Edition one · US-listed stocks and ETFs · end-of-day closes

Say the thesis out loud. buffet tells you what history did with it.

Backtesters want tickers, weights and a rebalance rule — which is the part you cannot do. A chatbot will invent an equity curve and sound certain about it. buffet takes the sentence, shows you how it read the sentence, then runs the arithmetic over real end-of-day price history.

The whole interface

Thesis

if I had put $10,000 into nuclear in 2015

write it the way you’d say it

Run the thesis

That is the entire product. One field, no account, no configuration. The first verdict is free and complete.

See a specimen verdict

A printed chart on almanac paper: one dark ink line rising unevenly, a flat grey line behind it. — Plate i — the thesis in ink, the index in grey. That is the whole argument.

§ 01

The problem

You hold four strong opinions and cannot check any of them.

Nuclear is going to run. Just buy the index. European defence is cheap. Sell in May. Every one of them is testable against history, and none of them has been tested. Here is what is available today.

Option one

A backtester built for quants

It wants tickers, allocation percentages and a rebalance rule. Translating your sentence into those is the hard part, and it is the part you were hoping for help with.

Option two

Asking a chatbot

It answers in seconds with a confident equity curve. It is not running anything against price history, so the number is fiction that looks like arithmetic.

Option three

Buying it and finding out

It drops by a third and you learn what you could have learned for nothing. This is the option most people take.

§ 02

What comes back

One page. One number. Green if it beat the S&P 500, red if it did not.

A thesis that returned +219.0% over 2015–2026 still lands red if the S&P 500 returned +246.0% over the same period. The colour is the judgment, not the sign of the return. That one decision is the whole of buffet.

Pick a thesis

Verdict — specimen

/v/nuclear

The reading

$10,000 into a basket of US-listed nuclear and uranium equities.

Equal weight, rebalanced yearly. Dividends reinvested.

2015–2026. Benchmark: S&P 500.

lost to the index+219.0% against +246.0% for the S&P 500, 2015–2026

$10,000 became $31,900 over 2015–2026. The same money in the S&P 500 became $34,600.

the thesisthe S&P 5002015–2026

What you sat through

Worst stretch: −58.2%, January 2016 to November 2016 — you would have sat through ten months of that.

Counter-test

Same rule, idle cash in T-bills: $32,400.

Where the thesis was wrong

The thesis was right about the direction and wrong about the entry. Almost all of the gain arrived after 2023. For the eight years before it, this basket lost to the index in every calendar year but one.

What this verdict does not include

US-listed stocks and ETFs only. End-of-day prices. No tax, no trading fees, no slippage. Survivorship is not corrected for. The figures in this specimen are illustrative; run your own thesis and every number on the page is computed from stored closes.

§ 03

How it works

Sentence to verdict in under two minutes.

You write the thesis

Plain English, the way you would say it in a thread. No tickers required, no percentages, no allocation table.

buffet reads it back

“$10,000 into SPY. Monthly contributions. 2005–2026. Dividends reinvested.” Every line is one click to change. Touch nothing and it runs.

It runs on real closes

Three lines with real counts, not a spinner: “Loaded 5,284 daily closes.” “Ran 21 years.” “Writing it up.” Twenty to forty-five seconds.

The verdict lands

The number, the chart with the index in grey, the worst stretch you would have sat through, and where the thesis was wrong even when it won.

§ 04

Where the number comes from

Everything here can compute a return. Only some of it is reading history.

What you type

Where its number comes from

Price

Portfolio Visualizer

What you type

Tickers, weights, a rebalance rule

Where its number comes from

Real price history

Price

$30–55 a month

Composer

What you type

A strategy flowchart, plus a brokerage account

Where its number comes from

Real price history

Price

$32 a month

QuantConnect

What you type

Python

Where its number comes from

Real price history

Price

Free

A chatbot

What you type

A sentence

Where its number comes from

Invented on the spot

Price

Free

buffet

What you type

A sentence

Where its number comes from

Our own database of end-of-day closes

Price

First verdict free, then $5

Prices as published by each service, August 2026

§ 05

The Library

A reference-book index page in warm paper, dense columns of entries separated by hairline rules. — Plate ii — the Library, as it is meant to end up.

Every verdict is a permanent page with its own URL.

Which means a verdict is something you can paste into a thread and let people argue with. Public by default, private in one click, and the URL keeps working either way.

A year of that is the point: the largest public library of ordinary investment beliefs, checked, and readable by anyone who googled the question.

Read the Library

§ 06

The line we do not cross

buffet is a history machine, not an adviser.

It never says buy or sell. Not once, not softened, not implied.

It never touches money, never connects to a broker, never asks what you own.

It never asks a model what the return was. A model is only asked what your sentence means.

It prints what it could not test on every verdict, next to the number.

Ask buffet what to buy and it answers once, plainly, and does not pretend to be coy: buffet only tests history. It cannot tell you what to buy, and it will not.

What it costs

The first verdict is free and complete.

No blur, no teaser, no trial. Every verdict after the first one is $5, bought one at a time. No subscription, no credits, no account, and no email before the number.

Run your first thesis

Nothing is collected. It is just the URL.

buffet.nanocorp.app · one verdict, $5, no subscription