Every data point, every layer transition, every decision traced back to its source. The full audit trail behind a single trading directive — what was observed, when, and how each finding propagated downstream.
How evidence propagated from macro observation to final directive.
Nine asset-pool observations, regime detection, global liquidity posture.
| TIMESTAMP | SOURCE | OBSERVATION | VALUE |
|---|---|---|---|
| 09:45:02 | FRED · DFF | US effective fed funds rate — stable macro anchor | 4.33% |
| 09:45:08 | TCMB EVDS | TRY benchmark rate snapshot | 47.50% |
| 09:45:14 | FRED · WALCL | Fed balance sheet — net liquidity expanding | $6.82T · +$8B 7d |
| 09:45:19 | CoinGecko | Global crypto market cap | $2.48T · +1.4% 24h |
| 09:45:24 | CoinGecko | BTC dominance — capital rotating to alts | 53.7% · −0.8% 7d |
| 09:45:30 | DefiLlama | Stablecoin supply — dry powder | $218B · +$2.1B 7d |
| 09:45:37 | Binance | SOL/USDT futures open interest | $1.82B · +12% 7d |
| 09:45:43 | Binance | SOL funding rate 8h — contained | +0.012% |
| 09:45:48 | Binance | Long/short ratio — bearish positioning | 0.74 |
| 09:45:54 | Finnhub | Fear & Greed index | 21 · extreme fear |
| 09:46:01 | Computed | CFI-2 market flow score | +0.46 · inflows |
| 09:46:08 | Computed | CFI regime classifier · GARCH + cross-asset vol | EXPANDING |
{ regime: "EXPANDING", flow_score: +0.46, risk_appetite: "moderate", preferred_pool: "crypto", alt_rotation: true }
L1 greenlights the asset scanner with expanding regime and positive flow score. Crypto pool favored, alt-rotation flagged — SOL eligible for deep scan.
Seven sequential gates — each observes, scores, and decides block or pass.
| STEP | GATE | EVIDENCE | SCORE / VERDICT |
|---|---|---|---|
| S1 | Market Pulse | CFI regime EXPANDING, L1 flow positive, BTC dominance falling | 6.2/10 · PASS |
| S2 | Asset Scanner | SOL ranked #3 by composite score · volume 0.3× · spread 0.010% | 8.5/10 · PASS |
| S3 | Safety Check | Liquidity 81.3 · contract clean · no honeypot · 24h vol $1.2B | 5.5/10 · PASS |
| S4 | Timing | 4H BBs squeezed then released · MA alignment bullish · RSI 65 | 9.0/10 · PASS |
| S5 | Trade Plan | Entry $112.05 · SL $107.07 · TP1 $127.74 · R:R 3.7:1 | 9.5/10 · PASS |
| S6 | Trap Check | Bull trap detected at $113.86 · layering pattern · wash trades | 2.0/10 · BLOCKED |
| S7 | Final Verdict | Weighted aggregate of S1-S6 · S6 veto triggers AVOID | 3.6/10 · AVOID |
{ score: 3.6, verdict: "AVOID", blocker: "S6_trap", direction: "LONG", tech_features: {...38 features} }
L2 passes raw technical features to L3 despite AVOID verdict. MLIS Pro will receive the full feature set and form an independent ML-based second opinion.
Four modules independently score; ensemble weighs them into a unified verdict.
| MODULE | WEIGHT | EVIDENCE | SCORE |
|---|---|---|---|
| K1 | Technical · 40% | 8 bullish vs 2 bearish signals · multi-TF aligned · 2 patterns active · Ichimoku bull | 94/100 |
| K2 | Momentum · 25% | RSI 65 neutral · MACD hist +0.71 · hidden divergence · Stoch 72.3 approaching OB | 60/100 |
| K3 | Volatility · 15% | BB expansion · ATR 3.66 (3.27% of price) · squeeze released · IV > HV | 75/100 |
| K4 | Volume · 20% | Relative volume 0.3× below avg · OBV rising · A/D accum · smart money LONG 85% | 60/100 |
| K5 | Ensemble | Weighted: (94·0.40)+(60·0.25)+(75·0.15)+(60·0.20) = 73 | 73/100 · BUY |
{ mlis_score: 73, verdict: "BUY", confidence: 73%, dual_engine: "PARTIAL", conflict: "trap_only_in_L2" }
L3 disagrees with L2 on conviction — agrees on direction. Panel review requested: conflict must be adjudicated by agent experts before L5 receives a final directive.
Four specialists review independently · VOLTRAN synthesizes · SENTINEL holds veto.
"Bullish structure holds — 8 signals aligned, price above both MAs, Ichimoku confirms, Elliott impulse wave 3. Tweezer Top introduces short-term hesitation but does not invalidate trend."
"R:R 3.69 favorable on paper, but layering, wash trading, and untested liquidity compromise the math. If taken: cap at 50% of planned size ($37 risk, 8.5% portfolio)."
"Smart money LONG, 85% confidence, rising 7d accumulation. Fear & Greed at 21 (extreme fear) with SM divergence — classic smart-money-ahead-of-crowd setup."
"Four red flags: layering, wash trading, bull trap at $113.86, liquidity grab. $4.5M long liq clustered $106–110 — this is a target zone. Do not trade."
{ panel_verdict: "AVOID", directional_bias: "LONG", veto_source: "SENTINEL", if_override: "cap 50%, monitor closely" }
Panel directive: AVOID. If user overrides, L5 receives instructions to enter conservatively with active monitoring.
Gate decisions, sizing math, entry/exit geometry — all traceable to upstream layers.
| GATE | SOURCE LAYER | EVIDENCE | VERDICT |
|---|---|---|---|
| G1 | L1 · CFI | Macro filter — EXPANDING regime, flow +0.46 | ✓ PASS |
| G2 | L2 · 7-Step S3 | Liquidity min — score 81.3 ≥ 50 threshold | ✓ PASS |
| G3 | L3 · K3 | Volatility band — ATR within acceptable range | ✓ PASS |
| G4 | Portfolio | Position limit — no existing SOL exposure | ✓ PASS |
| G5 | L3 · K5 | Confidence — 58% below 60% threshold | ✗ FAIL |
| G6 | L2 · S6 + L4 SENTINEL | Trap check — bull trap at $113.86 (escalated via SENTINEL) | ✗ FAIL |
{ status: "WAIT", blueprint: "ready", trigger: "gate_5_and_6_clearance", fallback: "manual_override_50pct_cap" }
Blueprint is fully formed — entry orders are NOT placed. System waits for G5 (confidence ≥60%) and G6 (trap invalidation). If user overrides, L5 executes with 50% size cap per L4 directive.
The full causal chain from first observation to final directive.
| LAYER | TRIGGER EVENT | CAUSED | DOWNSTREAM EFFECT |
|---|---|---|---|
| L1→L2 | EXPANDING regime detected (09:46:08) | Asset scanner activated | SOL eligible for 7-Step · flow +0.46 passed through |
| L2.S1→S5 | Technical structure aligned + R:R 3.7 | Trade plan drafted | $112.05 entry · $107.07 SL · $127.74 TP1 |
| L2.S6 | Bull trap pattern match @ $113.86 | S6 blocks pipeline | 7-Step score 3.6 · verdict AVOID · trap flag propagates |
| L2→L3 | Features forwarded despite AVOID | MLIS Pro engaged | ML-based second opinion requested · 38 features processed |
| L3.K1-K4 | Weighted composite = 73/100 | MLIS BUY verdict | Direction agreement with L2 · conviction disagreement |
| L3→L4 | Dual-engine PARTIAL state | Panel adjudication triggered | Four agents activated for independent review |
| L4.SENTINEL | Layering + trap + wash trading detected | SENTINEL ALERT raised | Veto overrides 3-agent LONG consensus |
| L4.VOLTRAN | Security veto rule applied | Panel synthesis: AVOID | L5 receives AVOID with override instructions |
| L5.Gates | G5 confidence 58% + G6 trap fail | 4/6 gates pass | WAIT-state · blueprint held · no orders placed |
The chain holds. L1 and L3 favored the trade (macro expanding, ML synthesis bullish). L2, L4, and L5 blocked (trap pattern, security alert, confidence threshold). The decisive evidence was a single observation — bull trap at $113.86 — which propagated through L2.S6 → L4.SENTINEL → L5.G6, amplifying at each step.
Every data point observed by any layer is logged with a UTC timestamp, source identifier, and payload hash. Layer transitions are recorded as signed events — the receiving layer must acknowledge the sender's payload before processing. No analysis occurs on data that cannot be traced back to a recorded source.
(1) Every verdict is a function of its declared inputs; (2) every blocker cites the specific observation that triggered it; (3) every conflict between layers is resolved by an explicit published rule (never by silent override); (4) the full audit trail can be reconstructed at any later time from stored evidence. This enables regulatory review, dispute resolution, and post-mortem analysis.
Arrows (→) represent direct dispatch: Layer A's verdict is an input to Layer B. Layer verdicts are never merged ad-hoc; each layer produces exactly one structured output, and the next layer consumes it deterministically.
No Investment Advice. This Evidence Chain is an audit document. It traces how a directive was derived; it does not instruct the reader to act on that directive. Past performance does not indicate future results.
Evidence Freshness. Timestamps reflect when each observation was captured. On a 4H timeframe, the effective shelf-life of the evidence is 2–8 hours. Re-verify before acting on an aged chain.
Layer Independence. Each layer is designed to form its opinion without being influenced by the conclusions of other layers — only the structured inputs. When conflicts arise (as on this chain), the published resolution rules apply.
Data Source Accuracy. All data is sourced from third-party public APIs (Binance, FRED, TCMB EVDS, CoinGecko, DefiLlama, Finnhub) and is believed reliable at capture time. TraderPath makes no warranty of accuracy, completeness, or timeliness of any upstream source.
Audit Limitations. This chain records what the system observed and decided. It cannot record observations the system did not make, nor can it account for unrecorded off-platform information the user may possess.
Risk Disclosure. Trading involves substantial risk of loss. Only trade with capital you can afford to lose.
Intellectual Property. © 2026 TraderPath. The BILGE engine, 7-Step Analysis, MLIS Pro, VOLTRAN, AutoEdge v2, Capital Flow Intelligence, Evidence Chain protocol, and report templates are proprietary. Registered with TÜRKPATENT.