Inside the workstation, not beside it
Not a chat in another window: it sits next to the chart, sees your panels and works on the data in front of you.

One workstation, with a chatbot that builds strategies, indicators and backtests with you. Six plugins extend it, and automated trading has a section of its own.
In the browser, nothing to install. Born on crypto, built for any market that produces a tick.
Diagram of the workstation: the symbol bar on top, the tool column on the left, in the centre a candlestick chart with bubbles for the trades and the volume profile along the right edge, below it the band of net flow per bar. On the right three panels: book and depth, agents, computation engine. At the bottom the status bar with the window, the step and the seed of the series.
Diagram of the workstation — not a screenshot of the product
It lives inside the workstation: describe a strategy or an indicator in words, and it writes the code, charts it and backtests it. The code stays yours.
Not a chat in another window: it sits next to the chart, sees your panels and works on the data in front of you.

Strategies, indicators and tools described in words become code you can read, change and run again.

What it writes goes straight to a backtest, bar by bar, with window, step and seed declared.

Ask what is happening on a symbol: the answer carries the measurements behind it, and can be reproduced.

Filters and rankings across hundreds of symbols, including ones you build by asking in words.

The scripts you already have on TradingView come along; TradeStation and MultiCharts are phases on the roadmap.

The models are never ours: you connect your own keys across providers and pay them — no meter, no markup. Several models reasoning on one question is in the works.
Features land one by one, and each one states where it stands.
Optimised tools, built on the research we publish: reading the chart stays yours. They sit on the base, and each one you add changes what the others tell you.
The volume traded at every price, bar by bar.
Small diagram of volume by price level: the value area band crosses the figure, on the right the volume histogram by level with the most traded level highlighted, on the left the candles of the series.
Buying and selling at each bar price.
Small diagram of trading inside the bar: three bars side by side, and in each one the two sides of the trade mirror around the price column; imbalanced levels are outlined, the diamonds in the margin mark forced closures, and under each bar there is its balance.
Where liquidity waits, and where trades hit it.
Small diagram of resting liquidity: a heat map with time across and price down, the order book in a column on the right, the price line crossing the map and three diamonds where trades meet it.
Length, phase and validity of every cycle, in time.
Small diagram of cycle decomposition: the price line above, below two waves of different period around a zero line, with the duration ticks and the projected stretch beyond the vertical line that closes the measurement.
Systems as versioned modules, with optimisation when you need it.
Small diagram of the strategy park: on the left the blocks of a strategy wired together — two conditions, the combinator, the action and the risk constraint — on the right four rolling test windows, each split between the part the system was tuned on and the part it has never seen.
Positioning read from the options flow. A measurement, never an instruction.
Small diagram of the options chain: on the left two wings of horizontal bars around the strike column, with the central strike highlighted; on the right the implied volatility curve by strike, shaped like a smile.
A plugin on Elliott waves is also in preparation: it joins when it is finished.
The figures on the cards are declared diagrams: real screenshots arrive when the platform opens.
Execution on your own venue, with your own API keys. Funds stay on your exchange: we never hold them. The AI proposes and measures; you authorise.
They read the market and write their reasoning in the clear. You decide how much autonomy they get, and no strategy goes live on its own.

Maximum position, maximum loss per strategy and per portfolio, VaR, order frequency, cooldown after an event. You write them, and the engine measures them.

Every strategy has a version, a status and a history. Hyperparameters are declared, and the future is never granted.

Dedicated sub-accounts on your own broker, each with its own strategies and limits. Not there yet: it is a declared direction.
Deterministic rules before and after. What the model proposes goes through a statistical gate declared in advance, and what fails is written down with the reason.
Funds stay on your exchange: we never hold or move them. You authorise, and you can revoke.
Five label-and-value rows describing one pass of the chain: inbound rules passed, three hypotheses raised by the model, only one of them through the statistical gate, one module left as a candidate, and automatic promotion declared as not provided.
The discarded hypotheses stay in the ledger, with the reason for the discard.
Diagram of the control chain — not a screenshot of the product
The procedure is not on display. It lives in the volumes of the series, in the training paths and in the papers, and each one is paid for.
Volumes that carry the method in full, with runnable notebooks that redo the sums page by page.




8 top-level recorded paths, and every one opens the platform with the modules of its own domain. Hands-on work is watched live, in the masterclass.
Every paper carries the data and the code that redo it from scratch. The record with abstract, DOI and BibTeX stays public; the full text is not free.
The platform opens with an account, in the browser. Meanwhile you can join the waiting list, and the book’s preview notebooks open without an account.
The six modules