An independent finance site by Luis Diaz Busquets, a finance student at Bryant University with minors in Applied AI and Law.
Most research hides its assumptions. Here, every number links back to where it came from and every model shows its inputs, so you can disagree with a conclusion and see exactly where.
The site sits alongside the main site, which covers who I am and my experience, and LDB Lab, which is about machine learning.
What it sells, who pays for it and why customers stay.
Revenue, margins and cash flow, with the quality of earnings checked.
What it's worth under bull, base and bear cases, and what the price already assumes.
What would change the view, and the dates to watch.
Every public company gets a Buy, Hold or Sell from the same rule, whoever owns it. The rule starts from the report’s expected value (below) and the price at writing.
Because the value depends on the discount rate, the same assumptions are first applied to the whole S&P 500: a 5.15% 10-year Treasury yield and a 4.5% equity risk premium. On those assumptions the index is worth about 68% of its price, so part of every stock’s gap is the method, not the stock. The rating compares each stock with the market on that same basis:
Relative, not absolute. A Buy means cheaper than the market on the same assumptions, not necessarily cheaper than its own value. So each rating also carries an absolute label, based on the expected value against the price: undervalued (value at least 10% above the price), near fair value (within 10%) or overvalued (value at least 10% below). A stock can be a relative Buy and still overvalued on its own, when the whole market is.
Ranges, not three cases. Three cases weighted 25/50/25 is a convention, not a measured probability. So each report’s own model is rerun 4,000 times with every input drawn from a range: the base case is the middle of each range, and the bear and bull cases are its 10th and 90th percentiles, so each side can be a different width. Growth, margins, multiples and capital spending share a common factor (correlation 0.6), because they tend to go well or badly together; the discount rate varies on its own. The expected value is the average of the 4,000 results, and it is what the rating uses. Each report also shows the likely range (10th to 90th percentile), the chance its value beats the price, and the chance it is cheaper than the market. The draws are seeded, so the numbers only change when a model does. The code is montecarlo.py.
Each report shows both numbers, the absolute value and the comparison with the market, and says whether the rating would change if the index grew 6% or 10% a year instead of 8%. Private companies are not rated, since there is no price to buy them at; the Costco report is university coursework and is left as written.
| Index growth, years 1–5 | Value / price | Implied equity premium |
|---|---|---|
| 6% | 61% | 1.9% |
| 8% | 68% | 2.4% |
| 10% | 76% | 3.0% |
420 S&P 500 companies with complete data, $50.0 trillion of market value and $1,654 billion of free cash flow over the last twelve months. Growth fades to 3% by year ten. Rebuilt with _source/models/market.py; the ratings use the 8% row, as of October 22, 2026.
Everything on this site is for education. It is not investment advice, and nothing here is a recommendation to buy or sell any security.
Every build checks that each report agrees with its own model and with itself. That cannot catch a figure copied wrong from a filing, so reports are also checked by hand against the primary source: the company’s own release or SEC filing, never a news summary. Corrections are listed here, not made silently.
| Date | Report | Checked | Primary source | Result |
|---|---|---|---|---|
| Oct 23, 2026 | Nvidia | Revenue and growth, data-center revenue, gross margin, operating income, Q3 guidance and its China assumption, shareholder returns, buyback authorization (10 figures) | Q2 FY2027 release (8-K) | All match |
| Oct 23, 2026 | Amazon | Revenue, AWS sales, growth and profit, advertising growth, operating income, net income and the Anthropic gain, Q3 guidance, cash, debt, free cash flow and capital spending (13 figures) | Q2 2026 release (8-K) | All match. Two clarifications: the two free-cash-flow figures now name their definitions, and the $66 billion rise in capital spending is now labelled net of proceeds |
| Oct 23, 2026 | JPMorgan Chase | Net income and EPS with and without securities losses, revenue, ROTCE, CET1, book and tangible book value, NII, markets, equities, IB fees, expenses, charge-offs, dividends, buybacks (16 figures) | Q3 2026 release (8-K) | All match. NII guidance comes from the earnings call, which the release does not cover |
| Oct 23, 2026 | L3Harris | Revenue and growth, orders, book-to-bill, backlog, segment margin, Missile Solutions growth, 2026 guidance, debt, cash, dividend (12 figures) | Q2 2026 release (8-K), dividend release | One correction: the dividend is $1.25 a quarter ($5.00 a year, raised in January), not $4.90 a year. It does not enter the valuation |
| Oct 23, 2026 | ExxonMobil | Earnings, EPS, free cash flow, dividends and buybacks, production, Permian and Guyana figures, Q3 guide, share count (11 figures) | Q2 2026 release, Q3 considerations (8-K) | All match |
| Oct 23, 2026 | Alphabet | Revenue and growth, operating income and margin, Search, YouTube and Cloud revenue, Cloud profit, net income, capital spending, free cash flow, the June stock and bond sales, the at-the-market program, cash and debt (17 figures) | Q2 2026 release (8-K), 10-Q | One correction: net cash was about $100 billion, which counted the SpaceX stake twice. Since the June IPO the stake sits in marketable securities (non-marketable securities total only $131.5 billion, less than the two stakes), so net cash is about $50 billion before the stakes. Value $249 → $245; rating unchanged (Hold). One clarification: the $49.6 billion raised was $30.5 billion of common stock and $19.1 billion of mandatory convertible preferred |
| Oct 23, 2026 | Meta | Revenue and growth, cost growth, the legal charge, operating income and margin, EPS with and without the charge, daily active people, ad impressions and price, capex, operating and free cash flow, buybacks, cash, debt, Q4 and 2026 guidance (16 figures) | Q3 2026 release (8-K) | All match. One clarification: January 19, 2027 is when the charter change allowing the Class C dividend takes effect; the record and payment dates are not yet set |
| Oct 23, 2026 | Microsoft | Revenue, operating income and growth, the three segments, Azure growth, commercial backlog, EPS and the Anthropic gain, shareholder returns (12 figures) | Q4 FY2026 release | All match. Returns of $10.2 billion are as the release states; the cash flow statement shows $11.3 billion, a difference the release does not explain |
| Oct 23, 2026 | Broadcom | Revenue and growth, AI and software revenue, free cash flow, Q4 revenue and AI guidance, cash, debt, debt repaid, dividend (12 figures) | Q3 FY2026 release (8-K) | All match. The gross-margin and 2027–28 AI figures come from the earnings call |
| Oct 23, 2026 | Visa | Net revenue, adjusted EPS, payments and cross-border volume, transactions, buybacks, dividend, cash, debt, diluted shares (11 figures) | Fiscal Q3 2026 release (8-K) | All match |
| Oct 23, 2026 | UnitedHealth | Revenue, earnings from operations, adjusted EPS, medical care ratio, reserve releases, Optum Health profit, 2026 guidance, debt, cash, debt-to-capital, buybacks and dividends (13 figures) | Q3 2026 release (8-K) | All match |
| Oct 23, 2026 | Johnson & Johnson | Sales and growth, Innovative Medicine and MedTech growth, the Stelara drag, reported and adjusted EPS and the talc charge, 2026 sales and EPS guidance (9 figures) | Q3 2026 release (8-K) | All match |
| Oct 23, 2026 | RTX | Sales and organic growth, adjusted and reported EPS, the $0.25 one-off, free cash flow, backlog, segment margins, 2026 sales, EPS and cash guidance (14 figures) | Q3 2026 release (8-K) | All match. One clarification: the $0.25 is what RTX calls IEEPA tariff cost recovery, not a refund |
| Oct 23, 2026 | Lockheed Martin | Sales and growth, EPS, free cash flow, backlog, Missiles and Aeronautics growth, the PAC-3 award in backlog, 2026 guidance (11 figures) | Q3 2026 release (8-K) | All match. The $58.6 billion ceiling of the PAC-3 agreement is not in the release; it comes from the award announcement |
| Oct 23, 2026 | General Dynamics | Revenue and growth, EPS, free cash flow, orders, book-to-bill, backlog, Gulfstream deliveries and margin (8 figures) | Q2 2026 release (Exhibit 99.1) | All match. Guidance comes from the earnings call, which the release does not cover |
| Oct 23, 2026 | Inflation note | Which measure the Fed’s 2% target uses | Federal Reserve statement on longer-run goals | Corrected: the target is headline PCE, not core PCE |
185 figures in 15 reports checked so far; two errors found and corrected (one changed a value, neither changed a rating), five figures clarified. The sample grows as reports are revised after earnings.
Every published version of every report is also kept, frozen as it was published, so any past call can be checked against what it actually said. When a report is revised, its earlier versions are linked at the foot of its summary.