When someone says "AI trading signal", most people picture a tip with a confidence score. A useful signal is something more boring and more valuable: the output of a chain of checks that turns raw market data into a risk-bounded, fully specified trade idea.
This article walks through the general stages of that chain. It describes how signal systems work as a category, not SuperKamouBot's strategies, parameters or scores, which are not published.
Stage 1: raw market data (OHLCV)
Every signal starts with data. For crypto futures the core data is OHLCV: open, high, low, close and volume for each candle on each market.
Data quality matters
Garbage in, garbage out. Before any analysis, data should be checked for:
- Gaps: missing candles from exchange downtime or API issues
- Stale data: candles that have stopped updating
- Outliers: impossible prices from API glitches
A well-built system rejects stale or gapped data before it enters the pipeline.
Stage 2: technical indicators
From clean data the system computes a set of measurements. Common families are:
- Momentum: is price accelerating or decelerating?
- Trend: is price trending, and how strongly?
- Volatility: how much is price moving? This drives stop placement.
- Structure: where are recent highs, lows and levels?
No single indicator generates a signal. They are inputs to the next stage.
Stage 3: scoring and ranking
The system combines indicators into a view of "how good does this setup look right now?", often weighting them differently depending on the market regime: trending, ranging or volatile. The output is a ranking, not a yes or no. A ranking says not just that one market looks good but that it looks better than the alternatives, which matters when only a few ideas can be acted on.
Stage 4: risk filters
Hard filters then reject any candidate that would break a rule that must never break. Typical examples are:
- Risk per trade: would this idea risk more than the allowed share of the account?
- Total exposure: would it push overall exposure past its limit?
- Concentration: is the account already exposed to the same move?
- Regime match: does the strategy fit current conditions?
These filters are non-negotiable by design. Many bot projects make them advisory, which is how accounts get blown up. The specific limits are not published.
Stage 5: defining the trade
A signal is only useful if it is fully specified. A SuperKamouBot signal names the asset, the direction, the entry, the stop, a +1R target and an 8-hour window. The stop and target come with the signal, so you know what you are risking before you commit.
Stage 6: delivery and settlement
Signals are delivered to a subscriber dashboard. Afterwards the outcome is settled by what price actually did: target reached first, stop reached first, or expired or unclear. That record is what feeds aggregate evidence.
You place every order yourself on your own exchange account. SuperKamouBot does not custody funds, does not ask for exchange credentials and does not trade your account.
Where AI fits
A model can be used to rank or filter ideas, but it cannot predict price. For transparency: whatever produces a candidate, rules, statistics or a model, it has to pass the same shadow test before any signal reaches a subscriber.
The whole chain in one sentence
Clean data, then indicators, then ranking, then risk filters, then a fully defined trade, then delivery and honest settlement.
Every stage exists for a reason. Without ranking you take random ideas, without risk filters you eventually take a ruinous one, and without settlement you never learn whether any of it works.
See it in practice
The results page shows which result cohorts are published and why the others are not yet; losses count when they are. The how-it-works page explains the evidence lifecycle. See pricing for the one plan on sale.