What twenty years of filings actually show
We test our own signals in public, count each company once, grade against the right benchmark, and publish the misses and the corrections next to the hits. Every piece links to the backtest behind it and to the stocks it points at today.
What works
Tested on the full record, with the sector breakdown.
Dividend yield: a modest, real edge — until it gets extreme
Ranking payers by their declared yield at each annual report, higher yield beat lower yield over the next year. Above roughly 8% the market is usually right that the dividend is at risk.
Price-to-book still sorts, across all sizes
The oldest value signal there is. Ranking non-financials by price-to-book each year, the cheapest fifth beat the most expensive fifth over the following twelve months in most years, with th…
Cheap small caps on book value: the widest spread in our data
Among non-financial companies under $2 billion, the fifth cheapest on price-to-book at each annual report beat the most expensive fifth by a wide margin over the next year — and by more in …
When an insider raises their stake by 30% or more during a sell-off
The size of the commitment matters. Buys that raised the insider's existing holding by at least 30% are the strongest single signal we have found, and the Industrials cell was pre-registere…
Insider buying pays in sell-offs, not in calm markets
Copying every opportunistic insider purchase barely beats small caps when the market is calm. The same trades beat the market clearly when it is being sold off.
Cluster buys: three insiders, one week
Three or more insiders buying in the same ten-day window is the most quoted insider signal. It works, but mainly in stress, and the bank version of it works only in a panic.
Officer buys versus director buys
Named executives (CEO, CFO, president) buying their own stock carry a small, consistent edge in every market state. Directors carry more in stress and less in calm.
Ranking companies as equity bonds: the top fifth beats the bottom fifth by a wide margin
Treat each company as a bond whose coupon is its owner earnings, growing at a capped rate, and rank the universe by the ten-year return that implies. The top fifth beat the bottom fifth by …
The equity-bond lens works for steady compounders, barely for growers
Split by growth type, the ranking sorts steady names cleanly and moderate growers well. For fast growers the ten-year return leans on the growth assumption, and the lens loses most of its p…
What does not work
Negative results, published with the same care.
Net-nets: the ratio does not sort, and the strict rule is a lottery
Ranking companies by net current asset value over price barely separates winners from losers. The strict Graham rule (price below net current assets) did return a high average, but on fewer…
Big dollar buys are the weakest insider signal we track
A $250,000-plus purchase sounds decisive. Across 9,000 of them it produced no edge in calm markets and a modest one in stress. Dollar size measures the insider's wealth, not their informati…
A first-ever insider purchase is not a signal in calm markets
We pre-registered the idea that a brand-new insider position in retail during a calm market would be informative. It was not. First purchases only matter when the market is already frighten…
Insider buying does not tell you a company will survive
Across the 20-year corpus, 93.4% of companies that later went bankrupt had a qualifying opportunistic insider buy in their final twelve months. So did the survivors. Insider purchases say n…
A 20%+ owner-earnings yield is the market disagreeing with the coupon
The richest-looking rows in the equity-bond table are almost never bargains. A yield above roughly 20% means the market expects the cash flow to shrink: a commodity peak, a business in run-…
Corrections
Where our own numbers were wrong, what caused it, and what changed.
Correction: a low beta made pipeline equity look barely riskier than Treasuries
Our cost of equity came from beta. A beta of 0.56 produced a 7.6% discount rate for leveraged midstream equity. Fixed 2026-09-08: the floor is now 8.5%, the ten-year Treasury plus a four-po…
Correction: we overvalued holding companies by up to six times
Plains GP Holdings consolidates all of Plains All American but its shareholders own a minority of it. We divided the whole enterprise's cash flow by the parent's share count and printed a $…
Correction: filers that tag only aggregate debt read as nearly debt-free
Air Products stopped tagging a separate long-term debt line in 2023 and tags only the aggregate with current maturities. We read the current portion as total debt: $716 million against $17.…
Evidence pages
The full tables behind the findings.
Evidence: every screen on the site, compared on one yardstick
Top fifth minus bottom fifth, twelve months ahead, versus the S&P 500 and small caps, with the market-state split, for every factor we screen on. The Screen Lab lets you test any other comb…
Evidence: the owner-earnings backtest in full
Every table behind the equity-bond ranking: quintiles, growth types, sectors and years, with the method and its limits.