signal_source is the disclosure field on every ATLAS Sovereign signal that tells you *how* the signal was generated: AI means it originated from the machine-learning order-flow model, while RULE means it originated from a deterministic, rule-based detection logic. It exists so that you always know whether a given XAUUSD order-flow signal came from a learned model or from fixed, human-defined conditions — no guessing, no blending of the two under a single opaque label.
Why We Label the Origin of Every Signal
Most signal services present every alert the same way: a direction, an entry, maybe a target. That presentation hides the most important question a skeptical trader should ask — *what actually produced this?* A machine-learning model and a hand-written rule behave very differently:
- A RULE signal follows a fixed, transparent condition. If the condition is met, it fires; if not, it doesn't. Identical inputs produce identical outputs, every time.
- An AI signal comes from a model that has learned patterns from historical order-flow data. It can adapt to structure a static rule cannot express, but its reasoning is statistical rather than mechanical.
Both approaches have legitimate strengths. Neither is a magic answer. What matters for your process is knowing which one you're looking at, because you evaluate them differently. A rule-based absorption print can be verified by eye against the tape. A model-flagged signal asks for a different kind of trust — trust in the training and validation process rather than in a single visible condition.
Hiding that distinction would be easy. Showing it costs us nothing and gives you everything you need to audit your own reliance on the service.
What AI vs RULE Looks Like in Practice
When you receive a signal, the signal_source field appears alongside the usual context — direction, instrument, and the order-flow observation that triggered it. A RULE tag means you can, in principle, reconstruct the exact market condition that fired the alert from the data in front of you. An AI tag means the model identified a pattern statistically associated with the setup, one that may not reduce to a single simple condition.
One important boundary on what these labels mean: ATLAS classifies observed lot-size tiers — small, medium, large, and whale-tier activity — based on order size alone. The labels never assert *who* is trading. A burst of whale-tier lot-size activity is a fact about order sizes on the tape, not a claim about exchange members, prop desks, or any other participant class. We hold that line in every signal, AI or RULE, because inferring identity from size is where most order-flow marketing drifts into fiction.
Why Honesty Here Matters More Than Any Metric
Order flow on CME gold futures is one of the most transparent markets available to a trader. The CME Group contract specifications are public: each GC contract covers 100 troy ounces, trades with a $0.10 minimum tick, and every tick is worth $10 per contract. Those are independently verifiable numbers — anyone can check them, and so should you. We apply the same standard to ourselves: the market-structure facts we cite are checkable, and the provenance of each signal is disclosed rather than implied.
A service that tells you *which engine* generated a signal is a service that expects you to think. One that flattens everything into a single stream is asking you not to. For traders migrating from spot XAUUSD to GC futures — where per-tick economics and depth-of-book behavior differ meaningfully from OTC spot pricing — that provenance matters even more, because the intuitions you built in spot don't transfer automatically.
How to Use the signal_source Field in Your Own Process
Treat it as a routing label, not a quality score. RULE signals invite tape-level verification — check the absorption, the imbalance, the tier shift yourself. AI signals invite pattern-level review — over time, note how model-flagged setups behave relative to your own read of the session. Many subscribers track the two sources separately precisely so their own statistics, not ours, drive their conviction.
That's the deeper point: transparency about signal origin isn't a feature we added for marketing. It's a prerequisite for anyone serious about building a repeatable process on top of order-flow signals.
If you want to work with signals that tell you exactly what they are — AI or RULE, whale-tier lot-size activity or plain imbalance, no participant-identity storytelling — ATLAS Sovereign publishes that context on every alert. Start the 14-day free trial and audit the signal_source field yourself, or go deeper on order-flow mechanics with live worked examples in the ATLAS Academy, available to Sniper and Sovereign subscribers.