Cerebras Systems (CBRS)

Value asks what you pay for each dollar of earnings, sales or cash the business produces. A high multiple isn't automatically bad - fast growers earn theirs - but a price far above the company's own history means the market expects a lot to go right. We compare each multiple to the company's own past and to its sector, never to a universal rule.

Growth here is what the company has actually reported to the SEC - not a forecast. We look at the last year, the three-year pace, whether growth is speeding up or slowing down, and whether it's been consistent rather than one lucky year. Hatched bars are fourth quarters we derived from annual filings (companies file a full-year 10-K rather than a Q4 report).

Quality asks whether the growth makes real money. Margins show how much of each sale survives costs; return on equity shows what shareholders earn on their capital; and the profit-to-cash comparison catches companies whose accounting profits never turn into actual cash.

Health asks one question: can the business survive a bad year? We check whether near-term bills are covered, whether debt is modest and shrinking, whether earnings comfortably pay the interest, and - for loss-makers - how many years of cash are left at the current burn rate.

A company can trumpet billions in buybacks while quietly issuing nearly as much stock to employees. What matters to you is the net effect: is the share count actually falling? If not, the 'return' is mostly recycling.

The trend read is the daily Ichimoku picture, the same indicator drawn on the chart above. Five names, and what each one is: the TENKAN is the midpoint of the last 9 sessions' high and low, the fast line; the KIJUN is the same over 26 sessions, the slow one; the two SENKOU spans are midpoints drawn 26 sessions into the FUTURE, and the band between them is the KUMO cloud - because it is drawn forward, the support for the coming weeks is already fixed and a change of its direction can be seen coming; the CHIKOU is today's close compared with the close 26 sessions back. A trend confirms itself in that order and the checks follow it: price clears the cloud, holds above both lines, the Tenkan crosses above the Kijun, the cloud ahead turns up, and the Chikou stands higher than the price a month ago. Six of six therefore means all of that at once; a price sitting inside the cloud, or below its own lines, cannot get there. The last check is the one Ichimoku cannot answer - whether the S&P 500 did better over 12 months (252 sessions), measured on the SPY ETF with dividends included, on the same adjusted basis as the stock's own price. Our figures are the last close we hold, refreshed every trading night; the chart is live. None of this says anything about the business, and it is not advice.

Most insider filings are not trades: stock grants, option exercises and tax withholding are how executives get paid, and gifts are estate planning. The signal lives in open-market transactions - a buy means an insider chose to spend their own cash on the stock. Sells are murkier: many are pre-scheduled 10b5-1 plans set months in advance. That's why the chart counts only open-market activity, and the table labels every row.

Fund managers with more than $100 million in US stocks report their holdings every quarter on Form 13F, up to 45 days after the quarter ends, so the fund figures are as of the last quarter end. A fund family reports as one manager, and funds together can report more than 100%: shares lent to short sellers are bought again by other funds, and both report them. Anyone holding more than 5% of a class files a Schedule 13D or 13G, and officers and directors report what they hold on Form 4 after every trade, so those rows carry their own dates. Short positions are not reported anywhere. A stake is the holder's shares as a share of all the company's shares, with other share classes converted at their market value.