Skip to content

COMPLEX_MODE_FLOW.md

Latest commit

 

History

History
218 lines (153 loc) · 6.19 KB

COMPLEX_MODE_FLOW.md

File metadata and controls

218 lines (153 loc) · 6.19 KB

Complex Mode Flow

Version: 1.0.0
Last Updated: 2025-01-17
Status: Initial public documentation

Overview

Complex mode analyzes custom candle sequences such as [1, 1, -1, 1] rather than a plain N-candle streak. It extends the simple streak engine with pullback-quality analysis, signal scoring, short-signal logic, and richer chart-oriented outputs.

High-Level Flow

graph TD
    A["API request"] --> B["Build AnalysisContext"]
    B --> C["Validate context"]
    C --> D{"Complex mode?"}
    D -->|Yes| E["Load and prepare data"]
    E --> F["Calculate indicators"]
    F --> G["Match complex pattern"]
    G --> H{"Pattern found?"}
    H -->|No| I["Return no-match payload"]
    H -->|Yes| J["Analyze pullback quality"]
    J --> K["Apply filters and score signals"]
    K --> L["Compute C1 / C2 performance"]
    L --> M["Calculate volatility stats"]
    M --> N["Run optional short-signal and interval studies"]
    N --> O["Build response payload"]
Loading

Execution Sequence

  1. Request reaches backend/modules/streak/router.py.
  2. AnalysisContext validates the complex pattern and determines the mode.
  3. Shared data loaders return a normalized OHLCV DataFrame.
  4. Complex-mode logic calculates indicators required for quality scoring.
  5. Pattern-matching logic finds every matching sequence.
  6. Pullback and volume quality are measured.
  7. C1/C2 performance, volatility, and optional side analyses are computed.
  8. The frontend receives a structured payload for tables, filters, and charting.

Detailed Steps

Step 1. Calculate technical indicators

Complex mode relies on indicator columns beyond the base OHLCV set. Typical fields include:

  • ATR
  • RSI
  • disparity
  • volume-derived helpers

Step 2. Match the complex pattern

Pattern matching is handled through the shared streak utilities and checks that the ordered candle sequence matches the requested array.

Example input patterns:

[1, 1, -1, 1]
[1, 1, 1, -1, -1, 1]
[-1, -1, 1, 1]

Step 3. Analyze pullback quality

The quality step measures whether the matched sequence contains a healthy pullback rather than random noise.

Typical signals:

  • retracement ratio
  • pullback volume ratio
  • momentum context

Step 4. Apply filters and calculate signal scores

Signal scoring combines pattern quality and context indicators into a ranked output. The goal is not only to say that a pattern occurred, but also to estimate whether it happened in a stronger or weaker setup.

Step 5. Calculate C1 / C2 outcomes

Complex mode keeps the same candle definitions as simple mode:

  • C1: first candle after pattern completion (T+1)
  • C2: second candle after pattern completion (T+2)

Computed outputs include:

  • c1_success_count
  • c1_total_count
  • c2_success_count
  • c2_total_count
  • average profit or body metrics for C1 and C2

Step 6. Calculate volatility statistics

The module computes post-pattern drawdown and rise behavior, including practical and extreme downside metrics.

Typical outputs:

  • average dip
  • average rise
  • dip standard deviation
  • ATR percentage
  • practical max dip
  • extreme max dip

Step 7. Run short-signal logic

Complex mode can also test short setups when the pattern ends in an overheated bullish configuration.

Typical checks:

  1. filter overbought cases
  2. define an entry threshold
  3. test whether price reaches the entry zone
  4. measure fill rate and short win rate

Step 8. Run interval analysis

Interval studies bucket matched cases by derived indicators such as RSI or disparity and then compute:

  • C1 probability by bucket
  • Wilson confidence intervals
  • Bonferroni-adjusted significance flags

Key Concepts

Pattern notation

Value Meaning Rule
1 Green candle close > open
-1 Red candle close < open

Quality metrics

Metric Preferred zone Interpretation
Retracement ratio 20-40% Healthy pullback depth
Volume ratio < 0.8 Lower pullback volume is healthier
RSI 45-65 Neutral-to-stable momentum context

C1 and C2

  • C1 (T+1): first candle after pattern completion
  • C2 (T+2): second candle after pattern completion
  • success is defined relative to the analysis rule being applied

Key Function Chain

run_complex_analysis()
  -> compute indicators
  -> find_complex_pattern()
  -> analyze_pullback_quality()
  -> calculate_signal_score()
  -> compute C1/C2 performance
  -> build chart and table outputs

Important Files

  • Complex strategy: backend/strategy/streak/complex_strategy.py
  • Shared utilities: backend/strategy/streak/common.py, backend/strategy/streak/statistics.py
  • API endpoint: backend/modules/streak/router.py
  • Controller: backend/strategy/streak/__init__.py

Complex Mode vs Simple Mode

Capability Simple Mode Complex Mode
Input form n_streak complex_pattern array
Example 3 green candles in a row [1, 1, -1, 1]
Pullback-quality analysis No Yes
Signal scoring No Yes
Indicator-based filtering Limited Expanded
C1 analysis Yes Yes
C2 analysis Yes Yes
Chart-oriented payloads Minimal Richer
Best use case Repeated simple streaks Repeated structured sequences

Optimization History

v1.0.0

  • Replaced sum(1 for x in pattern if x == 1) with pattern.count(1)
  • Consolidated multiple C1 / C2 loops into a single pass
  • Replaced separate confidence-count loops with list-comprehension paths

Known Constraints

Pattern complexity

  • Very long patterns can slow down matching
  • practical recommendation: keep pattern length near 5-7 candles when possible

Memory use

  • large match sets can increase memory use
  • chart payload generation involves extra copying and formatting

Filter philosophy

  • max_retracement was removed from hard filtering in favor of scoring and user interpretation

Related Docs

Change History

v1.0.0 (2025-01-17)

  • Initial document creation
  • Added Mermaid-style flow documentation
  • Added detailed execution stages, key concepts, and optimization notes