Manifesto

Published August 2026

Trading tools have split into two camps: casinos that happen to show prices, and museums that happen to still run. Tikker is an attempt at a third thing: an instrument.

Why Tikker exists

Most retail platforms are engagement machines. They reward checking, not understanding. Confetti on order fills, streaks, notifications tuned like slot machines. The chart is a backdrop for the buy button.

Institutional tools have the opposite problem. The data is good, and the workflows calcified twenty years ago. They assume a desk, a license, and a tolerance for pain.

In both camps the data itself is suspect: resampled, aggregated, delayed, massaged by intermediaries you never chose. If the input is noise, everything downstream is noise with confidence.

And around all of it, the actual workplace is a zoo. One terminal for charts, another for the book, notebooks for research, alert bots in chat apps, scripts on a server that dies at night. Active traders end up working as system administrators of their own tooling.

The workplace today, and the shape Tikker replaces it with.

What Tikker is

Tikker is a programmable trading workspace for people who read markets for a living, or intend to.

The workspace is a canvas with three primitives on it. Widgets: chart, order book, ticket, risk panel, drawn over live data. Workflows: node chains over event streams that keep running with the browser closed, in the cloud or on your own runners. And the workspace itself: a serializable object you can save, share, fork and version like a file. All three are declarative configurations, not code, which is what makes the rest possible.

Three primitives and the agent above them; everything is a configuration.

The agent builds on those primitives through the same interfaces developers use. Describe an instrument and it comes back as a typed proposal: a readable diff you preview and apply, never a silent change. Its market analysis is grounded: it cites the data it reads, and when it has no source it makes no claim. Models are picked by capability tags, not vendor names; bring your own provider or a local model, and prompts stay inside your walls.

Charting stays precise under a cursor, not pretty in a screenshot: aggregation is first-class, because time is only one way to slice price. Data is ingested from venues directly and normalized in one pipeline we control. Hyperliquid perpetuals come first; other venues follow. When something is wrong with a feed, we want to be the ones who know first.

Automation is treated like a loaded instrument. Strategies start on a paper venue. Live runners are supervised. The right to execute is a lease that has to be renewed, and a kill-switch ends everything at once. If the control plane loses a runner, the runner stops trading. Silence never means consent.

What Tikker refuses to be

Some of what Tikker will not do, written down so we can be held to it.

  • No gamification. A terminal is not a game, and an order fill is not a celebration.
  • No engagement mechanics. How often you trade is your business, not our metric.
  • No fake proof. No simulated screenshots, no invented numbers, no purchased testimonials. This site has none, and that is deliberate.
  • No advice and no oracle. Tikker shows structure, the agent cites its sources; neither predicts prices nor picks trades for you.
  • No unsupervised automation. Nothing trades without a lease, and nothing outlives the kill-switch. That includes everything the agent builds.
  • No mystery data. Venues are named, normalization is deliberate, and gaps are shown as gaps.

Where it stands

Tikker is pre-launch. The charting engine, the ingestion pipeline, the paper venue and the agent's runtime exist and are being hardened into one workspace. Early access will open when using Tikker daily feels right to us, not when a growth calendar says so.

If this reads like restraint, that is the point.