Last cycle
failed with receipt
Live — what the agent last did
Last cycle produced no reviewable plan
Every candidate was set aside. Reasons recorded: BTCUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; ETHUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; SOLUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; BNBUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; XRPUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; AVAXUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; LINKUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; DOTUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; POLUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; LTCUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; ATOMUSDT — THIN_VOLUME,NO_DIRECTIONAL_EDGE; ADAUSDT — ANALYST_REFUSED:UNPARSEABLE_OUTPUT.
An empty cycle is a real outcome and is recorded as one. The scan below still ran.
Today’s scan — every symbol considered
| Symbol | Score | Read | Why |
|---|---|---|---|
| ADAUSDT | 61 | SHORT | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| ATOMUSDT | 33 | THIN_VOLUME | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| POLUSDT | 33 | THIN_VOLUME | Base and higher timeframe both up. MACD histogram agrees with the trend direction. |
| XRPUSDT | 26 | THIN_VOLUME | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| LINKUSDT | 25 | THIN_VOLUME | Base and higher timeframe both up. Volume is 0.60x its 20-bar average — no confirmation. |
| BTCUSDT | 23 | THIN_VOLUME | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| BNBUSDT | 21 | THIN_VOLUME | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| AVAXUSDT | 19 | THIN_VOLUME | Base and higher timeframe both up. Volume is 0.72x its 20-bar average — no confirmation. |
| DOTUSDT | 9 | THIN_VOLUME | MACD histogram agrees with the trend direction. -0.42% over the momentum lookback. |
| SOLUSDT | 5 | THIN_VOLUME | MACD histogram agrees with the trend direction. Volume is 0.71x its 20-bar average — no confirmation. |
| LTCUSDT | 2 | THIN_VOLUME | MACD histogram agrees with the trend direction. -1.27% over the momentum lookback. |
| ETHUSDT | 0 | THIN_VOLUME | Volume is 0.53x its 20-bar average — no confirmation. %B is 0.52, leaving room before the far band. |
Scored in plain arithmetic before any model is called, so this part costs nothing and exists for every cycle. Only symbols with a directional read are worth paying to analyse.
Cycle gha-36617788266-1-ai-trader — FAILED_WITH_RECEIPT, finished 18 h ago at Sep 29, 2026, 12:15 PDT. Screened 12, analysed 1, spent $0.0026 (estimated). Value class: none.
Today’s scan — every symbol considered
| Symbol | Score | Read | Why |
|---|---|---|---|
| ADAUSDT | 61 | SHORT | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| ATOMUSDT | 33 | THIN_VOLUME | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| POLUSDT | 33 | THIN_VOLUME | Base and higher timeframe both up. MACD histogram agrees with the trend direction. |
| XRPUSDT | 26 | THIN_VOLUME | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| LINKUSDT | 25 | THIN_VOLUME | Base and higher timeframe both up. Volume is 0.60x its 20-bar average — no confirmation. |
| BTCUSDT | 23 | THIN_VOLUME | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| BNBUSDT | 21 | THIN_VOLUME | Base and higher timeframe both down. MACD histogram agrees with the trend direction. |
| AVAXUSDT | 19 | THIN_VOLUME | Base and higher timeframe both up. Volume is 0.72x its 20-bar average — no confirmation. |
| DOTUSDT | 9 | THIN_VOLUME | MACD histogram agrees with the trend direction. -0.42% over the momentum lookback. |
| SOLUSDT | 5 | THIN_VOLUME | MACD histogram agrees with the trend direction. Volume is 0.71x its 20-bar average — no confirmation. |
| LTCUSDT | 2 | THIN_VOLUME | MACD histogram agrees with the trend direction. -1.27% over the momentum lookback. |
| ETHUSDT | 0 | THIN_VOLUME | Volume is 0.53x its 20-bar average — no confirmation. %B is 0.52, leaving room before the far band. |
Scored in plain arithmetic before any model is called, so this part costs nothing and exists for every cycle. Only symbols with a directional read are worth paying to analyse.
How a cycle runs
01Scanplain code
Recent candles for a short watchlist, from a public read-only feed that has no trading surface on it.
02Screenplain code
Trend, momentum, volatility and volume computed in ordinary arithmetic. Anything stale, thin or too volatile is dropped here, for free.
03Analysethe model
Only the survivors reach a model, routed cheapest-first through the shared dispatcher. It returns a thesis with levels, invalidation conditions, the evidence each claim rests on, and what data it lacked. This is the only paid step.
04Riskplain code
A pure function returns approved, watchlist or rejected. The model has no vote and cannot size a position.
05Receiptplain code
Written for every cycle including the empty ones, with what it spent and what class of value it produced.
Approved does not mean take this trade. It means every machine check passed and a person should now look.
What the risk gate refuses
| Condition | Why it stops there | Code |
|---|---|---|
| Data older than the freshness limit | A plan built on a stale bar is a plan for a market that has moved. | STALE_DATA |
| Stop on the wrong side of entry | Absolute distances hide an inverted plan — it reports a healthy ratio while being structurally backwards. | LONG_STOP_NOT_BELOW_ENTRY |
| Reward-to-risk below the floor | A thesis that needs to be right most of the time is not an edge. | LOW_REWARD_RISK |
| Position would breach exposure | Tight stops buy large notionals; the cap binds regardless of conviction. | EXPOSURE_LIMIT |
| Daily loss or drawdown reached | The system stops for the session rather than trading its way back. | DAILY_LOSS_LIMIT |
| Position cannot be sized | No stop distance means no risk unit, so there is nothing to size against. | INVALID_SIZE |
| Anything other than research mode | Fixed in code, not configuration. Approval is unreachable under any other mode. | MODE_NOT_ALLOWED |
Boundaries
No order placement, no wallet connection, no exchange credential exists anywhere in this service. Market data comes from a public read-only host. The research-only mode is fixed in code rather than configuration, so changing it takes a code change and a review.
No fallback ever invents a trade. When the budget runs out, a provider declines, or fixtures are in use, the answer is no setup with the reason attached — never a fabricated entry, stop and target. A made-up plan is indistinguishable downstream from a real one.
It owns no model provider, no API key and no pricing table of its own. Routing, caching, budget and cost evidence all come from the shared dispatcher.
This agent has never run against a live provider. The wiring is verified end to end including its failure paths; its judgement is not. Nothing here is financial advice.