Ujjawal Patel

Signal Confidence Engine — Multi-Timeframe Confidence Scoring

Overview: The Signal Confidence Engine converts raw market data into a single, interpretable confidence score by evaluating indicator confluence across multiple timeframes. Instead of relying on isolated signals, the system emphasizes agreement between trend, momentum, volume participation, and Fibonacci price structure, with higher-timeframe confirmation used to filter noise.

Key inputs

• OHLCV price data — daily and weekly timeframes.
• Moving averages (SMA 50 & SMA 200) — trend strength and structure.
• RSI (14) — momentum and exhaustion analysis.
• Volume participation — relative volume confirmation.
• Fibonacci pivots — price positioning within key structural levels.

Scoring logic (high level)

1. Evaluate trend strength using short- vs long-term moving average alignment.
2. Score momentum based on RSI positioning and continuation strength.
3. Validate price action through volume participation relative to recent averages.
4. Assess price structure using Fibonacci pivot proximity and bias.
5. Aggregate weighted scores into a Daily Confidence Score.
6. Blend Daily (60%) and Weekly (40%) signals to produce a Final Confluence Score.

Outputs

• Daily Signal Confidence — tactical, timeframe-specific analysis.
• Weekly Trend Confirmation — higher-timeframe validation.
• Final Confluence Score — noise-reduced, decision-grade signal.
• Indicator-level breakdown with human-readable reasoning.

Application Link

Professional note

This public overview intentionally abstracts internal weighting logic, scoring thresholds, and tuning parameters. For detailed methodology, indicator calibration, backtesting results, or system architecture discussions, feel free to reach out and I can share a technical brief.

Quantitative Sector Rotation & Risk Management

Overview: This project addresses the challenge of deriving actionable insights from dynamic market data by unifying complex financial modeling and interactive visualization. The deployed web platform ingests real-time data to assess the relative strength and momentum of US equity sectors, providing institutional-grade decision support.

Key Features & Implementation

1. Risk Metrics: Integrated calculations for standard and advanced risk-adjusted performance indicators, including Sharpe Ratio, Sortino Ratio, Alpha, and Beta, allowing for efficient portfolio comparison and risk assessment.
2. Dual-Momentum Strategy: Engineered a customizable backtesting engine allowing users to simulate a rule-based momentum strategy, complete with performance key performance indicators (KPIs) and benchmark comparison against the S&P 500.
3. Technology Stack: Built the calculation core in Python, deployed the responsive front-end application using Streamlit, and utilized Plotly for dynamic, high-fidelity financial charting and interactivity.

Application Link

Fibonacci Central Pivot Range (F‑CPR)

Overview: Traditional pivot systems compute single levels which can misrepresent price structure during volatile regimes. F‑CPR defines a central pivot range (CPR) and refines support/resistance using Fibonacci ratios (e.g., 0.382, 0.5, 0.618) applied to derived S/R spreads. This produces adaptive bands that align with both volume-weighted structure (VWAP) and short-term volatility (ATR), improving the precision of entries, stop placements and profit targets.

Algorithm (high-level)

1. Initialize parameters: SuperTrend Factor, ATR Length, ATR Multiplier, Session Time, Fibonacci ratios (r = [0.382, 0.5, 0.618]).
2. Daily pivot geometry (using previous-day OHLC):
   • Central pivot range (CPR) is computed from prior session OHLC and used to derive intra-day bands.
3. Base support/resistance (classic pivots):

   Detailed algebraic formulas and implementation pseudocode are intentionally withheld. If you'd like the exact pivot math and code snippets, please contact me for the full technical brief.

4. Fibonacci refinement: For each base S/R pair, compute Fibonacci-scaled offsets and overlay them as intermediate bands:

   S1_f_r = S1 + r × (P - S1)
                                       R1_f_r = R1 - r × (R1 - P)

   where r ∈ {0.382, 0.5, 0.618}.
5. Technical overlays:
   • VWAP — volume-weighted typical price to bias bias and intraday fair-value.
   • ATR — Average True Range for volatility scaling.
   • SuperTrend bands — HL2 ± (Factor × ATR), used for trend confirmation and trailing stop placement.
6. Decision rules (examples):
   • Long bias when price > VWAP and above SuperTrend lower band; prefer entries at Fibonacci-refined S1_f_0.618 with tight ATR‑scaled stop below S2.
   • Short bias when price < VWAP and below SuperTrend upper band; prefer entries at Fibonacci-refined R1_f_0.618 with ATR-scaled stop above R2.

Why Fibonacci adds value

Applying Fibonacci ratios to the CPR-derived spreads converts single-point pivots into probabilistic bands that frequently coincide with natural retracement levels and liquidity clusters. Empirically, this reduces noise in entry signals and yields tighter risk-adjusted returns vs. classical single-level pivot rules.

Implementation notes

Core pseudo-code and step-by-step formulas are omitted here to protect proprietary calculations. Reach out to request the implementation notebook and Pine/strategy scripts.

Empirical notes

The attached images (strategy report) show backtest snapshots where applying F‑CPR reduced drawdowns at key sessions and tightened exits. The unique selling point is aligning CPR geometry with Fibonacci scaling rather than using Fibonacci only on raw price swings — this preserves structural context and yields bands with stronger confluence.

Implementation notes: be mindful of data resolution (session vs. intraday), ensure ATR uses the same bar frequency as trading signals, and optionally weight Fibonacci ratios by volume at-price clusters to further refine bands.

Parameters & Summary

Input	Purpose	Recommended
ATR length	Volatility window for ATR	14
ATR multiplier	Stop / band scaling	1.5 – 3.0
SuperTrend factor	Trend-band sensitivity	3
Fibonacci ratios	Refinement levels for S/R bands	0.382, 0.5, 0.618
Session	Pivot computation source	Previous trading day (daily)

F‑CPR Diagram

Drifted Bollinger Bands (DBB)

Overview: DBB is a structurally disciplined breakout system. It signals entries on Bollinger Band breakouts but only when the long-term trend and a volatility filter agree. The approach reduces whipsaws and focuses capital on high-probability drift moves. Risk is controlled via fixed stop-loss & take-profit, with an optional ATR-based trailing stop to lock gains during strong trends.

Formal definitions (high level)

The strategy uses standard Bollinger components (center SMA, band width via standard deviation) combined with trend and volatility screening. Exact algebraic expressions and custom adjustments are available on request.

Algorithm (DBB v1)

1. Inputs: BB_length = 14, BB_k = 2.1, TrendEMA = 365, Vol_SD_len = 15, Vol_SMA_len = 28, StopLoss (fixed %), TakeProfit (fixed %), ATR_len = 14 (for trailing), Trail_ATR_mult.
2. Compute Bollinger Bands using BB_length and BB_k.
3. Trend filter: compute EMA_{365} on Close. Require: For Long trades price > EMA_{365}; for Short trades price < EMA_{365}.
4. Volatility filter: compute SD_{15} (current volatility) and its 28-period SMA. Only allow trades when SD_{15} > SMA_{28}(SD_{15}). This ensures volatility is above recent average.
5. Generate entry signals:
   • Long entry: when Close crosses above Upper Band AND Trend filter = Long AND Volatility filter = true.
   • Short entry: when Close crosses below Lower Band AND Trend filter = Short AND Volatility filter = true.
6. Risk rules (fixed): place StopLoss at a fixed distance (e.g. 1.5% or ATR_scaled) from entry. Set initial TakeProfit at a multiple (e.g. 1.5×) of risk (R). Use position sizing to cap risk per trade (e.g. 0.5–1.0% of equity).
7. Trailing stop (optional): after trade moves favorably by X pips or Y × ATR, convert stop to ATR_len × Trail_ATR_mult below (for long) / above (for short) price and move it up as price advances to secure gains.

Parameters & recommended defaults

Input	Purpose	Recommended
BB length	Bollinger window	14
BB k	Band width multiplier	2.1
Trend EMA	Long-term trend direction	365
Vol SD length	Short-term volatility	15
Vol SMA length	Volatility baseline	28
ATR length	Trailing stop/scale	14
StopLoss	Initial fixed stop (percent or ATR)	1.0–2.0% (or 1.0–1.5 × ATR)
TakeProfit	Initial target (R multiple)	1.5 × Risk
Position size	Risk per trade	0.5–1.0% equity

Implementation notes

Complete pseudocode, risk equations and trailing-stop tuning are not published here. Contact me for the full strategy package including backtest notebooks and Pine Script examples.

Risk management & performance notes

DBB intentionally trades only when volatility and trend align with a breakout to avoid structural noise. Typical performance characteristics observed in backtests:

• Lower trade frequency vs naive BB breakouts, higher win-rate due to filtering.
• Improved maximum drawdown vs unfiltered breakouts when combined with fixed stop sizing and ATR-based trailing stops.
• Works best on liquid, intraday tickers where the 365 EMA provides meaningful structural bias (e.g., futures, FX majors).

Implementation notes

Ensure all indicators use the same bar timeframe. When using percent fixed stops, convert percent to instrument ticks for precision. For portfolio deployment, calibrate trade frequency and volatility thresholds per instrument to avoid overfitting.

GAPR — Gap Reversal Capture

Overview: GAPR identifies high-probability reversal setups at the open by combining a gap threshold with short-term momentum exhaustion (RSI) and volatility-scaled risk management. The system targets limit reversion while adapting stop/target levels to current instrument volatility via ATR.

Key inputs

• Gap threshold (percentage) — defines what constitutes a significant open gap.
• RSI length & thresholds — overbought/oversold levels to confirm exhaustion.
• ATR length & multipliers — for dynamic stop-loss and target sizing.

Rules (high level)

1. Compute gap % = (Today’s Open − Yesterday’s Close) / Yesterday’s Close × 100.
2. Identify Gap Up (≥ threshold) and Gap Down (≤ −threshold).
3. Calculate RSI over the chosen lookback window.
4. Entry signals:
   • Short when Gap Up AND RSI indicates overbought (e.g., >70).
   • Long when Gap Down AND RSI indicates oversold (e.g., <30).
5. Risk management: use ATR-derived stop and target distances; scale position to fixed % risk per trade.

Professional note

This page intentionally omits the full mathematical formulas and proprietary tuning constants. If you’d like the complete specification (exact gap formula implementations, RSI/ATR parameters used in live backtests, and Pine Script strategy code), contact me and I’ll share the technical brief and backtest results.

...

Details...