The API demo

What the front page is showing you, and why it is built the way it is.

BTCX-USD is a synthetic instrument. The series the demo agent trades was generated, not observed. It is not a market, its prices are not quotes, its future days are published in advance, and nothing on this page is a forecast or a performance claim. Everything you need to recompute the demo yourself is linked below.

What you are looking at

A rules-based strategy, running continuously inside the sandbox against a published synthetic price series called BTCX-USD, on 15-minute bars. Every bar, signal, order and fill is recorded as an event in the run's journal; the front page shows the equity curve and the latest events, and this page shows more of the journal with the price series behind it.

The strategy is an ordinary trend-follower, deliberately simple so its behavior is readable. It wins some windows and loses others. That is the point of showing it rather than a screenshot: the demo is a verifiable execution, not an advertisement.

loading the strategy

The viewer

BTCX-USD, a synthetic instrument · the published window

loading

The chart shows the same bars the rules saw. A green arrow is a buy, a red arrow a sell; the two lines are the moving averages the rules compare. Where nothing is marked, no rule fired. The days ahead exist in the dataset but are not drawn: the series is revealed bar by bar, as a market would be.

Time (UTC)KindWhat happened

How the series is made

The generator is a log-price random walk with slowly switching drift and volatility regimes, intraday wicks, and a U-shaped daily volume profile. It trends, it chops, it draws down. Its identity, seed and calendar anchor are published with every dataset download:

generator seed genesis (day 0) bar size1 minute · 1,440 bars per day window

Deterministic, append-only

Days are generated in order from genesis, so each day opens exactly where the previous one closed and the same seed produces the same bytes on any machine. A day is generated once and never revised. The published window is two days back and two days ahead; at midnight UTC the oldest day leaves the window and one more future day joins it. Nothing is regenerated: the window is a view onto one continuous stream, and every past day stays available from the archive.

Anchored to reality, loosely

A synthetic Bitcoin sitting at $5,000 while the real one trades at $80,000 would be a distraction, so each newly generated day carries an anchor: a recent real Bitcoin daily close, rounded to the nearest 100, recorded at the moment the day is generated. The walk is pulled gently toward that level (about a one-day half-life). The level tracks reality; the path is invented. Anchors are recorded once, never changed, and published so the series can be reproduced exactly.

DayDate (UTC)Anchor level

Verify it yourself

  1. Download the current window (JSON: generator, seed, genesis, anchors, and every bar as [open_ms, open, high, low, close, volume]). Any past day: /v1/demo/dataset?day=N.
  2. Read the strategy above. It is the exact document the engine executes; the demo endpoint returns its English rendering.
  3. Fetch the narrated journal: /v1/demo/events. Every signal, order and fill carries the bar time it happened at.
  4. Reproduce: feed the bars to the same rules, the engine is deterministic, so the fills must match. The generator is ~200 lines of dependency-free Rust (crate synth); given seed and anchors it reproduces any day.

What the demo is not

Real markets

Every signed-in user runs against real prices from a market-data account they hold: a free key from Alpaca, Twelve Data or Finnhub, connected on the account page. The sandbox streams as you, for your runs only. That is why the API demo cannot show a real quote: no one has licensed it to, and we would rather show you a synthetic series and say so than show you a real one and stay quiet.

Sign in and connect your own key

See also the disclaimer and the terms. This page describes the demo as deployed; the dataset endpoint is the authority on what the series actually is.