PTQ is in early access — one plan, one price, and we build in the open.
PTQ.
How it works

From payment to armed, in order

Nine steps. None of them works until the one before it does. Including the parts that are not finished yet.

Every screenshot here is the real app. Illustrations are labelled as illustrations.

Before you start

What you need in front of you

Step 01

Buy

$997 per year, one plan, everything in it.

Checkout runs through Stripe. Card details never touch our servers. We get your name, email, what you bought and the amount.

The payment mints your key, not the sign-up. An account with nothing paid holds no key and no access.

jurisdiction checked, then payment — and the payment is what mints the key (illustration)

Step 02

Download and install the OS

The hosted platform opens in a browser. The OS is a copy you own.

Your account serves the archive and its checksum. Check it before unpacking. The build is not code-signed yet, so that check is the integrity story, and the notes say exactly what your computer will warn you about.

It checks your versions, builds its own Python environment and puts the ptq command on your path. Run it twice and the second time changes nothing.

your machine

# 1 — check the archive against its checksum

shasum -a 256 -c SHA256SUMS

# 2 — unpack and run the installer

./install.sh

[checks node + python, sets up the engine, links the ptq command]

# 3 — activate this device

ptq activate PTQ-…

[device leased · credit allowance attached]

the shape of the install, schematically — an illustration, not a captured session

Step 03

Activate this device

One command, one device, one signed lease.

Your device swaps the licence key for a short-lived lease, signed by a key that only exists on our server. Every request checks that signature. An edited lease fails.

The lease carries your credit allowance. Two device seats, counted on the server, not on the honour system.

the server signs the lease, your device verifies it, and it expires (illustration)

Step 04

Connect a broker — your keys, your machine

Optional, local-only, and refused if the venue does not recognise them.

Your keys go straight into your computer's own secret store. Keychain on macOS, DPAPI on Windows, Secret Service on Linux. They are never sent to us, and the website refuses any request carrying one.

They are tested against the broker before anything is saved. If the broker says no, nothing is written. Only your own machine can reach these routes.

the keys go into your machine's own store; the hosted side refuses them outright (illustration)

Step 05

Build or clone a strategy

Explicit rules, not code — and the risk engine reads the draft as you write it.

Start from a template or from nothing: when to buy, when to sell, how much, where you get out. What trades is still yours.

Break one of your own limits and the builder says so. A rule the engine cannot run is refused with the reason, never approximated.

Strategy builder — a real strategy draft, risk engine ruling live
Strategy builder — a real strategy draft, risk engine ruling live

Step 06

Let the gates try to kill it

A backtest on real bars, then the statistics that ask whether it means anything.

Your rules run against real market history, and every run is stamped with where its data came from. If the engine is down the run fails and says so.

Then the gate runs, hunting for luck. Most strategies fail. When one does you are told which test caught it, and by how much.

Validation gates — a real verdict — this run was blocked
Validation gates — a real verdict — this run was blocked

Step 07

Paper trade it

Real quotes, real hours, no hindsight — and every fill labelled simulated.

Whatever survives trades on paper. Real prices, fake money, every fill labelled simulated. Paper orders pass the same risk checks a live one would.

This part is slow on purpose. Weeks of paper is how a strategy proves itself without hindsight. The loop can run it on a schedule while you are at work.

Execution — the paper choke point — simulated labels on every route
Execution — the paper choke point — simulated labels on every route

Step 08

Arm it yourself

Local only, explicitly confirmed, scoped and expiring.

You arm it on your own machine, one strategy at a time, with a scope and an expiry you set. No web page can arm anything. Change your limits afterwards and the arm dies rather than quietly inheriting them.

Plainly: the live path is experimental, single-shot, and never tested against the live venue. Automated execution, multi-broker support and production execution safety do not exist yet. Unattended live trading is built but switched off pending a compliance answer, so an armed strategy runs paper.

the arm step — illustration, not a screenshot: live exists only on your machine, scoped and expiring

Step 09

Review what actually happened

A report over your own record that suggests and never adopts.

The review loop reads your own journal and answers four questions from real rows: what happened, why, what the gates said at the time, and what to try next.

Suggestions arrive as comparisons you can run yourself. Nothing is applied for you. If the engine is offline the report says so and invents nothing.

Journal — the record the review reads — a simulated fill, labelled
Journal — the record the review reads — a simulated fill, labelled
Where it stops you

The route has brakes, and they are the point

Every one of these is the product working. If you want a tool that never says no, this is the wrong purchase.

A field of horizontal lines displaced outward by a circle at the centre, every line clipped flat where it meets a heavy rule above and below.

Still deciding? The product page walks the same route from the maths side, the security page covers where your keys and your data sit, and the FAQ answers the rest.

PTQ

Start at step one

If a strategy dies at step six, that is money the gate just saved you.