Data Pipeline

• Source: 1-minute SPY OHLCV bars, Jan 2000 – Oct 2025 (4.5M bars)
• Aggregation: 1m → 3m, 10m, 1h, 4h candles (all hours, excluding 20:00 bar to match TradingView session). Daily/weekly candles use RTH only (9:30–16:00) with 2% wick clipping for bad ticks.
• ATR: Wilder's RMA (alpha=1/14), matching TradingView's ta.atr(14)
• ATR Levels: Intraday tables use daily ATR reference (matching TradingView's request.security(ticker, 'D', ta.atr(14)))
• EMAs: Standard EWM (span=N, adjust=False), matching TradingView's ta.ema()
• Phase Oscillator: EMA(((close - EMA21) / (3 * ATR14)) * 100, 3)

Validation

• Pivot Ribbon EMAs: 0.000% match vs TradingView export (10m bars)
• ATR Levels (daily ref): 0.00–0.07% match
• Phase Oscillator (10m): 0.5–3.5% after convergence
• Phase Oscillator (60m): 0.00% match with full history warmup

Golden Gate Definition

The Golden Gate is the zone between the 38.2% and 61.8% ATR levels. “Entry” = price reaches 38.2%. “Completion” = price reaches 61.8% within the same session.

Core Finding

The 60-minute Phase Oscillator at the moment of Golden Gate entry (38.2% hit) is a strong predictor of whether the Golden Gate will complete through 61.8%. When the 1-hour PO is aligned with the trade direction, completion rates jump significantly and time-to-completion is cut roughly in half.

The Bilbo Conditions

• Bullish Bilbo: 60m Phase Oscillator is above +61.8 (High zone) and rising at the moment the +38.2% ATR level is hit. Completion rate: 77.7% vs 63.0% baseline. Mean time: 34.8 min vs 65.4 min baseline.
• Bearish Bilbo: 60m Phase Oscillator is below -61.8 (Low zone) and falling at the moment the -38.2% ATR level is hit. Completion rate: 90.2% vs 65.0% baseline. Mean time: 24.3 min vs 54.4 min baseline.

Why 60 Minutes?

The 10-minute Phase Oscillator was also tested but provides 5–12x less predictive edge. The 10m PO captures short-term noise — bar-by-bar swings don't reflect sustained momentum. The 60m PO filters for genuine directional conviction built over multiple hours, which is what drives Golden Gate completions.

Key Data Columns for Replication

• Use ind_10m for Golden Gate entry/completion detection (filter to RTH: 09:30–15:59)
• Use ind_1h for the 60-minute Phase Oscillator snapshot at trigger time
• Join with pd.merge_asof(df10, df60, on='timestamp', direction='backward')
• Golden Gate entry: atr_upper_0382 / atr_lower_0382
• Golden Gate completion: atr_upper_0618 / atr_lower_0618
• PO zone classification: High (>61.8), Mid (-61.8 to 61.8), Low (<-61.8)
• PO slope: compare phase_oscillator to its previous bar value

Read the full Bilbo Golden Gate study →