Reading a Recommendation Card
Every trade IronGate proposes arrives as one card. After this lesson you'll read one the way a pilot reads instruments — knowing where every number comes from and what would make you say no.
3.1Meet the card
This is a real example of what the recommendation engine produces (from your own system's documentation). The numbered badges are our tour stops:
GBPUSD is showing the strongest trend signal today (ADX 34.2) with low spread (1.1 pips). USD weakness after yesterday's soft CPI print is creating a sustained bid for GBP. The pair is in its London session — its highest-liquidity window…
…three more paragraphs, one for each remaining choice.
3.2Stops ① – ⑤ · the decision
Pair. Not chosen at random — every configured pair was scored this cycle and GBPUSD won. The scorer checks: is there a trend worth trading ()? Is volatility healthy ()? Is the cheap right now? Is the pair in its busy ? Does news help or hurt? Are you already exposed to a correlated pair?
Strategy. First the classifier answers "what kind of market is this — trending, ranging, or chaotic?" Then a regime→strategy map picks the tool that suits it. A trending market gets a trend-follower like MA Crossover; a ranging one gets a mean-reverter. Same strategy in the wrong regime = losses; this stop exists to prevent exactly that.
Timeframe. Chosen from evidence, not habit: test results per timeframe are compared, and the one with the best out-of-sample Sharpe for this pair+strategy wins. Here H1 beat H4 and D1.
Direction. The strategy's actual signal — for MA Crossover, the fast EMA crossing above the slow one says BUY (you know from Lesson 1 that SELL would be a perfectly normal trade too).
Score. The pair scorer's 0–100 verdict on how good conditions are right now. Treat it as a confidence dial, not a promise: an 87 means many independent checks lined up. Low-scoring opportunities never even reach a card.
3.3Stop ⑥ · the risk block — you can now derive it yourself
Nothing in the risk line is arbitrary. It chains together everything from Lessons 1 and 2, starting from a single measurement of current volatility:
⚙ Build the risk block yourself
Raise the ATR (a wilder market) and watch the stop widen and the lot shrink — risk in dollars never changes. That's the whole risk stack cooperating.
3.4Stop ⑦ · the four "why" paragraphs
The numbers above are computed; the prose is generated from them. IronGate hands the scores and signals to an LLM, which writes the four explanations — why this pair, this strategy, this timeframe, these risk settings. If no AI key is configured (or the call fails), the system falls back to shorter rule-based explanations rather than staying silent.
How to read them critically: every claim in a "why" paragraph should trace back to a number you can see — "strongest trend today" ↔ ADX 34.2, "low spread" ↔ 1.1 pips, "highest-liquidity window" ↔ London session. If a claim has no number behind it, that's a question to raise, not a reason to trust. The card is evidence, and you're the judge.
3.5The whole pipeline in one picture
3.6Your job at the gate
Recommendations wait in the and expire automatically if you do nothing — a stale signal is a wrong signal, so inaction is always safe. When you do act, you have four moves: accept it, reject it, modify it (say, halve the lot size while you're learning), or let it time out.
Before accepting any card, answer three questions out loud: Do the numbers back the story? (every "why" claim has a figure) · Is the R:R at least 1:2? · Would I survive being wrong? (risk ≤ 1%). Three yeses or no trade. That ritual — not any indicator — is what will make you a trader.