Where does a mortgage payment go? Why can a computer get a decimal sum wrong? What does an AdSense application actually require? This site takes questions about technology, money and the internet and works through the answers, with examples you can follow and sources you can check.
That is the whole product. There is no course, no newsletter funnel, no signals group and no account to create. If you want to know what a 4% stop does to your position size, the answer is one page away and you owe nothing for it.
This page exists because a site giving people numbers about their money should be legible: who makes it, how, with what tools, paid for by what, and where its limits are.
What is here
The articles are the substance. Each one shows the mechanism properly — how the thing actually works, a worked example with every number written out, and an honest section on what it leaves out. Where a piece is built on inputs you might want to change, the figures in it are computed from a panel on the page, so you can read the argument about your own numbers rather than mine.
The tools are for the jobs prose cannot do: a check you point at something and read an answer off. There are few of them, and there is meant to be. None requires an account.
How a page is actually made
Articles are files in a version-controlled repository. Each one is a text file with structured metadata at the top. There is no database in the path between writing a sentence and it appearing: the site is compiled to static HTML ahead of time, so a page you load has already been built and the server is doing nothing but handing it to you. Every change to every article is a diff with a date attached.
The numbers come from one shared library, not a script per page. It lives in a single package of pure functions — no framework, no network access, no clock, no randomness. Given the same inputs they return the same outputs, every time, which is what makes them testable. Each one has a suite of cases with hand-checked values, boundary and zero inputs, and property tests asserting the identities that must hold no matter what is entered — that proceeds minus cost basis really does equal the reported gain, that an amortisation schedule really does end at zero.
That same library backs a public API, so the numbers on this site and the numbers any future app shows come from one implementation. Two implementations of the same formula eventually disagree, and the disagreement is always discovered by a reader rather than by the author.
The interface is deliberately plain. Black, white and grey; no web fonts; no UI framework; no animation; hand-written CSS. Light and dark both work from the first line of code, and there is a visible toggle. This is not minimalism as a style statement — it is what makes pages open quickly on a phone on a bad connection, which is how most people will arrive.
What is measured, and what is not
Two counters, both aggregate: how many times an article was read on a given day, and how many times a tool was run on a given day. That is the entire analytics surface.
There is no record of who you are. The database has no table for people, deliberately — not an oversight to be filled in later, a decision that makes the question “what personal data do you hold about me?” answerable with “none”. No IP address is stored against a view, no user agent, no session, no fingerprint, no cross-site identifier.
No cookie is set to measure readers. The only browser storage this site uses for itself is a single entry remembering whether you chose light or dark.
Advertising is the exception, and it deserves a straight answer. This site is intended to be funded by display advertising, because the alternatives are a paywall or sponsored content, and both of those quietly change what gets written. When advertising is switched on, the advertising code is third-party code and it sets its own storage. Where the law requires consent — the EEA, the UK and Switzerland among others — it is asked for first through a certified consent tool, refusing is as easy as accepting, and refusing does not take anything away from the page. Ad slots have their space reserved in the layout so nothing jumps under your thumb, and no ad is placed where it can be mistaken for a result or a control.
Where AI is used, and where it is not
Being specific about this is both a policy commitment and, for content about financial matters published in the EU, a legal one.
Used for: research and source-gathering; drafting and restructuring prose; writing and reviewing the code behind the site and its tools; checking a result against an independent implementation.
Not used for: deciding what is true. Every article published here is read line by line by a person before it goes out, every worked example is recomputed by hand or against the tool’s own test suite, and every claim that could not be verified is either cut or explicitly marked as unverified.
The rate limit is the point. The failure mode that regulators and search engines both target is volume without review — many pages, no judgement, nobody accountable. So publishing here is capped at what can genuinely be reviewed, which is a small number of pieces, slowly. If this site ever starts producing forty articles a week, something has gone wrong with it and you should trust it less.
Articles that were drafted with AI assistance say so. A page carrying no such note was written without it.
What this site will not do
- Predict prices. Not in an article, not in a tool, not in a headline. Nothing here forecasts what anything will be worth.
- Promise a return. No “guaranteed”, no “risk-free”, no back-tested strategy presented as an expectation. Every projection follows from assumptions you supplied, and it inherits every flaw in those assumptions.
- Give personalised advice. Nothing here knows your circumstances, and reading it creates no advisory relationship. It is general information and education.
- Take payment for coverage without saying so. If a link is an affiliate link, it is labelled before you click it. If a piece is sponsored, it says so at the top.
- Interrupt you. No pop-up asking for an email, no interstitial over the article, no auto-playing anything, no ad wedged into the middle of a sentence.
- Collect what it does not need. The privacy position above is not a maximum — it is the design.
When something here is wrong
Some of it will be. A figure gets transcribed with a sign error, an assumption goes unstated, a fee model turns out not to match how a venue actually charges.
Corrections are made in place, the update date on the page changes, and anything substantive gets a line saying what changed and when. Nothing is quietly rewritten to look as though it was always right. If you find an error — especially in a number — tell me and include the inputs you used, because that is what makes it reproducible.
Where to start
If you have never sized a trade against a stop, read how to size a position so a loss costs what you decided: it is one division, and the version most people use understates the loss.
If you are considering providing liquidity, impermanent loss is a ratio problem explains the one number that decides the outcome and the fee yield a position needs to be worth holding.
Everything on this site is general information and education. It is not financial, investment, tax or legal advice, no advisory relationship is created by reading it, and no content here is an inducement to buy or sell anything. Crypto assets are volatile and leveraged products can lose more than you put in. Where a decision matters, take advice from someone regulated to give it in your jurisdiction.