Alpha Report
Gold | 2023–2026

Market SectorMetalsModelPythia-v0.5.0
Date FromJan 02 2023Date ToMar 31 2026
Published May 9, 2026

1. Executive Summary

Signal Coverage — Gold

Asset ClassTrading SymbolName
FuturesGCGold Futures (COMEX)

GC Macro Review

2023: Gold demonstrated remarkable resilience throughout 2023, climbing from approximately $1,815/oz in Q1 to over $2,070/oz by year-end, representing a gain of roughly 14%. The metal benefited significantly from banking sector turmoil in Q1 (Silicon Valley Bank, Credit Suisse collapse) and persistent geopolitical tensions surrounding the Russia-Ukraine conflict, which sustained safe-haven demand. The Federal Reserve's pivot toward a more dovish stance in Q4, with markets pricing in rate cuts for 2024, provided additional tailwinds as real yields compressed.

2024: Gold experienced a historic breakout year, surging past $2,400/oz by Q3 and establishing new all-time highs above $2,700/oz in Q4, marking gains exceeding 30% annually. Central bank purchases remained robust, particularly from emerging market economies seeking dollar diversification, while escalating Middle East tensions following the October 2023 Hamas-Israel conflict and subsequent regional spillover effects sustained geopolitical premiums. The Federal Reserve's rate cutting cycle, which commenced in Q3 with an initial 50bp reduction, fundamentally shifted the interest rate environment and compressed opportunity costs for holding non-yielding assets.

2025: Gold consolidated its 2024 gains in a volatile sideways pattern, trading between $2,200-$2,500/oz as markets digested the previous year's dramatic rally. Geopolitical risks remained elevated with ongoing tensions in Eastern Europe and periodic flare-ups in the South China Sea, though reduced intensity compared to 2024 allowed for some risk-on rotation. The Fed maintained a measured approach to policy normalization, delivering gradual 25bp cuts in Q1 and Q2 before pausing as inflation showed signs of re-acceleration, creating a more neutral backdrop for precious metals.

Signal Performance Overview

2023: The strategy demonstrated strong risk-adjusted performance throughout the year, with particularly impressive quarterly Sharpe ratios exceeding 2.0 in Q1, Q2, and Q3, indicating consistent alpha generation relative to volatility. The signal showed excellent risk control with maximum drawdowns remaining well-contained below 2% in most quarters, suggesting effective position sizing and stop-loss mechanisms. Q3 stood out with the highest win rate of 60.38% and strongest Sharpe ratio of 2.495.

2024: Signal performance remained solid but showed more variability across quarters, with Q2 delivering the strongest risk-adjusted returns while Q3 experienced the most challenging conditions with a Sharpe ratio near zero. The strategy maintained disciplined risk management throughout, though Q3's maximum drawdown of -2.33% represented the year's most significant stress period. Overall consistency was evident in the win rates hovering around the mid-50s range.

2025: The year presented the most challenging environment for the strategy, marked by the only negative quarterly return in Q1 with a Sharpe ratio of -1.058. Performance recovered strongly in Q2 with exceptional risk-adjusted returns and the highest quarterly win rate of 62.12%. The relatively low maximum drawdowns throughout the year, particularly the minimal -1.52% in Q2, highlight the strategy's continued emphasis on capital preservation.

2026: Based on the partial year data through Q1, the strategy appears to have entered a highly favorable regime with exceptional risk-adjusted performance and strong win rates above historical averages. The maximum drawdown of -3.81% represents the highest single-quarter risk exposure in the dataset, yet the strong Sharpe ratio suggests this drawdown was quickly recovered with profitable signals.

2. Trading Strategy

In order to produce the metrics below we use the signal in combination with the trading strategy below:

  • Leverage: No leverage is applied for this strategy and metrics
  • Positions:
    • Entry positions: Every 5 minutes (between 09:45 and 15:30 ET) we decide to take a long, short or no position using 1/70 of our starting portfolio for the day (there are 70 possible openings per day). Each long/short position is then split into 5 parts and executed on each minute for the next 5 minutes following the decision. There is no sizing adjustment.
    • Exit positions: We exit all positions at the end of the day. The exits are split over five minutes (15:55–16:00 ET).
  • Costs: 1.5 bp round-turn assumption. Extra exchange/clearing fees not included.
  • Contract series & roll: Front-month continuous. Switch at the open T–5 trading days before expiration; stop trading the expiring contract and start trading the next.

For detailed examples, flowcharts, and a full walkthrough of the trading strategy, see Benchmark Trading Strategy.

3. Model Training Data and Timeframe

CategoryValue
Model FamilyPythia
Versionv0.5.0
ExchangeCME Globex
DataLevel II Limit Order Book (10 levels)
Trained Time Period21Q1 to 24Q1
Final Validation Period25Q1 to 26Q1

4. Performance Metrics

Table 1: Quarterly and Annual Metrics

QuarterReturn (%)SharpeWin (%)CalmarAnn. Vol (%)MDD (%)
20267.8822.52857.6317.06810.648-3.808
26Q16.6312.53658.3188.06212.105-3.808
20256.6431.41554.8542.7704.590-2.345
25Q42.4641.61455.4954.5356.290-2.239
25Q30.4920.67051.4441.3213.544-1.799
25Q24.3013.48362.12112.6945.536-1.519
25Q1-0.636-1.05850.319-1.7422.865-1.741
20247.4511.74254.7223.0454.276-2.446
24Q42.3332.13357.06710.5914.078-0.821
24Q30.2580.17852.8030.2663.490-2.334
24Q24.1582.82955.1218.2756.277-2.146
24Q10.5200.96653.5863.0123.150-1.011
20236.9971.87554.4882.8563.631-2.384
23Q41.3401.13852.2842.9655.024-1.929
23Q31.5772.49560.3806.0782.320-0.952
23Q21.2402.04350.8718.1502.524-0.633
23Q12.6862.59254.20010.6774.164-1.011

5. Next Steps

Download historical predictions using the Client API and confirm performance in your own test harness.

6. Contact

Please reach out with any questions or comments at: info[at]quantumsignals.ai