Would you trust this market signal?
Freeze a real market sample. Introduce a fault. See which decisions survive.
A signal is only useful when your team can explain why it deserves attention.
Signal Wind Tunnel
A controlled experiment on a copy of observed public trades. Compare an unguarded rule with a quality-aware decision using the exact same frozen window.
Freeze a window to capture source, timestamp and observed trades.
The same window. Three views.
Inspect the checks and experiment limits
Missing-event counts and coverage are knowable here because this experiment controls the injected fault and retains the original sample. They do not measure completeness of the live exchange feed. This is a deterministic local replay of sampled public data, not a production reliability certification or trading recommendation. Price shocks are synthetic.
Start by freezing a market window. No experiment runs until you choose to run it.
Market observatory
The source context behind your experiment · independently refreshing
Bitcoin BTC / USD
Waiting for a public ticker
Fetching real market candles…
Activity, observed and expected
Completed Flink window · sampled trades
No Flink signal received. Forecast values appear only after the pipeline publishes them.
Attention monitor
Rules for the selected market
Surface feed interruptions and sharp observed price changes before they become customer support escalations.
Latest observed trades
Deduplicated by product and trade ID
| Time (UTC) | Taker side | Price (USD) | Size |
|---|---|---|---|
| Waiting for recent trades | |||
Taker side is the opposite of Coinbase’s reported maker side.
Follow the data
Visible provenance from source to screen
This dashboard reads the public API directly. Cloud streaming and forecasting are not claimed as active.