# TITAN SIGNAL SCALPING ADVANCED PLAYBOOK (1m-4h) This document is the advanced version for manual trading in `SIGNAL`, without bot dependency. It is intended for disciplined execution on all Alpaca timeframes between `1m` and `4h`. Main objective: - increasing the quality of inputs - reducing drawdown - clear protocol for outputs at `TRAIL` and `CHAOS` - fast but standardized decisions --- ## 1) Correct mental model: what the signals mean Signals are not certain predictions. They are "probabilistic trigger + management". - `BUY` / `SELL` = input trigger - `TRAIL` / `EXIT` / `CHAOS` = management / output trigger Prefixes on the chart: - `O` = Oracle - `G` = GPU - `F` = Fusion Critical rule: - entry is taken on `BUY/SELL` - exit is executed on `TRAIL/EXIT/CHAOS` - do not automatically reverse the position at the first exit signal --- ## 2) MTF stack: 3-level professional method Do not trade "single timeframe". It uses 3 levels: - `Anchor TF` (direction context) - `Trigger TF` (signal) - `Execution TF` (effective input) Recommended preset: - for 1m execution: anchor=15m, trigger=5m - for 3m execution: anchor=30m, trigger=10m - for 5m execution: anchor=30m, trigger=15m - for 10m execution: anchor=1h, trigger=15m - for 15m execution: anchor=1h, trigger=30m - for 30m execution: anchor=4h, trigger=1h - for 1h execution: anchor=4h, trigger=1h - for 4h execution: anchor=1d (context), trigger=4h Rule: - if the anchor is clearly opposed to the setup, skip the trade - if the trigger and execution are aligned, the trade has priority --- ## 3) Source selection (O/G/F) depending on the regime ### 3.1 Clean trending - primary: `FUSION` - secondary: `ORACLE` - `GPU` only for fine input timing ### 3.2 Mean reverting - primary: `ORACLE` - secondary: `FUSION` - `GPU` with size reduced ### 3.3 Chaotic / high noise - primary: none for aggressiveness - fallback: `ORACLE` with minimal risk - avoid overtrading on `GPU` Practically: - if you see many `CHAOS` in series, reduce the frequency and size --- ## 4) Entry Quality Score (0-100) - do not fall below the threshold Quickly calculate an internal score before clicking: - fresh signal on closed bar: 0-20 - alignment with Anchor TF: 0-20 - price structure (HH/HL or LH/LL): 0-15 - normal volatility (no extreme spike): 0-10 - no `CHAOS` in last 2 bars: 0-15 - reasonable stopping distance (ATR): 0-10 - minimum risk/reward 1.8: 0-10 Execution threshold: - `>= 70`: normal execution - `60-69`: reduced size (50%) - `< 60`: no trade --- ## 5) Advanced position sizing (risk-first) Simple formula: `qty = (equity * risk_pct) / abs(entry - stop)` Recommended settings: - 1m-5m: `risk_pct` 0.20% - 0.40% - 10m-30m: `risk_pct` 0.35% - 0.60% - 1h-4h: `risk_pct` 0.50% - 0.80% Hard limits: - max daily loss: 2.0% equity - max 3 consecutive losses -> pause 30-60 min - do not double size after loss --- ## 6) Exit Decision Tree (obligatory) Priority order: 1. hard invalidation (technical stop) -> immediate exit 2. `CHAOS` -> fast defensive output 3. `TRAIL` -> lock profit 4. opposite signal confirmed -> closing rest Partial exits model (recommended): - TP1: 40% - TP2: 30% - rest 30% managed with `TRAIL` When `CHAOS` appears after entry: - on 1m/3m/5m: closing 70-100% - on 10m+: minimum discount 50% --- ## 7) Execution rules on sessions ### US Open (first 20-30 min) - high volatility, variable spread - use reduced size for the first 2 entries ### Mid-session - usually the best area for clean runs - strictly follows the MTF protocol ### The last hour - avoid new entries on marginal signals - prefer management of existing positions --- ## 8) Anti-patterns that destroy performance - change `SRC` in the middle of the trade - you enter the unclosed bar from FOMO - treat each `SELL` as a mandatory short - ignore `CHAOS` for "may return" - overtrade after 2-3 losses --- ## 9) Minimum KPI for real performance control Monitor separately on `O/G/F` and each TF: - win rate - expectation (R/trade) - average win / average loss - max adverse excursion (MAE) - max favorable excursion (MFE) - average hold time - output percentage on `TRAIL` vs `CHAOS` What are you looking for: - the source that produces constant positive expectancy - TFs where `CHAOS` occurs too often (high noise) --- ## 10) Weekly tuning protocol (no improvisation) Once a week: 1. extract the last 50-100 trades 2. segment by source (`O/G/F`) and TF 3. mark trades with entry score <70 (should be removed) 4. adjust only 1-2 variables, not the whole system at once 5. the same version runs for 1 week Rule: - don't trade based on a single trade --- ## 11) New ideas (practical roadmap) Below are concrete ideas that are worth implementing: ### 11.1 Signal Health Panel in UI Show live: - active source (`O/G/F`) - `BUY count`, `SELL count`, `TRAIL count`, `CHAOS count` on the last N bars - feed status: `LIVE`, `STALE`, `0 markers` ### 11.2 Confidence ribbon on the chart Discreet overlay with "quality zone": - green: good setup - yellow: medium setup - red: avoid entry ### 11.3 Optional Mode Consensus Entry only if 2/3 sources confirm the direction in the short time window. It can reduce frequency, but increase robustness. ### 11.4 Auto cooldown after CHAOS After `CHAOS`, block new entries X bars on the same symbol/TF. ### 11.5 News risk guard Protection window around major events (e.g. +/- 5 min): - reduce size automatically - or block new entries ### 11.6 Slippage-aware execution hints In UI: - "spread high -> reduce size" - "execution quality low -> skip" ### 11.7 Session templates Preset profiles: - Open Aggressive - Mid-session Balanced - Late-session Defensive --- ## 12) Ultra-fast checklist (10 seconds) Before entering: - [ ] MODE = SIGNAL - [ ] SRC fixed (I don't change it in trade) - [ ] signal on closed bar - [ ] Entry Score >= 70 - [ ] Anchor TF is not clearly opposed - [ ] risk/trade within limits - [ ] I know exactly the output trigger (`TRAIL` / `CHAOS`) If a point is missing: `NO TRADE`. --- ## 13) Operational script for decision (mental) Question 1: "Do I have a valid input signal?" - no -> expect - yes -> question 2 Question 2: "Is the setup aligned MTF and has a score >= 70?" - no -> skip - yes -> question 3 Question 3: "Controlled risk + clear exit plan?" - no -> skip - yes -> execution --- ## 14) Conclusion The real edge comes from: - better entry selection - disciplined exit execution - strict risk control Not from "multiple signals". Use signals as an execution tool, not as a promise.