When EURUSD and GBPUSD Both Signal: Choosing the Stronger Bet
Limits simultaneous exposure to correlated positions, reducing correlated position exposure by 40% through signal strength tiebreaking.
The Correlation Exposure Problem
When EURUSD and GBPUSD both generate L1 buy signals on the same bar, a naive system opens both positions. But these pairs typically correlate at 0.80+. Two positions in correlated pairs is effectively 1.8x the intended risk, not 2x independent bets. S12 addresses this by limiting simultaneous entries in pairs with rolling correlation above 0.70.
When multiple correlated signals fire simultaneously, S12 selects the signal with highest L1 confidence and blocks the others. The blocked signals are logged but not executed. This tiebreaking reduced correlated position exposure by 40% in the V7 backtest.
Implementation Details
S12 maintains a live correlation matrix updated every 50 bars. When a new L1 signal fires, the system checks whether any open position or pending signal has correlation above 0.70 with the new signal's instrument. If so, the higher-confidence signal wins.
The 0.70 threshold was chosen through grid search. Lower thresholds blocked too many legitimate diversifying trades. Higher thresholds allowed too many correlated pairs through. At 0.70, the system blocks the clearly correlated pairs (EURUSD/GBPUSD, AUDUSD/NZDUSD) while allowing genuinely uncorrelated opportunities.
Net Impact on Performance
The 40% reduction in correlated exposure had a modest negative impact on total R (approximately -8R over 7.5 years) because some blocked trades would have been winners. But it reduced max drawdown by 0.12% and reduced the frequency of multi-loss days. In FTMO terms, reducing the chance of two correlated losers hitting on the same day is worth far more than the missed winners. S12 is another module where the value is entirely in risk reduction, and the system is better for it.