one of them is wrong; this page shows where that happens today and how often the model was right.

Star Group, L.p.: the price implies cash-flow growth of -12.0% a year; our model expects 3.2%.

What the price implies What our model expects

Lens

What growth in cash flow is the price paying for?

The growth in owner earnings (free cash flow after stock-based pay) that would justify the share price in our discounted-cash-flow model, against the growth the model itself assumes from the company's record.

Biggest gaps today readings at September 30, 2026

Companies whose price implies less than our model expects
CompanyPrice implies ○ vs our model ●GapTrack record
SM SM Energy Co -25.9% vs 13.0% +38.9 pts Beat the S&P 500 in 7 of 17 quarters held
COE 51talk Online Education Group -20.8% vs 13.0% +33.8 pts Beat the S&P 500 in 2 of 9 quarters held
GRNT Granite Ridge Resources, Inc. -21.3% vs 9.6% +30.9 pts Beat the S&P 500 in 3 of 11 quarters held
F Ford Motor Company -28.3% vs 2.5% +30.7 pts Beat the S&P 500 in 12 of 24 quarters held
CHRD Chord Energy Corp -12.8% vs 17.8% +30.6 pts Beat the S&P 500 in 5 of 14 quarters held
PERI Perion Network Ltd. -25.6% vs 4.0% +29.6 pts Beat the S&P 500 in 18 of 41 quarters held
PR Permian Resources Corp -13.7% vs 13.0% +26.6 pts Beat the S&P 500 in 15 of 25 quarters held
USNA Usana Health Sciences Inc -24.4% vs 2.0% +26.4 pts Beat the S&P 500 in 11 of 26 quarters held
MNR Mach Natural Resources LP -12.8% vs 13.0% +25.8 pts Beat the S&P 500 in 0 of 5 quarters held
RGP Resources Connection, Inc. -21.0% vs 4.0% +25.0 pts Beat the S&P 500 in 1 of 16 quarters held
NOG Northern OIL & GAS, Inc. -24.3% vs 0.5% +24.8 pts Beat the S&P 500 in 18 of 46 quarters held
CALM Cal-maine Foods Inc -20.5% vs 4.0% +24.5 pts Beat the S&P 500 in 16 of 27 quarters held
SHOE Shoe Station Group Inc -21.9% vs 2.0% +23.9 pts Beat the S&P 500 in 9 of 13 quarters held
PRDO Perdoceo Education Corp -14.9% vs 8.7% +23.6 pts Beat the S&P 500 in 12 of 35 quarters held
VIPS Vipshop Holdings Ltd -21.5% vs 2.0% +23.5 pts Beat the S&P 500 in 8 of 18 quarters held

Only companies with a reading for the latest quarter are listed. A company whose latest filings are too old for this lens has no current reading and is left off, not shown with an old one.

Has trading the gap paid?

Since June 30, 2010, buying the companies the market was too pessimistic about returned +4.0% a year against the S&P 500 (no reliable evidence). Buying those and shorting the ones it was too optimistic about earned +2.6% a year, and was ahead in 61.5% of quarters: No reliable evidence. The companies bought moved 1.33 times as much as the S&P 500 (beta); after allowing for that, and for the cash rate, their return (alpha) was +0.2% a year (no reliable evidence), so much of their lead over the index came from market exposure. The long-short line's beta was 0.07 (t 0.9): not reliably different from zero. Its alpha was +2.0% a year (no reliable evidence).

$0$50.0k$100k$150k 201020122014201620182020202220242026 $164k$98.4k
StrategyS&P 500, buy and hold $10,000 invested in June 30, 2010, before costs, over the same 65 quarters as the strategy table below. Strategy: an equal-weight portfolio of every company the market was too pessimistic about, rebalanced each quarter. Benchmark: the S&P 500, bought and held. For reference, an equal weight of every company this lens read would have reached $107,781, and the market-neutral long-short spread $15,132.

Is it skill or luck?

In plain English: It passes 1 of 4 checks (the best pairs kept working in later years) and fails 3 (beats the market after allowing for its risk; works within each sector; clearly positive in both halves of the period). The evidence is mixed: treat it as a pointer for research, not a strategy.

Four questions a sceptic would ask, each at the bar this page uses for strong evidence (a t-statistic of 2 or more).

  • Beats the market after allowing for its riskAlpha of buy and short (t 0.77) or of matched pairs (t 1.78) at t ≥ 2No
  • Works within each sectorMost pessimistic fifth minus most optimistic fifth inside each sector, so no industry bet (t 1.70)No
  • Clearly positive in both halves of the periodBuy and short at t ≥ 2 in 2010–2017 (t 1.31) and in 2018–2026 (t 0.78)No
  • The best pairs kept working in later yearsChosen on 2010–2017, checked on 2018–2026 (t 2.73)Yes
Does a bigger gap mean a better return? Average return against the S&P 500 a quarter, with companies split into five equal groups by gap
+0.2% 1 −0.1% 2 +0.1% 3 +0.3% 4 +1.1% 5

Group 1: the market most optimistic against our model; group 5: most pessimistic. The steps are not in order: the gap does not rank returns cleanly. Top minus bottom group: +3.0% a year (t 1.53).

Return and risk of each strategy

StrategyReturn a yearTypical yearly swingSharpe ratioBiggest dropMoves with market (beta)
Buying onlyOwn the companies the market is too pessimistic about, equal weight, rebalanced each quarter. +18.8% 22.9% 0.81 -37.3% 1.31
Buy and short*Also bet against the companies it is too optimistic about. Market rises and falls largely cancel out. +2.6% 9.3% 0.32 -23.8% 0.07
Matched pairs*Each buy paired with a short in the same sector with a similar beta, so market and sector moves cancel. +3.3% 7.6% 0.46 -18.5% -0.02
S&P 500, buy and holdThe yardstick. +15.1% 14.7% 0.95 -23.9% 1.00

Buying only, buy and short and the S&P 500 cover the same 65 quarters; matched pairs cover 64 (the quarters with at least three matched pairs). Before trading costs. Return is compounded; swing is volatility; biggest drop is the worst fall from a previous high. Sharpe ratio is return above cash divided by the swing: buying only and the S&P 500 subtract the 3-month Treasury bill rate (averaging 1.5% a year over the period, from FRED). *The long-short strategies already pay for their buys with their shorts, so their whole return counts.

Why is the Sharpe ratio of buy and short lower than the market's?

Its return is only the gap between the cheap and dear companies, +2.6% a year, with none of the market's own rise; its Sharpe ratio was 0.32. The S&P 500 earned +15.1% a year over the same quarters, a Sharpe ratio of 0.95, high by its own history: these years were a strong stretch for the index. Averaging all five lenses into one score (long the top fifth, short the bottom fifth each quarter, a different rule from the thresholds here) earned +0.1% a year with a Sharpe ratio of 0.07 over 65 quarters, no better than this lens's buy and short. Buying only has the higher Sharpe ratio largely because it rides the market: its beta of 1.31 means it moved more than the index.

Buy and short replaced 28.7% of its holdings each quarter; a trading cost above 65 basis points a trade would have wiped out its gain.

Full statistics, for the detail-minded

A threshold backtest can flatter a signal. These tests ask the same readings different questions: does a bigger gap earn more, step by step; is the result skill or market exposure; does it survive within sectors, across periods and over longer holds?

1 strong 3 some 9 no reliable evidence across 13 tests of this lens.

TestResultt-statisticEvidenceWhat it checks
Extra return after removing market moves (alpha) +2.0% a year, beta 0.07 0.77 No reliable evidence Return not explained by moving with the market.
Extra return vs an equal mix of all companies +0.9% a year, beta 0.13 0.39 No reliable evidence Also removes the small-company effect.
Buying only: extra return after market moves +0.2% a year, beta 1.33 0.05 No reliable evidence Buying only: how much is skill, not market exposure.
Do bigger gaps rank next quarter's returns? (rank correlation) 0.008 on average, positive in 53.8% of quarters 0.79 No reliable evidence Across all companies, not just the extremes.
Works within each sector +2.7% a year 1.70 Some evidence Top fifth minus bottom fifth of the gap inside each sector.
Worked in 2010–2017 +2.7% a year 1.31 No reliable evidence Long-short, first half.
Worked in 2018–2026 +2.5% a year 0.78 No reliable evidence Long-short, second half.
Holding for six months +3.0% a year 1.23 No reliable evidence Does the gap close slowly? Six-month holds overlap, so the t-statistic uses non-overlapping periods only (about 32).
Stricter rules (a gap of 15 points or more) +4.0% a year 1.72 Some evidence Fewer, more extreme gaps than the standard rule (a gap of 10 points or more).
Looser rules (a gap of 5 points or more) +2.2% a year 1.32 No reliable evidence More companies, smaller gaps.
Holding for a year +4.9% a year 1.44 No reliable evidence Year-long holds overlap, so the t-statistic uses non-overlapping years only (about 15).
Pairs matched by sector and market sensitivity +3.3% a year; alpha +3.9%, beta -0.02 1.78 Some evidence Each long paired with a short in the same sector whose beta (two years of weekly prices) is within 0.2; about 117 pairs a quarter.
Best pairs from 2010–2017, tested on 2018–2026 +1.7% a quarter vs +0.2% for all pairs 2.73 Strong evidence 35 of 842 pairs reached t ≥ 2 in 2010–2017, against about 27 expected by chance alone; 33 of 49 chosen pairs stayed positive.

Five lenses were tested; with that many tries a t-statistic near 2 can arise by chance. A Bonferroni-adjusted bar is about 2.6. Every figure on this page uses the same sample: readings at the edge of the solver's range and quarters with fewer than 25 readings are left out.

Alpha
Return left over after allowing for how much a portfolio simply moves with the market. Positive alpha is the part that could be skill.
Beta
How much a portfolio moves when the market moves. 1 means in step with the S&P 500; 0 means unaffected by it.
t-statistic
How far a result stands out from random noise. Around 2 or more is usually treated as meaningful; below 1.5 it could easily be chance.
Sharpe ratio
Return above cash per unit of ups and downs, over a year. Higher is better. Professional funds aim for above 1; the S&P 500 has been about 0.4–0.5 over the long run, higher in 2010–2026.
Worst fall
The biggest drop from a previous high before recovering.
Long-short
Owning the companies the market is too pessimistic about and betting against the ones it is too optimistic about, so market rises and falls largely cancel out.

All five lenses compared

The same three strategies, run on each lens, all from one sample (the evidence column is the t-statistic of the buy-and-short return in the same quarters). Return a year and Sharpe ratio (return per unit of swing; above 1 is what professional funds aim for, the S&P 500 managed 0.95 over the period).

LensBuying onlyBuy and shortMatched pairsHoldings a quarter
Return a yearSharpeReturn a yearSharpeEvidenceReturn a yearSharpe
Cash flow +18.8%0.81 +2.6%0.32 No reliable evidence t = 1.30 +3.3%0.46 175 long, 246 short117 pairs
Earnings +15.8%0.76 +2.4%0.32 No reliable evidence t = 1.27 +3.0%0.40 351 long, 211 short134 pairs
Growth vs P/E +13.5%0.68 +3.0%0.48 Some evidence t = 1.76 +1.8%0.33 250 long, 163 short122 pairs
Sales +18.6%0.87 +0.3%0.08 No reliable evidence t = 0.34 +2.5%0.29 151 long, 545 short128 pairs
Book value +16.2%0.74 −0.3%0.00 No reliable evidence t = 0.02 −0.1%0.02 151 long, 954 short139 pairs
S&P 500, buy and hold+15.1%0.95The yardstick, same quarters.

Returns compounded, before trading costs. Buying only is the total return of the companies bought, not the gap over the S&P 500. Sharpe for buying only and the S&P 500 subtracts the Treasury bill rate; buy and short and matched pairs pay for themselves, so nothing is subtracted. Results with stricter and looser buy rules are in each lens's full statistics.

How the test works, and its limits
  • Quarters already reported, figures as later restated. At each quarter end since March 31, 2010 the models are rebuilt from quarters ending at least 90 days earlier (and no more than 200 days before that, so an old quarter is never used as if it were current). The figures are the latest filed for those quarters, including any later restatement, not necessarily what was first reported.
  • Prices. Valuations use each day's close, adjusted for later share splits and spin-offs so that price and share count are on the same basis; returns use total return, dividends included.
  • Rules fixed in advance. A company is bought when our model is at least 10 points above the market's growth (5 points for margin and return on equity; a PEG under 0.75) and shorted on the reverse (a PEG over 3), for one quarter. Stricter and looser settings are shown so no single setting is cherry-picked.
  • Earnings lens. Net income is what is left for shareholders after interest, so it is valued at the required return on equity and only cash is added: debt is not subtracted a second time.
  • Evidence. "Strong" means the average result is at least twice its own noise (a t-statistic of 2 or more); "some" between 1.5 and 2; below that the result could be chance. "Evidence: worse" means the result was reliably below zero.
  • Data checks. 275 companies are left out where the traded and dividend-adjusted prices disagree (a split recorded in one only) or history is too short. Companies whose published valuation is under review are kept in the backtest but left off today's lists.
  • Limits. Today's list of companies (firms that failed or were taken over are missing, which flatters the longs), no trading or borrowing costs. A backtest describes the past under stated rules; it is not a forecast or investment advice.