The future of trading is one decided edge.
Signal Trader is a self-learning quant engine for the Micro E-mini S&P 500. A council of AI analysts mediates every tick into a single, auditable decision: entry, stop, target and the conviction behind it, with the risk rules already built in.
The live command center runs at signaltrader.cre8ivelabs.ai
The command center
Watch it think, in real time.
The desk scans, locks a signal, and projects entry, stop and target, with the forecast cone and depth ladder updating live. This is the actual cockpit operators run.
Inside the engine
Discipline you can audit.
Deterministic composite
One rule, everywhere. A fixed, auditable scoring layer whose logic is identical across every code path, and drift-tested so it can't quietly change on you.
Multi-timeframe confluence
5m, 15m, 30m and 1h have to agree. Signals pass sequential gates before a single entry is allowed.
Order-flow microstructure
Reads the order book itself: depth, imbalance and pressure at the L2 level, not just the last price.
An AI council, analysis-only
Four analysts and a governor weigh in with a conviction, entry, stop and target. It analyzes; it never places an order.
Risk governance
Hard gates the model can't override: position limits, dynamic stops, a kill switch and a circuit breaker on the whole desk.
A self-learning loop
The Ghost Engine replays shadow variants of every trade and missed entry. Changes are proposed only when the evidence is statistically real, and a human adopts them.
From tick to decision
How a trade is made.
- 01
Ingest
Live MES price and the L2 order book stream in from the NinjaTrader bridge.
- 02
Confluence
Multi-timeframe signals must line up through sequential gates before anything is considered.
- 03
Distill
The deterministic composite and the AI council each score both sides. One conviction emerges.
- 04
Gate
Risk governance sizes the trade or vetoes it. Nothing gets past the hard limits.
- 05
Execute
Entry, stop and target are mapped and monitored live on the execution graph.
- 06
Learn
The Ghost Engine replays counterfactuals; evidence-gated improvements are proposed for a human to adopt.
FAQ
Questions from the desk.
Signal Trader is decision-support software for Micro E-mini S&P 500 (MES) futures. It combines a fixed, auditable scoring layer with multi-timeframe confluence gates, order-book microstructure, and an analysis-only AI council that proposes an entry, stop and target with a conviction score. Hard risk limits sit outside the model. It is not investment advice.
No. Signal Trader is a decision-support and trade-execution tool for futures markets. It is not investment advice, not a recommendation to buy or sell, and not personalized to anyone's circumstances. Trading futures carries substantial risk of loss and is not suitable for everyone.
The AI council never does — it analyzes and proposes, and it does not place an order. Signal Trader is described as a decision-support and trade-execution tool, so an execution layer exists: entry, stop and target are mapped and monitored on the execution graph, and hard risk governance can size or veto a trade before anything reaches the market. What the engine cannot do is act on its own conviction without clearing those gates.
MES is a smaller-sized S&P 500 futures contract traded on the CME. Signal Trader is built specifically around it: live MES price and the Level 2 order book stream in through a NinjaTrader bridge, and every signal the engine produces is for that one instrument.
It means the scoring layer is fixed rather than learned at runtime: one rule, identical across every code path, and drift-tested so the logic cannot quietly change. That is what makes a decision auditable after the fact — you can reconstruct why a signal scored the way it did.
Four analysts and a governor each weigh in with a conviction, entry, stop and target, and the ensemble resolves to a single verdict. The council is analysis-only. Signals must also clear multi-timeframe confluence gates — 5-minute, 15-minute, 30-minute and 1-hour have to agree — before an entry is allowed.
Position limits, dynamic stops, a kill switch, and a circuit breaker on the whole desk. These are hard gates outside the model rather than inputs to it, so risk governance sizes a trade or vetoes it regardless of how convinced the council is.
A counterfactual learning loop. It replays shadow variants of every trade taken and every entry missed, and proposes a change only when the evidence is statistically real. A human reviews and adopts any change; the system does not rewrite its own rules unattended.
No. No performance record, win rate, backtest or return figure is published. The prices, conviction scores and consensus values in the interface mockups on this page are generated sample values shown for illustration, not a record of trades or of achievable results. Nothing here is a prediction, guarantee or promise of performance or profit.
Access is opening to a small group of operators by request rather than public sign-up. Tell us about your desk through the request form on this page. The live command center runs at signaltrader.cre8ivelabs.ai for operators who have access.
Trade with a system, not a hunch.
Signal Trader is opening to a small group of operators. Tell us about your desk and we’ll be in touch.
Signal Trader is a decision-support and trade-execution tool for futures markets. It is not investment advice, and nothing here is a prediction, guarantee, or promise of performance or profit. Trading futures carries substantial risk of loss and is not suitable for everyone.