3 rules in 30 seconds
- Stocks: does today's cash yield, plus per-share growth, clear the 10-year Treasury plus 3 points?
- Moonshots: how much future growth is already in the price?
- Bitcoin: holder cost, miner revenue, and size versus gold, each against its own last four years.
How the score is built
The short version
A score, not a published fair value.
Stocks use cash flow and the balance sheet. Moonshots ask whether the growth is already in the price.
Bitcoin has no earnings. Ether is scored as a commodity, not on fees or staking. Other coins are scored only on cash that reaches holders.
Basemark scores a stock by the yearly return in its cash flow, not by a published target. A stock has two numbers. The value score compares that return with the 10-year Treasury plus 3 points. The cycle reading is the chart. They are not added. Neither is a price target.
This month
The 10-year close is still loading.
The base bar is the 10-year plus 3 points. Cash flow per share that swings (a 10-year coefficient of variation above 0.5), or revenue per share that swings by more than 12% a year, adds 1 point, and only once. Debt above 1.5 times equity adds 0.75. A current ratio under 1 adds 0.75, except for banks. Steady companies with high returns and a decade of positive free cash flow get 0.5 points back.
Why growth is now in the hurdle
A company that grows its cash flow per share is worth more than one that doesn't, even at the same yield today. So we estimate a yearly return in two parts: the yield in the formula, plus how fast cash flow per share has grown. That yield is labeled owner earnings on the chip. It is the median of trailing free cash flow, the 3-year average, and the forward earnings yield, not the raw trailing free-cash-flow figure beside it. Growth comes from the company's own reports over 5 and 10 years, never from the share price. It is capped at 9% and floored at −2%, and it cannot run more than 2 points above sales per share. A flat 9% is a ceiling, not a starter we plug in. Above the economy's long-run pace (about 4.4%) it fades to that pace within ten years. VOO uses the same fade on the 5-year S&P earnings and dividend pace. The high end of its range is the trailing earnings yield, not a forward estimate. We compare that return with the 10-year Treasury plus 3 points, a little more for cyclical or indebted companies and a little less for steady, high-quality ones. We only call a stock cheap with a margin of safety when it beats that bar by 1.5 points or more. The price chart is still shown separately, and it never changes the value score. Tested on today's large companies since 2016, adding growth made the rule less wrong, but it still didn't beat the market. Treat it as a screen, not a prediction.
Why the oil line and the through-cycle yield
A cyclical company is scored on a through-cycle cash yield: up to ten years of free cash flow per share (earnings for a bank), restated in today's dollars, averaged, and divided by today's price. Oil and gas producers are scored on the cash flow that the last decade of results imply at a mid-cycle oil price, the 10-year average WTI in today's dollars, not at today's spot price. A boom year is not the whole yield. The long-run average is. If oil explains little of the cash flow, the card says so and stays on the through-cycle yield. Midstream names that are spending heavily are scored on the distribution, not on free cash flow that is being reinvested. Autos with a captive finance arm are not scored on consolidated free cash flow.
Digital-asset treasuries
Strategy, BitMine, and SharpLink are not scored on owner earnings. The value mark is mNAV: the stock's market cap divided by the bitcoin or ether it holds. 1× pays a dollar for a dollar of coins. Under 1× is cheaper than the coins. A 40% discount or premium reaches the end of that part of the meter. Thirty percent of the mark is where that coin sits against its own 200-day average. Below the line is cheaper, and above it is richer. The stock's own chart stays the cycle mark and is not added.
Value weights
The value score is −20 per point of gap through 3 points, then 5 per point, so ±100 takes about an 11-point gap rather than 5. The chip is the yield in that formula, labeled owner earnings. Banks and insurers use the median of trailing and forward earnings yield, times one minus growth divided by return on equity. VOO, SCHD, QQQ, and VXUS show a range: dividend yield plus per-share growth, up to trailing earnings yield plus that growth. VOO's growth is the 5-year S&P pace and fades the same way a stock does. The other three use the fund's own dividends per share, capped at 9%. The score uses the midpoint, and the fund is expensive only when the top of the range is still below the bar. Missing growth is not filled in from the share price. The S&P page leads with names at or below −20. The halfway-back rule is only on the sale list. Capping financials and energy at two names is a filter, not the valuation.
Cycle weights, kept apart
Ordinary: 200-day 3, one-year versus VOO 7, five-year versus VOO 8. Compounder: 200-day 6, 52-week range 3, one-year 7, five-year 7. A fund: 200-day 15. A chart that already ran does not make the cash yield cheap.
Caps on the value mark
Each growth measure is kept between −10% and +30%, and the median is kept between −2% and +9%. The score itself stops at −100 and +100. The cycle mark is not added, even when the tape has already run.
Commodity energy
Oil, gas, refining, and the services around them. Pipelines are not this rule. The cash yield is the lower of trailing free cash flow and the average of the annual years the feed actually reports, up to five. A boom year is not the whole yield. A thin year is not lifted. Fewer than three annual years and the card says so and uses trailing. We do not invent a year, and we do not scale free cash flow by an oil price. If the cycle mark is already above +50, the value mark cannot print below zero. The tape already paid that year. Cycle is still not added.
Expected return
Today's cash yield is next year's cash. If per-share growth is above the economy's long-run pace, it fades to that pace by year 10, and the terminal value is the following year's cash divided by the return minus that pace. If growth is already slower, we do not lift it. Not a forecast.
Ten-year record
Names at or below −30 on the yield rule, and names at or above +30, rebuilt each quarter from 2016, equal weight, versus VOO. Banks are separate. Price is the split-adjusted close. Total reinvests dividends. In a quarter with fewer than five names, the strategy held the market. The line under the chart says how many quarters that was, and how many quarters actually held picks. It also shows undervalued and overvalued results for 6-month and 12-month holds, and how often each beat the market. Today's large companies only, so this is survivorship bias. GAAP earnings include one-off items. The windows overlap. There is no balance-sheet or quality factor in the book. So far the zero-growth rule has not beaten the S&P 500. Treat scores as a screen, not a prediction.
Zero-growth rule: cheapest fifth minus dearest fifth −11.8 points a year (12-month holds). Growth-adjusted rule: −10.3 points. Neither beat the market. Today's large companies only (survivorship bias).
Rebuilding the quarters from filings…
Moonshots
Not cash flow. Sales versus peers, dilution, runway, and how much of the move versus the S&P is already in the price. Pre-revenue names cannot look cheap on a sales multiple they do not have.
Bitcoin
Bitcoin has no earnings, so we don't pretend it does. We ask three questions, each compared with bitcoin's own last four years (about one halving cycle):
- Holder cost (MVRV): is the price high or low compared with what holders paid on average? Weight 50%.
- Miner revenue (Puell multiple): are miners earning unusually much or little right now? Weight 30%.
- Size versus gold: is bitcoin's value, as a share of all gold, above or below its recent norm? Weight 20%.
A score of −100 means cheaper than almost any day in the last four years. +100 means dearer than almost any day. The 200-day trend is shown separately as the cycle mark. We tested it, and a strong trend has tended to keep going, so it's not a value signal. We don't use stock-to-flow, because the supply schedule is fixed and says nothing about demand.
Ether
Ether is a commodity, so fees and staking yield are not a mark against it. The value score is holder cost versus its own last four years (46.7%), value versus bitcoin (20%), and how much stablecoin value it secures (33.3%). Price versus yearly fees and the real staking yield (staking reward minus new supply, versus the 10-year inflation-protected Treasury) are shown on the card. They are not in the score.
Other coins
The list is the 25 largest after memecoins and stablecoins are removed, plus wrapped and bridged tokens, liquid staking, and tokenized funds. Dogecoin, Shiba Inu, Pepe, Bonk, dogwifhat, Floki, and Official Trump are out, as are USDT, USDC, USDE, DAI, and FDUSD. Ethena's protocol token is not a stablecoin and can stay. Rank is not raw market-cap rank. Bitcoin and ether keep their own models and show a value meter and a cycle meter. For the other 23, the score is half fundamentals and half relative strength and trend, each ranked against that group. A high number means rich or strong, not cheap. Fundamentals use holder revenue (DefiLlama dailyHoldersRevenue only; revenue is used when a chain burns every fee), fees, supply growth, unlocks, and, for smart-contract chains, value secured. Relative strength is the last year versus bitcoin, ether, and the top-25 median, plus the 200-day and 50-week lines. A low score means 'cheap vs other alts', not 'cheap vs Bitcoin'. Coins outside the 25 stay on a page that says they are not scored. A coin with no holder cash keeps an unscored value meter, and the cycle meter still shows. Figures are dated and the source is named on the card.
How well has this worked?
Since 2014, months when bitcoin scored in the cheapest fifth were followed by a median 12-month gain of +94%. The dearest fifth was followed by −11%. For ether since 2018: +70% versus −33%. That's only about three bitcoin cycles and two ether cycles, so the evidence is thin. Cheap has often got cheaper first. This is a screen, not a prediction.
Metals and bonds
Position versus the trend. Not a value score. Gold, silver, oil, copper, and long Treasuries have no cash-flow cheap.