console — a live market, running the real engine in your browser

loading the engine (WebAssembly)…

DEMO mid— spread— last— step0 digest—

seed

Trade as “you”

engine.Process(order)

Your resting orders

engine.OpenOrdersFor("you")

Nothing resting. A limit order away from the touch will sit in the book — marked ● on the ladder.

Tape

MatchResult.Trades

No prints yet.

Depth ladder

engine.Snapshot(10)

Mid price

Snapshot.Mid —

Order-flow imbalance

signals.OFI.Observe(snap) —

contemporaneous, not predictive — the study

Volume delta (CVD)

signals.CVD.Observe(trades) —

Depth imbalance · top 5

signals.DepthImbalance(snap, 5) —

−1 · ask-heavy0bid-heavy · +1

Kyle's λ

signals.EstimateLambda(flow, ΔP)

—ticks / 1.0 qty warming up

Surveillance

surveillance.Monitor.Observe(event) 0 alerts

Quiet — the detector is watching every event this page generates. Provoke it: Run a spoof places layered size and pulls it, and the shipping SpoofDetector names the account here.

What this is. The real engine — pkg/matching compiled to WebAssembly — matching a market of sim.NoiseTrader agents, with every panel computed by the shipping library code and titled by the exact call that produced it. Same seed, same market, every time: that is the engine's determinism, demonstrated rather than claimed. What it is not. A trading signal. The noise market contains none by construction, and the OFI study found a ~540× R² gap between contemporaneous and predictive even in richer flow. Latency numbers are deliberately absent: browser-WASM timings would be noise presented as measurement — the honest ones live in BENCHMARKS.md. Adapt it: go get github.com/intrepidkarthi/orderbook — the console's whole market is ~40 lines against the public API (cmd/obwasm).