Where Tarn
stands

A tracker answers one question: what is it all worth right now. CoinStats and Delta answer it at scale, on the phone, for millions of people, and that scale sets their shape. Your holdings are mirrored into an account on their servers. The features are metered by subscription. The surface is a feed, built to be opened between other things, glanced at, and closed.

What a tracker cannot hold is the reason. There is no thesis on a position, no route drawn against it, no plan the position answers to, and no record of the decision once it closes. The number moves all day, and none of the movement is a decision. That is not a flaw in the trackers; it is the category. They are built for holders who want a balance, not for traders who work.

Tarn starts where the feed stops. It is local-first: it reads the venues itself, from your machine, with read-only keys in the macOS Keychain, no account on anyone’s server, no cloud mirror of your wealth, no subscription meter. Each venue is read in its own register, settlement verbatim, never collapsed into one vague number. And the number is the beginning of the job, not the end of it: every position is drawn against its route, carries the thesis you took it on, and lands on the record when it closes. A tracker shows you a balance. An instrument holds you to a plan.

Rotki is open source, local first, and built for accounting and tax reporting. It is strong on cost basis and history across many chains. It is not a terminal for live positions, and Hyperliquid’s builder dexes are outside its scope. Tarn covers the ground Rotki does not: live positions from each venue’s socket, Hyperliquid builder dex (HIP-3) assets rendered natively with live prices, per dex margin kept independent, and a journaling surface built around the route and the R multiple.

How Tarn reads each venue ↗

A journal answers the harder question: are you actually any good. Edgewonk and TraderSync take that question seriously, and both are built around the same motion: you trade somewhere else, then bring the trades in for review. Imports, tags, reports, on a subscription, in the browser. They were shaped by stocks, futures and forex, where every broker speaks a clean export. Crypto is supported the way a guest is supported.

The gap is time. A journal filled after the fact records what you can still remember, and the most important thing about a trade, the reason you took it, is the first thing to go. Discipline that lives in a separate tool becomes homework: the upload you owe, the tags you backfill, the review you skip in a losing week, which is exactly the week the review was for.

Tarn keeps the journal at the moment of the decision. The thesis is written when the position is taken. The route is drawn while it runs. The verdict, held or broke, is recorded at the close, next to the outcome, part of the same record by construction. Reflection is not a second app you owe an upload to; it is the last elevation of the same climb. Your live positions and your record finally agree, because they are the same data.

Walk the instrument, plan to record ↗

You could build it: a sheet per venue, an API key, a script that pulls balances on an interval, a tab for the journal you will definitely fill in. Every serious trader has built some version of it, because the alternative was trusting someone else’s dashboard, and you knew better.

Then a venue changes an endpoint, and the script breaks quietly. The sheet does not stream; it polls when you run it, shows the world as of the last pull, and never admits which figures have gone stale, which is the polite way of saying it guesses. It records what you type into it and nothing else, so the journal tab dies first. Maintaining the rig is a second job that pays in the feeling of control rather than the thing itself.

Tarn is that rig, built once, properly, by someone who refused the compromises. Streamed instead of polled. Stale marked instead of guessed. The thesis demanded at entry instead of promised for later. Settlement read verbatim from each venue instead of approximated. Native instead of a grid of cells. The honesty of the spreadsheet, with a standard behind it, and none of the maintenance.

What happens if Tarn is gone in two years ↗

Put plainly: Tarn does the tracker’s job, the journal’s job, and the daily practice neither of them holds, in one native macOS instrument. No tracker keeps your journal. No journal reads your venues live. Neither keeps your keys and your record on your own machine, and the Mac, when it is served at all, is served by a port or a browser tab.

If you want one of those jobs alone, the tools above do that job well, at scale, for a broad audience. Tarn is not for a broad audience. It is for the trader who wants the whole discipline in one place: the plan before the position, the route while it runs, the honest record after, read straight from the venues and owned outright. One instrument, one permit, and nothing else to assemble.