Korelacja par walutowych — arbitraż statystyczny
Krzysztof €10k konto, długo EUR/USD 1 lot + długo GBP/USD 1 lot. „Diversification!" — myślał. Reality: korelacja +0.92, 2× USD weakness exposure. USD strengthens 1% → -€200 (2× single position). Powinno być 0.5 lot każda lub jeden tylko. Tu pokazujemy korelacja framework + arbitraż.
Korelacja podstawy
Korelacja = statystyczny związek ruchów cen. Range -1 do +1.
Application 1 — risk management
Problem: long EUR/USD + long GBP/USD = 2× USD weakness exposure. NIE diversified.
Solutions
- NIE open both: choose strongest setup
- Reduce size 50% each: 0.5 lot każda zamiast 1.0
- Hedge with anti-correlated: long EUR/USD + long USD/CHF = USD risk hedged
- Position sizing matrix: 2% per uncorrelated, 1% per correlated
- Net portfolio risk < 6% total
Application 2 — divergence trading
Gdy correlated pairs diverge → return to mean expected.
Application 3 — statistical arbitrage
Institutional strategy, retail proxy simple version (triangular arbitrage):
- EUR/GBP implied = EUR/USD ÷ GBP/USD
- Compare implied vs actual EUR/GBP market
- Gdy diff > 5 pips = arbitrage opportunity
- Trade: buy undervalued, sell overvalued
- Tiny profit 1-5 pips per arb
- HFT funds capture milliseconds — retail too slow most cases
Tools dla correlation analysis
- myfxbook correlation matrix: free real-time, 28 pairs
- MetaTrader native: Custom indicator available
- TradingView: correlation coefficient indicator
- Custom Excel: CORREL() formula z historical data
- Python pandas: rolling correlation, advanced
„Korelacja par walutowych = invisible risk doubling. 90% retail open multiple correlated pairs sądząc diversification. Reality: 2× single risk factor. Solution: net currency exposure analysis. Pro standard 5-10 sec check przed entry."
Refresh frequency
Korelacje zmieniają się — refresh weekly minimum:
- Stable regimes (low vol): refresh monthly OK
- Volatile periods: refresh weekly
- Major events (FOMC, ECB, war): refresh after event
- Structural breaks (Brexit 2016): may permanently change
- Crisis (2020 COVID, 2022 Ukraine): historical correlations break
Krzysztof case
Wnioski
Korelacja par walutowych = statystyczny związek ruchów. Range -1 do +1.
Major correlations 2024-2026: EUR/USD + GBP/USD +0.90, EUR/USD + USD/CHF -0.95, AUD/USD + NZD/USD +0.90.
Application 1 — risk management: NIE 2 correlated long = double exposure. Net currency exposure analysis.
Solutions: NIE open both, reduce 50% each, hedge anti-correlated, position sizing matrix.
Application 2 — divergence trading: correlated pairs diverge → return to mean. Pair trade win rate 65-75%.
Application 3 — statistical arbitrage: institutional. Retail simple triangular arbitrage 1-5 pips per arb.
Tools: myfxbook correlation matrix (free), MetaTrader native, TradingView, Excel CORREL, Python pandas.
Refresh: weekly stable regimes, after major events, structural breaks (Brexit) may permanently change.
Krzysztof case: 1 lot EUR/USD + 1 lot GBP/USD "diversified" = -€200 USD strengthens 1%. Lesson: 0.5 lot each.
Pro standard: 5-10 sec correlation check przed każda entry.
Powiązane: pary walutowe charakter baseline, risk management framework, pair trading strategie deeper.
Głębsza analiza — correlation deep dive na ForexMechanics (~40 min, statistical arb + Python).
Źródła i bibliografia
-
Myfxbook Correlation matrix · free real-time www.myfxbook.com ↗
-
BIS Triennial Survey FX correlation data · industry reference www.bis.org ↗
-
Investopedia Statistical arbitrage · educational www.investopedia.com ↗
Najczęstsze pytania
Major pair correlations 2024-2026?
Major pair correlations (typical values 2024-2026): EUR/USD + GBP/USD: +0.85 do +0.92. Both USD-quoted, EUR + GBP somewhat similar economies (developed European). EUR/USD + USD/CHF: -0.92 do -0.97. Mirror image. CHF often EUR proxy. EUR/USD + USD/JPY: -0.3 do +0.3. Weak correlation. JPY safe haven dynamics. GBP/USD + EUR/USD: same as EUR/GBP. AUD/USD + NZD/USD: +0.85 do +0.95. Both commodity currencies, similar risk-on/risk-off behavior. USD/CAD + Oil: -0.7 do -0.85. CAD heavily oil-correlated. Oil up → USD/CAD down. AUD/USD + Gold: +0.5 do +0.75. AUD commodity export (gold). EUR/JPY + Risk sentiment: +0.7 do +0.85. EUR risk-on, JPY risk-off → EUR/JPY tracks SPX. USD/JPY + US 10Y yields: +0.75 do +0.90. Higher US yields → USD demand → USD/JPY up. Practical: track 5-10 key correlations. Refresh weekly. Major changes during crisis (2020 COVID, 2022 Ukraine).
Risk management application?
Risk management = primary correlation application. Problem: trader otwiera long EUR/USD + long GBP/USD. Both correlated +0.90 z USD weakness. Effectively = 2× position USD weakness. NIE diversified, double exposure single risk factor. Detection: portfolio rules — compute net exposure currencies. Each long EUR/USD = +1 EUR -1 USD. Long GBP/USD = +1 GBP -1 USD. Net = +1 EUR +1 GBP -2 USD. USD weakness biggest risk. Solution 1: NIE open both. Choose strongest setup. Solution 2: reduce size 50% each. €100k each EUR/USD + GBP/USD = €50k each. Total exposure same risk USD. Solution 3: hedge z negatively correlated pair. Long EUR/USD + LONG USD/CHF = USD risk hedged. Pure EUR vs CHF play. Anti-correlation pairs hedge: long EUR/USD + long USD/CHF = neutralize USD direction. Bet purely EUR/CHF dynamics. Position sizing matrix: 2% risk per UNCORRELATED setup. 1% risk per correlated pair. Net portfolio risk < 6% total open positions. Real-world example: trader €10k account. Long EUR/USD 1 lot (€100 risk) + long GBP/USD 1 lot (€100 risk). Total exposure USD = 2 lots = €200 risk if USD strengthens. Reduce 0.5 lot each = €100 total. Same effective risk USD.
Divergence trading strategy?
Divergence trading = pair trade gdy correlated pairs diverge from typical relationship. Setup: identify pair z normalnym +0.90 correlation (e.g. EUR/USD + GBP/USD). Track 30-day rolling correlation. Gdy spadnie do +0.50 OR EUR/USD up 1% + GBP/USD down 0.5% (divergence), expect return to mean. Trade: Long undervalued (GBP/USD if dropped). Short overvalued (EUR/USD if rallied). Wait dla return correlation. Exit when correlation back to typical. Why works: short-term divergences often noise. Both fundamentals similar (UK + EU economies linked) — mean reversion. Example: EUR/USD 1.1000 (+1% week). GBP/USD 1.2700 (-0.5% week). Typical correlation +0.90 broken. Divergence 1.5%. Expected: GBP catches up. Trade: long GBP/USD 1 lot, short EUR/USD 1 lot. Hold 1-2 tygodnie. Exit gdy correlation restored (e.g. GBP +0.8% catches up). P/L: long GBP/USD +1% (+€100), short EUR/USD -0.3% (-€30). Net +€70 dla 2-week trade. Win rate: 65-75% w well-defined divergences. Caveat: structural breaks (Brexit GBP weakness 2016, ECB policy divergence) = correlation permanently changes. Don't fight regime change.
Statistical arbitrage advanced?
Statistical arbitrage = institutional strategy, retail proxy possible. Concept: identify multiple correlated assets (e.g. EUR/USD, DXY index, EUR/CHF, gold, US 10Y yields). Construct "fair value" model. Gdy actual price deviates >2 std deviations from model, trade reversion. Tools needed: Python/R statistical analysis. Real-time data feeds. Risk management automated. Tier-1 retail = challenging (institutional-grade). Simple retail version: "Trinity trade" EUR/USD + GBP/USD + EUR/GBP. Math: EUR/GBP = EUR/USD ÷ GBP/USD. Gdy implied EUR/GBP from EUR/USD ÷ GBP/USD differs od actual EUR/GBP > 5 pips, arbitrage opportunity. Trade triangular. Example: EUR/USD 1.1000, GBP/USD 1.2500. Implied EUR/GBP = 1.1000/1.2500 = 0.8800. Actual EUR/GBP market = 0.8810. Diff 10 pips. Arbitrage: buy EUR/GBP (low at 0.8810), sell EUR/USD, buy GBP/USD. Tiny profit per arb (1-5 pips), ale repeats hundreds times daily. Reality retail: high-frequency funds capture these w milliseconds. Retail tools too slow. Spreads + commissions eat profit. Statistical arb retail: longer timeframe (hours-days), NIE milliseconds. Easier execute, smaller edge. 1-2 pips per pair per trade, frequency 5-10/tydz. Modest profit dla deep-pocketed traders €50k+.