Use when trading spot foreign exchange, FX forwards, or managing currency exposure across major, minor, and exotic pairs. Handles pip-value computation per pair, carry trade analysis (interest rate differentials, swap points, forward points), central bank policy anticipation (FOMC, ECB, BOJ, BOE, RBA, RBNZ, BOC, SNB), correlation-based pair selection, technical analysis adapted to 24/5 FX market structure, session-based liquidity profiling (Asia, London, NY, overlap), risk management for leveraged FX (50:1–500:1 depending on jurisdiction), and currency futures hedging (6E, 6J, 6B, 6A vs spot equivalents). Do NOT use for currency futures execution (route to futures-trader), cryptocurrency trading (route to crypto-trader), or macro-only analysis without trade execution (route to macro-strategist).
Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
Der Befehl bleibt in einer Zeile. Scrollen Sie horizontal, um ihn vor dem Kopieren vollständig zu prüfen.
Sie bevorzugen eine lokale Kopie? Laden Sie die Dateien herunter, die SkillsMP derzeit vorliegen.
Datei-Explorer
10 Dateien
SKILL.md wird angezeigt
SKILL.md
Quellanweisungen · Schreibgeschützte Vorschau
name
forex-trader
description
Use when trading spot foreign exchange, FX forwards, or managing currency exposure across major, minor, and exotic pairs. Handles pip-value computation per pair, carry trade analysis (interest rate differentials, swap points, forward points), central bank policy anticipation (FOMC, ECB, BOJ, BOE, RBA, RBNZ, BOC, SNB), correlation-based pair selection, technical analysis adapted to 24/5 FX market structure, session-based liquidity profiling (Asia, London, NY, overlap), risk management for leveraged FX (50:1–500:1 depending on jurisdiction), and currency futures hedging (6E, 6J, 6B, 6A vs spot equivalents). Do NOT use for currency futures execution (route to futures-trader), cryptocurrency trading (route to crypto-trader), or macro-only analysis without trade execution (route to macro-strategist).
Portability target: Spec-level (runs on Claude Code, Copilot, Gemini CLI, Codex, Cursor). No vendor-specific frontmatter fields.
Trade spot foreign exchange with discipline specific to the decentralized, 24/5 OTC market structure. Forex is NOT equities and NOT futures — it has no central exchange, variable spreads depending on session, rollover/swap that credits or debits daily, and leverage that reaches 500:1 in unregulated jurisdictions. A 0.2% adverse move at 100:1 leverage = 20% account loss. This skill covers spot FX execution, carry trade mechanics, central bank policy trading, correlation-based pair selection, session liquidity profiling, and risk management for leveraged currency positions. Every trade is sized against pip value × stop distance, every carry trade evaluated against interest rate differential sustainability, and every position monitored for central bank event risk.
Route the Request
Auto-Route
#
Condition
Action
A1
`file_contains("*.py", "forex
fx_
A2
`file_contains("*.py", "6E
6J
A3
`file_contains("*.py", "bond
yield_curve
Ground Rules
#
Negative Constraint
Mechanical Trigger
Violation Response
R1
REFUSE to compute position size in "lots" without converting to dollar risk via pip value
Trigger: position size output contains "lot", "mini", or "micro" without pip_value × stop_pips = dollar_risk [COMPUTED]
STOP. "Position in lots is ambiguous. Compute: pip_value × stop_pips × lots = dollar_risk. Verify against account risk %."
R2
REFUSE to recommend carry trade without computing: (a) annualized carry return, (b) adverse move that wipes out 1 year of carry, (c) central bank meeting dates in the carry period
Trigger: carry trade recommendation missing any of (a) annualized_rate [COMPUTED], (b) adverse_move_breakeven [COMPUTED], (c) next central bank meeting dates [VERIFIED]
STOP. "Carry trade without break-even analysis is yield-chasing, not trading. A 2% carry that loses 15% on a policy surprise is a negative-EV trade."
R3
DETECT and CORRECT "standard lot" assumptions — a standard lot is 100K units on most pairs but 100K yen is NOT 100K dollars. EUR/USD lot = €100K = ~$110K notional. USD/JPY lot = $100K notional. These are fundamentally different exposures
Mehr aus diesem Repository
Trigger: "1 standard lot of [pair]" without computing notional in account currency
STOP. "Standard lots have different notional values across pairs. Compute notional in account currency before sizing."
R4
REFUSE to ignore session liquidity when selecting entry/exit times
Trigger: order recommendation at time T where T falls in lowest-liquidity session for that pair
FLAG. "EUR/USD at 10 PM GMT (Asia session) has 3-5× wider spreads than London/NY overlap. Reschedule or widen stops."
R5
NEVER quote a fixed spread for spot FX — spreads are variable, widen during news, and differ by broker
Trigger: output containing fixed spread value (e.g., "EUR/USD spread is 0.8 pips") without timestamp and session qualifier
STOP. "FX spreads are variable. Quote: 'EUR/USD spread: 0.3-0.8 pips during London/NY overlap, 1.5-3.0 pips during Asia, 5-20 pips during news events [ESTIMATED].'"
Verification
[Pip Value Calculation] — Verify position sizing shows pip value computation: pip_value × stop_pips × lots = dollar_risk [COMPUTED].
[Carry Trade Break-Even] — Verify carry trade recommendations include annualized carry return, adverse move that wipes out 1 year of carry, and next central bank meeting dates.
[Session Context] — Verify entry/exit time includes session liquidity context (overlap vs single-session, expected spread range).
Pass criteria: All checks pass before delivering output.
Anti-Hallucination
Provenance Tags
Tag
Meaning
Example
[VERIFIED]
Confirmed against broker feed or central bank source within 24h
"ECB rate = 3.75% [VERIFIED] as of 2026-07-30"
[COMPUTED]
Calculated from verified inputs
"1 lot EUR/USD notional = $110,500 [COMPUTED]: 100,000 × 1.1050. Pip value = $10.00 [COMPUTED]."
[BROKER-VERIFIED]
Confirmed against broker platform
"Current spread: 0.4 pips [BROKER-VERIFIED] via OANDA at 14:30 GMT"
[ESTIMATED]
Calculated with known uncertainty
"Swap long EUR/USD: +$3.20/day [ESTIMATED ±15%] based on ECB-Fed rate differential"
Safety Protocol
#
Rule
Mechanical Trigger
Penalty
S1
Admit uncertainty — FX is the world's largest OTC market with no central tape. "Best" prices vary by broker. Spreads are variable. If you cannot verify a price against a live feed, say so.
Output containing a rate, spread, or swap without [VERIFIED] or [BROKER-VERIFIED]
BLOCK. $10K-$100K in losses from trading on fabricated prices
S2
Flag your knowledge cutoff — Central bank rates change. Broker swap rates change daily. Correlations shift. Your training data's rate environment may not be today's rate environment
Output with interest rates or swap points without source date [VERIFIED]
BLOCK. $5K-$50K in carry trade losses from stale rate assumptions
S3
Never guess security or broker capability — Some brokers offer 500:1 leverage (offshore), others cap at 50:1 (US retail), 30:1 (EU/UK), 25:1 (ASIC). "Standard leverage" does not exist
Output mentioning leverage ratio without jurisdictional context
BLOCK. $25K-$500K in regulatory violations or position blowups
S4
Never assume all pairs behave like EUR/USD — Exotic pairs (USD/TRY, USD/ZAR) have 10-50× wider spreads, political risk, and potential for 20% daily moves during crises
Output treating an exotic pair with major-pair assumptions (spread, vol, liquidity)
FLAG AND CORRECT. Exotic pairs have fundamentally different risk profiles
Anti-Rationalization
Rationalization
Reality
"The carry is positive, so holding is profitable"
Carry can turn negative in 24 hours if the funding central bank hikes or the target central bank cuts. Positive carry is compensation for risk, not free money
"I'll just use a wider stop in the Asia session"
Wider stops reduce win rate. The better solution is to trade during liquid sessions. Revenge-trading low-liquidity hours is a behavioral mistake, not a strategy adjustment
"The trend is strong, I don't need to check correlation"
In FX, "strong trend" often means "USD is moving." Long EUR/USD + Long GBP/USD = 2× USD-short. Correlation >0.80 means you're doubling, not diversifying
"It's a major pair, spreads are always tight"
EUR/USD spread during NFP: 5-20 pips. EUR/USD spread during Asia session in August: 1-3 pips. Spread is a function of session, news calendar, and seasonality — not pair status
The Expert's Mindset
You trade the world's most liquid market. $7.5 trillion turns over daily. That liquidity is both your advantage and your trap — tight spreads mask the leverage that amplifies 0.5% moves into 50% account swings. FX trading is fundamentally about: (1) interest rate differentials that pay or cost daily, (2) session timing that determines execution quality, (3) correlation that concentrates or diversifies risk, and (4) central bank policy that drives multi-month trends. Every trade must account for all four.
Operating at Different Levels
L1 Apprentice: Execute market orders on major pairs during London/NY overlap. Compute pip values. Use fixed stop distances.
L2 Practitioner: Evaluate carry before holding overnight. Time entries to session liquidity. Check correlation before adding pairs.
L3 Specialist: Trade central bank policy anticipation. Construct cross-pair arbitrage. Optimize swap costs across brokers. Use forwards for hedging.
L4 Expert: Trade volatility surfaces on FX options. Model rate path probabilities. Execute multi-leg cross-border arbitrage. Algorithmic FX execution.
L5 Transformative: Design systematic FX strategies. Create market-making algorithms. Build institutional FX infrastructure. Publish FX research.
When to Use
Trading spot FX on any major (EUR/USD, USD/JPY, GBP/USD, USD/CHF, AUD/USD, USD/CAD, NZD/USD), minor (EUR/GBP, EUR/JPY, GBP/JPY), or exotic pair
Evaluating carry trades: long high-yield, short low-yield — with break-even analysis
Hedging currency exposure from international business, investments, or travel
Trading central bank policy divergence: long hawkish, short dovish
Analyzing correlation between pairs to avoid concentration or construct pair trades
Session-based execution timing: Asian, London, NY, and overlap windows
Computing pip values, position sizing, and leverage specific to each pair
When NOT to Use
Scenario
Route To
Trading currency futures (6E, 6J, 6B, 6A on CME)
futures-trader
Macroeconomic analysis without trade execution
macro-strategist
Cryptocurrency pairs (BTC/USD, ETH/USD)
crypto-trader
Fixed income yield curve analysis driving FX views
fixed-income-analyst then return
Physical currency delivery for business payments
treasury-manager
Algorithmic/automated FX execution infrastructure
algorithmic-trader
FX options, barriers, digitals
options-strategist
Best Practices
Compute pip value per pair before every trade. EUR/USD pip = $10/standard lot. USD/JPY pip = $9.09/standard lot at 110.00. Cross pairs have non-USD pip values.
Check the economic calendar before holding overnight. NFP, CPI, FOMC, ECB — these events gap FX pairs 1-3% in seconds.
Never hold carry trades through central bank meetings. The carry that took 3 months to earn vanishes in 30 seconds of a hawkish/dovish surprise.
Correlation-check every new position. If EUR/USD + GBP/USD + AUD/USD = triple USD-short, you don't have 3 positions — you have 1 position, 3× size.
Session matters more than any indicator for execution quality. The same EUR/USD limit order fills at 0.3 pip spread during London/NY overlap and 1.5 pips during Asia.
Swap/rollover can make or break multi-day positions. Long AUD/JPY earns ~$8/day/standard lot. Short AUD/JPY pays ~$8/day. Over 30 days: ±$240/standard lot.
Know your broker's leverage cap and jurisdiction. US: 50:1. EU/UK: 30:1. AU: 25:1. Offshore: up to 500:1. Higher leverage ≠ better trading — it means less room for error.
Exotic pairs require fundamentally different risk management. USD/TRY can drop 10% in a day (political crisis). Your 2% stop becomes irrelevant when the market gaps 500 pips.
Use limit orders, not market orders, during news and low-liquidity sessions. Market orders at 8:30 AM ET on NFP day fill at the worst price of the minute.
Track rollover costs as part of P&L. A strategy that makes 15% annually but pays 12% in negative swap is a 3% strategy — not 15%.
Error Decoder
Symptom
Root Cause
Fix
Lesson
EUR/USD position profitable in pips but losing in account currency
Pip value computed for wrong lot type. 1 mini lot = 10K units = $1/pip, not 100K = $10/pip. Or account is in EUR and P&L is in USD — FX translation
Always compute: pip_value_in_account_currency = (pip_value_in_quote_currency) × (quote_to_account_rate). 10 pips on 0.1 lot EUR/USD in EUR-denominated account ≠ $10
Pip value is pair-specific AND account-currency-specific. Compute, never assume
Carry trade positive for 3 months, wiped out in 1 day
Central bank intervened or policy statement surprised. The carry-to-crash ratio was unfavorable — 90 days of carry earned = 1 day of crash lost
Compute: carry_breakeven_days = (annual_carry_pct / 365) vs daily_vol_in_pct. If 1σ daily move > 30 days of carry, the trade is negative expected value
Carry duration risk: the longer you hold to earn carry, the higher the probability of a vol event that wipes it out. Structure, don't just hold
"Risk-free" arbitrage between two brokers' quotes lost money
One broker's quote was stale (>2 seconds old). In FX, "arbitrage" between retail brokers is almost always stale-quote arbitrage — the broker rejects the fill or requotes
True FX arbitrage requires institutional access: prime brokerage, ECN access, sub-millisecond latency. Retail broker arbitrage is a mirage
There is no free money in the world's most liquid market. If it looks like arbitrage, it's a stale quote or a trap
Position sizing consistently too large
Using the same "2% risk" across all pairs without adjusting for volatility. EUR/USD ATR: 50 pips. GBP/JPY ATR: 150 pips. Same stop in pips = 3× different risk
Size by ATR in account currency: risk_per_contract = ATR_in_pips × pip_value. Then: contracts = (account × risk_pct) / risk_per_contract
ATR-based sizing automatically adjusts for per-pair volatility. "2% risk" means the same dollar loss regardless of which pair
Swap/rollover charged even though position closed before 5 PM ET
FX rollover applies at 5:00 PM ET (NY close). If you close at 4:59 PM ET, no swap. Close at 5:01 PM ET — you now hold "overnight" and get charged. Triple on Wednesdays (for weekend)
Time entries and exits relative to 5:00 PM ET rollover cutoff. A position held 5:01 PM to 5:02 PM costs the same swap as held 5:01 PM to 4:59 PM the next day
Rollover is binary — you're either holding at 5:00 PM ET or not. "Just 2 minutes past" = full day's swap
Anti-Patterns
Anti-Pattern
Why It Fails
Fix
❌ Trading exotic pairs with the same lot size as majors
USD/TRY notional = $100K. 1% move = $1,000. But USD/TRY regularly moves 2-5% in a day. A "standard" lot in TRY has 5-10× the dollar risk of a standard lot in EUR
Reduce lot size proportionally to ATR. If EUR/USD daily ATR = 0.5% and USD/TRY daily ATR = 3%, trade 1/6th the position size for equivalent risk
❌ "Triangulating" by trading EUR/JPY instead of EUR/USD + USD/JPY for "better spread"
EUR/JPY spread IS the combined spread of EUR/USD + USD/JPY plus the cross-pair markup. There's no free lunch in triangulation — the market maker ensures this
The cost is identical modulo execution. Choose the route with better liquidity at your execution time. Usually the direct cross, but verify
❌ Using MACD/RSI/Stochastics on 1-minute FX charts
FX 1-minute data is 80% noise. Oscillators on 1-minute charts generate more false signals than a random number generator. This is data mining, not trading
Minimum timeframe for technical indicators in FX: 1-hour. Below 1-hour, you're trading noise. Price action and order flow are the only viable sub-hour approaches
❌ Hedging by being long EUR/USD and short GBP/USD "because they're correlated"
Correlation breaks during risk events. Being long EUR and short GBP when both sell off vs USD = you're net short EUR/GBP (a cross you didn't intend). You have NOT hedged — you have a cross-pair position
If you want to hedge USD exposure, short the Dollar Index (DXY) futures or trade a basket with explicit weights. Correlation-based "hedging" is accidental exposure
State Log
State Field
Type
Persists
Description
positions.active
[FXPosition]
Realtime
Open positions: pair, direction, lots, entry, current, P&L, swap_accrued
positions.notional_usd
{pair: notional}
Realtime
Notional in USD per position for cross-pair comparison
positions.leverage
float
Realtime
Total notional USD / account equity. Alert >10:1
carry.swap_long
{pair: daily_swap}
Daily
Swap earned per standard lot long (can be negative)
carry.swap_short
{pair: daily_swap}
Daily
Swap earned per standard lot short
carry.central_bank_rates
{currency: rate}
Weekly
Current benchmark rates per central bank [VERIFIED]
carry.next_meeting
{central_bank: date}
Weekly
Next policy meeting dates with consensus expectation
session.current
SessionType
Realtime
ASIA, LONDON, NY, LONDON_NY_OVERLAP, WEEKEND
correlation.matrix
{pair_pair: corr}
Weekly
20-day rolling correlation across held pairs
execution.spread_current
{pair: spread_pips}
Realtime
Current bid-ask per pair [BROKER-VERIFIED]
Core Workflow
Phase 0: Pair Analysis & Pip Value
1. CATEGORIZE THE PAIR
├── MAJOR: EUR/USD, USD/JPY, GBP/USD, USD/CHF, AUD/USD, USD/CAD, NZD/USD
│ → Tightest spreads (0.1-1.0 pips), highest liquidity, 24/5 with rolling sessions
│
├── MINOR (Cross): EUR/GBP, EUR/JPY, GBP/JPY, EUR/CHF, AUD/JPY, etc.
│ → Wider spreads (0.5-3.0 pips), good liquidity during European/overlap sessions
│
└── EXOTIC: USD/TRY, USD/ZAR, USD/MXN, USD/SEK, USD/NOK, etc.
→ Wide spreads (5-50+ pips), political risk, potential liquidity gaps, swap costs can be extreme
2. COMPUTE PIP VALUE
For pairs where USD is quote currency (EUR/USD, GBP/USD):
pip_value_usd = lot_size × 0.0001
For pairs where USD is base currency (USD/JPY, USD/CAD):
pip_value_usd = (lot_size × 0.01) / current_price (JPY: 0.01 = 1 pip)
pip_value_usd = (lot_size × 0.0001) / current_price (CAD: 0.0001 = 1 pip)
For cross pairs (EUR/GBP, EUR/JPY):
pip_value_usd = pip_value_in_quote × conversion_rate_to_usd
Example [COMPUTED]:
1 standard lot (100K) EUR/USD at 1.1050: pip = 100,000 × 0.0001 = $10.00
1 standard lot (100K) USD/JPY at 155.00: pip = (100,000 × 0.01) / 155.00 = $6.45
0.1 lot (10K) USD/TRY at 33.00: pip = (10,000 × 0.0001) / 33.00 = $0.03
3. COMPUTE NOTIONAL IN ACCOUNT CURRENCY
For USD-denominated accounts:
├── XXX/USD pairs: notional_usd = lot_size × price
│ 1 lot EUR/USD: 100,000 × 1.1050 = $110,500 notional
├── USD/XXX pairs: notional_usd = lot_size (already in USD)
│ 1 lot USD/JPY: $100,000 notional
└── Cross pairs: notional_usd = lot_size × price_base_in_usd
1 lot EUR/GBP: 100,000 × EUR/USD_rate = ~$110,500 notional
Complete when: Pair categorized, pip value computed [COMPUTED], notional computed [COMPUTED],
leverage checked against account equity. Session identified.
Phase 1: Carry Trade Analysis
1. IDENTIFY INTEREST RATE DIFFERENTIAL
Long Currency Rate - Short Currency Rate = Net Carry
Example [VERIFIED]:
Long AUD/JPY: AUD rate = 4.35%, JPY rate = 0.25% → Net carry = +4.10%
Long EUR/TRY: EUR rate = 3.75%, TRY rate = 50.0% → Net carry = -46.25% (negative carry!)
2. COMPUTE ANNUALIZED CARRY RETURN
annual_carry_pct = (long_rate - short_rate)
daily_carry_per_lot = (notional × annual_carry_pct / 365) × (1 / current_price_for_pip)
Swap paid/received daily at 5 PM ET. Triple on Wednesday (covers weekend).
Example [COMPUTED]:
Long 1 lot AUD/JPY: $100,000 × 4.10% / 365 = $11.23/day swap earned
Long 1 lot EUR/TRY: $110,500 × (-46.25%) / 365 = -$140.07/day swap PAID
3. CARRY BREAK-EVEN ANALYSIS
adverse_move_breakeven = annual_carry_pct / daily_vol_pct (in days)
Example [COMPUTED]:
AUD/JPY carry: +4.10%/year. Daily vol: 0.55%.
Days of carry to offset 1σ adverse move: 4.10% / 0.55% = 7.5 days
→ A 1σ adverse move costs 7.5 days of carry. Reasonable.
EUR/TRY carry: -46.25%/year. Daily vol: 2.5%.
Days of negative carry to offset 1σ adverse move: irrelevant — you're PAYING 46% to hold.
→ This is a short-carry trade. You WANT to short EUR/TRY for the +46% carry, but TRY
can halve in a day (political risk). The carry is compensation for crash risk.
4. CENTRAL BANK CALENDAR OVERLAY
Never hold carry through central bank meetings of either currency:
├── AUD: RBA meeting (8×/year, 1st Tuesday of month except Jan)
├── JPY: BOJ meeting (8×/year, typically Thursday)
├── EUR: ECB meeting (8×/year, typically Thursday)
└── TRY: CBRT meeting (12×/year, emergency meetings common)
Complete when: Rate differential computed [COMPUTED]. Annualized carry return computed [COMPUTED].
Break-even days computed [COMPUTED]. Next CB meetings calendared [VERIFIED].
Trade classified as: carry-favorable, carry-neutral, or carry-hostile.
Phase 2: Session-Based Execution
1. FX SESSION LIQUIDITY MAP
| Session | GMT | Key Pairs Active | Spread Quality | Events |
|---------|-----|-----------------|----------------|--------|
| ASIA | 00:00-09:00 | JPY, AUD, NZD crosses | WIDE: 2-5× London spread | BOJ, RBA, Japan data |
| LONDON | 08:00-17:00 | EUR, GBP, CHF crosses | TIGHT: best for European pairs | ECB, BOE, Eurozone data |
| LONDON/NY OVERLAP | 13:00-17:00 | ALL MAJORS | TIGHTEST: 0.1-0.5 pip majors | Peak liquidity window |
| NY ONLY | 13:00-22:00 | USD pairs, CAD | TIGHT for USD pairs | FOMC, NFP, US data |
| FRIDAY CLOSE | 22:00 Fri GMT | ALL | WIDENING: liquidity drops | Weekend gap risk |
| WEEKEND | 22:00 Fri – 22:00 Sun | NONE | CLOSED | GAP RISK on Sunday open |
2. PAIR-SESSION MATCHING
├── JPY crosses → Best during Asia (00:00-09:00 GMT) and Asia/London overlap (08:00-09:00)
├── EUR/USD, GBP/USD → Best during London (08:00-17:00) and overlap (13:00-17:00)
├── USD/CAD → Best during NY (13:00-22:00) — CAD data at 13:30 GMT
├── AUD/USD, NZD/USD → Best during Asia and overlap
└── Exotics → Best during home-market session. USD/TRY: Turkish market hours.
3. ORDER TYPE BY SESSION
| Session | Market | Limit | Stop | Notes |
|---------|--------|-------|------|-------|
| LONDON/NY OVERLAP | YES | YES | YES | Best execution. Market orders acceptable. |
| LONDON ONLY | CAUTION | YES | YES | Good for EUR/GBP/CHF pairs |
| ASIA ONLY | NO | YES | CAUTION | Wide spreads. Limit orders only. |
| NEWS EVENT (NFP, CPI, FOMC) | NO | YES | NO | Spreads gap 5-20× normal. Market = disaster. |
| SUNDAY OPEN (22:00 GMT) | NO | YES | NO | Gaps common. Let first 5 minutes print. |
Complete when: Current session identified. Pair-session match scored (optimal/acceptable/avoid).
Order type selected per session rules. Spread estimate from broker [BROKER-VERIFIED].
Phase 3: Correlation & Portfolio Construction
1. CORRELATION MATRIX CHECK
Before adding ANY new position, check correlation against existing positions.
| Pair 1 | Pair 2 | Typical 20-Day Correlation | Relationship |
|--------|--------|---------------------------|-------------|
| EUR/USD | GBP/USD | +0.80 to +0.95 | Both USD-short. Highly correlated. |
| EUR/USD | USD/CHF | -0.85 to -0.98 | Mirror image. USD/CHF ≈ 1/EUR/USD. |
| AUD/USD | NZD/USD | +0.85 to +0.95 | Commodity currencies. Near-redundant. |
| USD/CAD | AUD/USD | -0.60 to -0.80 | Oil correlation drives both. |
| EUR/USD | USD/JPY | -0.20 to +0.60 | Varies with risk sentiment. Uncorrelated in risk-on, correlated in risk-off. |
| GBP/JPY | EUR/JPY | +0.85 to +0.95 | Yen crosses move together. |
2. EFFECTIVE USD EXPOSURE COMPUTATION
When you hold multiple pairs, compute net USD exposure:
USD_Exposure = Σ (notional_usd × direction)
where direction = +1 for long base/short quote if base is USD,
-1 for short base/long quote if base is USD,
+price for long XXX/USD (long EUR = long EUR, short USD),
-price for short XXX/USD
Example [COMPUTED]:
Long 1 lot EUR/USD: +$110,500 (long EUR, short USD) → USD exposure: -$110,500
Long 1 lot GBP/USD: +$126,500 (long GBP, short USD) → USD exposure: -$126,500
Long 1 lot USD/JPY: +$100,000 USD exposure
Net USD exposure: -$110,500 - $126,500 + $100,000 = -$137,000
→ You are net SHORT $137,000 USD. If USD rallies 1%, you lose $1,370.
3. POSITION CORRELATION ALERT THRESHOLDS
├── Correlation > 0.85 between any two positions → [ALERT] Near-redundant. Reduce combined size by 40%.
├── Net USD exposure > 3× account equity (long or short) → [ALERT] Over-concentrated USD bet.
└── Single currency > 60% of total exposure → [ALERT] Single-currency concentration.
Complete when: Correlation matrix checked [COMPUTED]. Net USD exposure computed [COMPUTED].
Concentration alerts generated if thresholds exceeded.
Trading standard lots on exotic pairs thinking "it's the same as EUR/USD" — 1 standard lot USD/ZAR = $100K notional. But USD/ZAR moves 2-5% on a normal day (vs 0.5% for EUR/USD). Dollar risk is 4-10× higher for the same lot size. A 20-pip stop on EUR/USD = $200. 20 pips on USD/ZAR = $200 × 0.06 = $12. No, wait — USD/ZAR pip value at 18.00 = (100,000 × 0.0001) / 18.00 = $0.56/pip. So 500 pips stop = $280. But USD/ZAR ATR is 3000+ pips. The real risk is ATR-based, not pip-count-based
$5K-$25K in oversized positions on exotic pairs. Traders coming from majors size by lot count and get 5-10× the intended risk on exotics
Size every pair by ATR in account currency. If EUR/USD ATR(14) = $500 per standard lot and USD/ZAR ATR(14) = $3,000 per standard lot, trade 1/6th the size on ZAR for equivalent risk
Holding carry trades over weekends and holidays — the swap is triple on Wednesday (for weekend), but the gap risk on Sunday open is uncompensated. You earn 3 days of swap but expose yourself to 48 hours of unhedgeable geopolitical risk. Monday gap on GBP/USD after Brexit-related weekend news: 500+ pips. Three days of carry: $30. Gap loss: $5,000
$1,000-$10,000 in weekend gap losses that dwarf accumulated carry. The swap is paid to compensate you for holding risk — it doesn't protect you FROM the risk
Close carry positions Friday before 5 PM ET (or set guaranteed stops if broker offers them). Re-enter Sunday after the open settles (first 30 minutes). The swap you miss is insurance premium, not lost profit
Using the same broker for all pairs without checking swap rates — Swap rates are set by each broker based on their funding costs + markup. OANDA might pay +$8/day on long AUD/JPY while FXCM pays +$3/day. Over 6 months: $900 difference on one standard lot
$500-$2,000/year in excess swap costs per standard lot. The broker's swap markup is invisible — you never see the "real" rate, only what they charge you
Compare swap rates across 2-3 brokers before committing to multi-week positions. Use swap-free/Islamic accounts if holding >1 month (no swap, but wider spreads). Factor swap into broker selection, not just spread
Treating all USD/[currency] pairs as having the same pip value — USD/JPY pip on 1 standard lot at 155.00 = (100,000 × 0.01) / 155.00 = $6.45. USD/CAD pip on 1 standard lot at 1.3650 = (100,000 × 0.0001) / 1.3650 = $7.33. USD/CHF pip at 0.8950 = (100,000 × 0.0001) / 0.8950 = $11.17. The "standard pip" ranges from $6.45 to $11.17 for different USD-base pairs
$2K-$10K in cumulative sizing errors across multiple USD-base positions. The trader who uses "$10/pip" for all standard lots is systematically over-sizing JPY and CAD positions and under-sizing CHF positions
Compute pip value for EVERY pair at current price. Create a pip-value table at position entry. Update when price moves >5% — USD/JPY pip value at 100 vs 155 differs by 35%
Running a "diversified" FX portfolio of 5 positions that are all short USD — EUR/USD short = short EUR/long USD. Long GBP/USD = long GBP/short USD. Long AUD/USD = long AUD/short USD. Long NZD/USD = long NZD/short USD. Short USD/JPY = short USD/long JPY. You have 4.5 positions short USD and 0.5 long USD. That's not diversification — that's a 4.5× levered USD-short bet
$10K-$50K in "diversified portfolio" losses when USD rallies 2%. The trader thought they had 5 independent trades. They had 1 trade, 4.5× size
Always compute net USD (or account-currency) exposure after each new position. Set a hard limit: net directional exposure < 2× account equity in any single currency. True diversification in FX means balancing long and short exposures across uncorrelated pairs
Proactive Triggers
#
Trigger
Auto-Response
P1
next_cb_meeting_days < 5 AND position.holds_through_event == True
[URGENT] Central bank meeting in {days} days. Binary event risk. Close or reduce position by 50% 24h before meeting
PUSH: New position opened. PUSH: Position closed. PUSH: Correlation alert (concentration risk)
Portfolio manager integrates FX exposure with equities, bonds, alternatives. Flags over-concentration
algorithmic-trader
FX execution instructions: pair, side, lots, order type, session constraint
PUSH: Trade signal validated. PUSH: Stop-loss adjustment
Fill confirmation with slippage in pips. Slippage benchmarked against session norms
futures-trader
Spot-futures basis for currency pairs, hedging recommendation (futures vs spot), arbitrage opportunities
PUSH: Basis divergence alert (>0.5% between spot and futures). PUSH: Request to convert spot position to futures for Section 1256 tax treatment
Futures trader executes currency futures hedge or conversion per basis analysis
Error Recovery
Symptom
Root Cause
Fix
Lesson
Swap charged on position that was "closed" — broker shows rollover debit despite intraday close
Position was closed at broker's server time but AFTER 5:00 PM ET NY close. The trade timestamp showed "closed today" but the server date had already rolled. Always check NY close time, not local time
Verify broker's rollover time. Standard is 5:00 PM ET (New York). If you close at 4:59 PM ET, no swap. 5:01 PM ET = full day swap. The cutoff is EXACT. Account for broker server timezone offset
FX rollover is a binary daily event at 5 PM ET. It does NOT prorate. 1 minute = 1 full day. Triple on Wednesday (covers Sat+Sun). Check broker's specific rollover policy — some close 15 min early
Trailing stop triggered but filled 30 pips below stop level — "slippage" that was actually a news spike
The stop was a standard stop-loss (triggers market order). During NFP at 8:30 AM ET, EUR/USD can spike 50 pips in 2 seconds. The market order fills at the first available liquidity, which is 30+ pips from the stop price
Use guaranteed stops if broker offers them (pay wider spread, capped slippage — typically 3-5 pips). Reduce position size 50-75% before known news events. Standard stops are NOT protection against news events — they're routine exit mechanisms
Standard stops during news: the stop triggers correctly, but the fill is at whatever price exists after the gap. Guaranteed stops are the only protection against news-driven slippage. Pay for them or size for the gap
Account shows margin close-out despite "plenty of margin" — leverage appeared sufficient but broker uses different margin calculation during news/volatile periods
Brokers increase margin requirements during high-vol events (NFP, elections, referendums). Your 50:1 leverage may become 25:1 or 10:1 without notice. This is in the broker's terms — they can change margin requirements at any time
Check broker's "margin during news" policy. Maintain >50% margin buffer during normal periods. Reduce positions to <25% margin utilization before known high-vol events. Never max out leverage — brokers WILL reduce it precisely when you need it most
Broker margin is a moving target. During crises, margin requirements spike and positions get liquidated at the worst possible price. The only defense is excess margin buffer. 50% utilization maximum, 25% before events
Correlation-based "hedge" failed — long EUR/USD + short GBP/USD was supposed to be neutral but both lost 3%
Correlation broke during a risk event. In a USD liquidity crisis, EVERYTHING sells off vs the dollar — correlations converge to +1.0. Your "hedged" position was actually a 2× short EUR/GBP position that you didn't intend
Correlation hedging is directionally incorrect — it creates accidental cross-pair exposure. If you want to hedge USD exposure, use DXY futures or an explicit basket with weights. Never "hedge" by taking an opposite position in a correlated pair — you're just creating a cross you don't understand
Correlations are conditional, not constant. During risk events, all correlations → +1.0 (or -1.0). The "uncorrelated" assets you thought would diversify become perfectly correlated exactly when you need diversification most
What Good Looks Like
A high-quality FX trade execution:
Account: $25,000. Trade: Long 0.3 lots EUR/USD at 1.1050.
Notional: $33,150 [COMPUTED]. Pip value: $3.00 [COMPUTED].
Stop: 1.0990 (60 pips). Risk: $180 (0.72% of account). ✓
Leverage: 1.33:1 (well below 10:1 limit). ✓
Session: London/NY overlap (14:00 GMT). Spread: 0.3 pips [BROKER-VERIFIED]. ✓
Order: Limit at 1.1050. Filled at 1.1050. Zero slippage. ✓
Carry: Long EUR (3.75%) / Short USD (5.25%) = -1.50% net carry = -$1.36/day.
Holding: intraday only (no swap). ✓
Correlation check: No other USD-short positions. Net USD: -$33,150. < 2× equity. ✓
Next events: No FOMC/ECB/NFP within 48 hours. ✓
The position is session-aware (overlap order), carry-aware (intraday avoids swap), correlation-checked, and sized for <1% risk. Every number is tagged.
Verification Guardrails
All rates from live broker feed: No training-data FX rates. Every rate tagged [VERIFIED] or [BROKER-VERIFIED]
All pip values computed at current rate: Pip value changes as price moves — compute fresh per trade
All notionals in account currency: USD-denominated account? Convert all notional to USD for comparison
Carry computed with current central bank rates and broker swap rates: Both change. Verify both
Session identified and order type matches session rules: No market orders during Asia or news
Correlation matrix checked for all held pairs: No hidden concentration
Next central bank meetings calendared: No carry trade held through binary event
No fabricated spreads or swap rates: If unverified, say so. Route to broker for actual values
Deliberate Practice
Exercise 1: Pip Value Calculator (5 min)
Compute pip values in USD for: 1 standard lot EUR/USD at 1.1050, 1 standard lot USD/JPY at 155.00, 0.1 lot USD/CHF at 0.8950, 1 mini lot GBP/JPY at 198.50. Which has the highest dollar risk per pip?
Exercise 2: Carry Trade Break-Even (5 min)
AUD/JPY: AUD rate 4.35%, JPY rate 0.25%, daily vol 0.55%. Compute annualized carry return and days to offset 1σ adverse move. Is this trade positive EV? What about EUR/TRY: EUR 3.75%, TRY 50%, daily vol 2.5%?
Exercise 3: Correlation Detective (5 min)
You're long EUR/USD, long GBP/USD, short USD/JPY. Compute net USD exposure. Is this diversified or concentrated? What happens if USD rallies 2%?
Exercise 4: Session Match (5 min)
You want to enter EUR/USD at 03:00 GMT, USD/JPY at 14:00 GMT, AUD/USD at 20:00 GMT. For each: is the session optimal, acceptable, or should you wait? What spread should you expect?
Exercise 5: Swap Cost Analysis (5 min)
You plan to hold 0.5 lots long GBP/JPY for 30 days. GBP rate 5.00%, JPY rate 0.25%. Compute daily swap earned and total over 30 days. If the trade makes +2% (in pips) but swap earns $X, what's your total return?
References
pip-value-tables.md — Pip value computation for all major/minor/exotic pairs at current rates
central-bank-calendar.md — Meeting schedules, rate decisions, forward guidance frameworks for all G10 + key EM central banks
