Correlations - do not take 5 of the same position

Diversification is not the number of positions. It is the number of independent risks. 5 long USD positions are 1 position with 5x leverage, only managed worse.

This lesson teaches you to measure your real portfolio exposure, avoid hidden leverage through correlations, and build a portfolio that survives when correlations go to 1.0 on a panic day.

Why the beginner’s “diversification” is an illusion

A typical beginner portfolio:

• Long EURUSD
• Long GBPUSD
• Short USDCHF
• Long XAUUSD
• Long AUDUSD

That looks like 5 different positions. In reality it is 5x short USD. The correlation between them is 0.7-0.9.

When DXY jumps +0.8%, you lose 5R at the same time. If you had one 1R position, you would lose 1R. With diversification you lost 5R for the same event.

The formula:

Real risk = Sum (Size_i * Correlation_i_to_dollar)

If you have 5 positions at 0.5R each and an average correlation to USD of -0.8, the real risk = 5 * 0.5 * 0.8 = 2R on a single DXY move. You thought you were risking 0.5R per position. You are risking 2R on one driver.

How to measure correlations - without a doctorate

You do not need an Excel covariance matrix to start. You need 3 levels:

Level 1: The exposure map (2 minutes before the session)

Write it down in your journal:

Portfolio today:
USD: -3 (short USD 3x) → 3 dependent positions
EUR: +1
GBP: +1
AUD: +1
JPY: 0
Gold: +1 (anti-USD)
SPX: 0
Total USD exposure: -4 (very bad, the maximum should be +-1.5)

If the total exposure to a single currency is >2, you have a concentration. Reduce it.

Level 2: 20D visual correlation

In TradingView add the Correlation Coefficient 20 on D1.

A quick assessment table:

Level 3: Beta to SPX and DXY (for advanced traders)

Every position has a beta to the two main drivers.

Beta_DXY = % change of your position / % change of DXY
Beta_SPX = % change of your position / % change of SPX

Example:
EURUSD Beta_DXY = -0.9 (almost perfectly inverse)
AUDJPY Beta_SPX = +0.8 (almost like SPX)

If the sum of Beta_DXY in the portfolio is >2.0 or <-2.0, you are too dependent on the dollar.

5 forbidden baskets (the most common mistakes)

Basket 1: Short USD x5 EURUSD long + GBPUSD long + AUDUSD long + NZDUSD long + XAU long = 5x short USD. One Fed decision and -4R.

Basket 2: Risk-on x4 SPX long + NAS100 long + AUDJPY long + BTC long = 4x beta to risk-on. VIX +15% and you lose everything.

Basket 3: Long JPY as a safe haven x3 Long USDJPY? No, short. But: short EURJPY + short AUDJPY + short GBPJPY = 3x long JPY. When the BoJ intervenes, 3x the loss.

Basket 4: Gold + Silver + Platinum The correlation of gold to silver is 0.8. You think 3 metals = diversification. It is 1 metal x3.

Basket 5: EUR x3 Long EURUSD + long EURJPY + short EURGBP? That is not 3 positions. It is 2x long EUR and 1x short. Net EUR exposure +1, but 3x the spread and commission.

The rule:

At most 2 positions correlated >0.6 in the same direction at the same time.
A third position = reduce all of them to 0.33R or close the weakest one.

How to build a portfolio resistant to a 1.0 correlation on a panic day

On a panic day (VIX +20%, DXY +1%) correlations go to 1.0 or -1.0. Everything risk-on falls together, everything safe rises together. Your diversification disappears.

That is why you build the portfolio on 2 axes:

Axis 1: Regime At most 60% of positions in one regime (risk-on). 40% in a neutral or the opposite regime.

A healthy example:

• 2x risk-on (AUDJPY long 0.5R, SPX long 0.5R) = 1R risk-on
• 1x defensive (USDCHF long 0.5R or cash)
• 1x independent (e.g. a commodity spread, or none)

Axis 2: Driver The maximum exposure to one driver (DXY, SPX, US10Y) = 1.5R in total.

The formula for your journal:

Driver DXY: sum (size * beta_DXY) < 1.5R
Driver SPX: sum (size * beta_SPX) < 1.5R
If >1.5R → REDUCE the technically weakest position immediately.

Formulas you must have in your spreadsheet

1. Total USD risk:

USD_exposure = Σ (direction_i * size_i) where direction = +1 if long USD, -1 if short USD
Allowed: -1.5 to +1.5

2. Risk adjusted for correlation:

If the correlation of A/B = 0.8 and you have A 0.5R + B 0.5R in the same direction:
Effective risk = 0.5 + 0.5*0.8 = 0.9R (not 1R, but close)
If the correlation = 0.95:
Effective risk = 0.5 + 0.5*0.95 = 0.975R ~ almost the sum.

3. The 2x rule:

If you have 2 positions with >0.6 correlation:
Position 1 = 1R max
Position 2 = 0.25R max or none
Total risk of the pair < 1.25R

When the plan says do not trade

1. You already have 2 short USD positions and you want a third because of a “nice setup”. No. Close the first one, or wait. No more than 2 correlated positions.
2. Portfolio heat >3R. The sum of open risks >3R in total at a given moment. With VIX >20, heat >3R = the risk of intraday ruin.
3. The correlations of all your positions to SPX are >0.7. You are 100% long risk. One tweet about tariffs = -3R.
4. You trade EURUSD and GBPUSD at the same time intraday. They are correlated 0.85 on M15. That is the same position with double the commission. Pick the cleaner chart.
5. You add a gold position while you are already long EURUSD and AUDUSD. Gold correlates 0.6 with both in risk-on. Instead of diversifying, you add the same beta. Close the weakest one, do not add a third.

Mistakes that cost the most

A measurable exercise

Portfolio correlation audit - 14 days

Every day before the session:

1. List all open and planned positions.
2. For each one write down: USD direction (+/-), R size, main driver (DXY/SPX/US10Y).
3. Calculate USD_exposure and the sum of Beta_SPX.
4. Write it into a table:

Date | Number of positions | USD_exposure | Beta_SPX sum | Max correlation in the portfolio | Is >1.5R? (Y/N) | Reduction decision

After 14 days:

• On how many days was USD_exposure >2?
• What was the correlation between your losses? Did the losses arrive in clusters on the same day?
• How many R did you lose to clusters versus to single positions?

The goal: proof that 70% of your losses are not bad setups, only correlated clusters.

Test yourself

1. You are long EURUSD 0.5R, long GBPUSD 0.5R, long AUDUSD 0.5R. How much are you really risking on DXY? a) 0.5R b) ~1.2-1.35R (because the correlations are 0.7-0.9) c) exactly 1.5R

Correct: b) Not the sum, but close to the sum because of the high correlation. That is hidden leverage.

2. You are long SPX and you want to go long BTC because of “crypto diversification”. The SPX/BTC correlation is 0.6. What do you do? a) I enter 1R, it is a different market b) I enter max 0.25-0.33R or I skip it, because it is the same risk-on beta c) I close SPX

Correct: b) Diversification is low correlation, not a different name.

3. On a panic day correlations go to 1.0. What does that mean for a 3x risk-on portfolio? a) Nothing, diversification protects it b) All 3 positions fall at the same time, a loss of 3R instead of 1R, diversification disappears c) It is an opportunity to add

Correct: b) In a crisis diversification fails, which is why a limit on exposure to a driver is crucial before the crisis.

Frequently asked questions

Q: Can I have 5 positions if each one is 0.2R? A: Yes, if the correlations are <0.5. If they are >0.7, then 5x0.2R = 1R on the driver, but 5x the spread and 5x the difficulty of managing it. Better one 1R position on the cleanest chart.

Q: Which correlation should I measure - 20, 50, 100 days? A: For swing trading 20-50D. For intraday 50-100 H1 periods. The shorter the period, the more it reacts to the current regime.

Q: Does hedging with an inversely correlated position work? A: For retail, usually not. Long EURUSD and short GBPUSD is not hedging, it is a position on EURGBP with extra costs and uncertainty. If you want a hedge, use a smaller size, not a second position.

Q: What if all my setups are on USD? A: That is normal - USD is in 88% of FX transactions. That is why you must pick 1-2 of the best setups, not take all of them. Selection > quantity.

---