/ai - AI market-structure read
Premium
/aigives you Claude's read of a coin on demand: regime, bias, the levels in play, scenarios and what would invalidate them. It is a labelled second opinion, not a deterministic signal - and it does not tell you what to trade.
What it does
/ai runs an on-demand AI analysis of a symbol on a chosen timeframe. It explains the
market structure: the current regime and bias with reasoning, which levels are active,
what has been confirmed and what has been broken, up to two scenarios each with the
condition that would confirm it and the invalidation that would break it, a structure-read
paragraph, and the main risks. It does not print a direction to trade, an entry, a stop or
targets - the same rule every alert follows. Every price and level it cites is checked
server-side against live data. Premium feature.
How to use
/ai SOLUSDT
/ai SOLUSDT 1h
- A ticker is required; timeframe is optional (default 15m). Intervals:
1m 5m 15m 1h 4h. - A short per-symbol/timeframe cooldown applies. Each run consumes one AI token (refunded if the analysis fails). Re-running on the same symbol does not cost extra quota.
What you see
A text analysis headed 📊 AI Analysis | #SOLUSDT | 15m, with:
- Regime and leaning - e.g.
📈 Regime: TRENDING UP | Leaning: 🟢 BULLISHwith the reasoning. "Leaning" describes where structure sits, not what to do about it. - 🧭 Structure read - the interpretation: where price sits relative to structure and what that arrangement means. When the AI sees no readable structure it says so.
- 📍 Levels in play - the levels the read is built on, each with a one-line note. The AI supplies a price; the bot names the level from its own volume profile, VWAP, swings, pivots and order-book walls. A price that matches nothing the engine computed is removed before you ever see it.
- ✅ Confirmed / ❌ Broken / 🔀 Contradicting data - the evidence split three ways, including the data that argues against the read.
- 👁 What the structure says next - up to two conditional observations ("acceptance above X keeps the range broken"), each with what would negate it. They are readings, never calls to act.
- Risks and a one-line takeaway.
- On a follow-up timeframe, a 🔄 Since the previous read line: what this timeframe adds, confirms or breaks relative to the read you already saw.
- A short caveat line when the AI quoted a price that matched no engine level, so a read with removed numbers is never shown as clean.
The buttons:
- 📊 1m / 5m / 15m / 1h - re-run the analysis on another timeframe; prior reads on the same symbol are fed in as context.
- 📷 Review - hand the bot a chart screenshot to analyze (see /review).
- ℹ️ Info - the symbol's raw stats.
A footer shows the AI model used and your remaining token balance. On 1m a note flags that
RSI and ATR are suppressed as noise-dominated.
From an alert. The 📊 5m / 15m / 1h buttons under every alert run exactly the same engine and deliver the same structural read as a separate message - the alert itself is never rewritten. Re-running on another timeframe feeds the earlier read in as context, so the new one also reports what changed.
Tips
- Treat the read as a second opinion on structure - use it to decide which levels matter and what would prove the read wrong, not as a signal.
- Step up timeframes (15m → 1h → 4h) with the buttons to check whether the read holds across horizons.