Three long USD trades at 1% each: one coin flip or three? Combined stop-out odds

[Fairman]

Three long USD trades at 1% each: one coin flip or three? Combined stop-out odds

You hold EURUSD, GBPUSD and AUDUSD longs, each risking 1%. They look like three trades with 1% each. Do they behave like three?

Use a simple model. Each trade has a 50% chance of stopping out. If they were independent, the chance that all three stop is 0.5 x 0.5 x 0.5 = 12.5%. If they were perfectly linked, so that one move in the dollar decides all of them, the chance is 50%, and the day is a single 3% trade.

Reality sits between, and a rough way to represent it is a mixture. Say that 80% of the time the three trades are one trade in effect (one dollar move decides them), and 20% of the time they are independent. Then:

P(all three stop) = 0.8 x 0.5 + 0.2 x 0.125 = 0.425, or 42.5%

That is more than three times the independent figure, and the loss is 3%, not 1%.

The mixture is an assumption, not a measured correlation. Run it yourself with the share of time you think they are linked, and for dollar pairs during US data days it will be higher than for a quiet Asian afternoon.

A cap on the combined risk in one currency is a cleaner rule than a cap per trade.

[FXS HongKong Scalper]

The mixture model is a good way to show the hidden concentration. In practice, three USD trades often move together most of the time, especially around US data. A rule that treats correlated positions as a single position for risk purposes, with a combined risk limit, helps avoid this.

[Amanhasnoname]

I learned this lesson on a US jobs day, long before I'd thought about it in these terms. I had EURUSD, GBPUSD and NZDUSD longs on, each at what I thought was a sensible risk, and the number came out stronger than expected. All three stopped out within about a minute of each other, and what I'd thought of as three small losses was really one large one.

Since then I treat my exposure by currency, not by trade. My limit is 1.5% total risk on any single currency at a time, so if I want to be long three dollar pairs, each gets half a percent. It means some winning days are smaller, but the losing days no longer have that sharp drop.

I also check correlation over the past twenty days before adding a second trade in the same direction. When EURUSD and GBPUSD were moving almost together, I'd skip the second one entirely. On weeks when the pound had its own news and they drifted apart, I'd allow both, since they really did behave more like separate trades.

[Fairman]

Counting by currency is the usual way to see the overlap. EURUSD long is long EUR and short USD. GBPUSD long is long GBP and short USD. AUDUSD long is long AUD and short USD. Three trades, three different long currencies, but a short dollar position in all of them, three times the 1% each. On a dollar rally day all three tend to lose together. A table of the open positions, one row per currency, with the risk in each, makes the concentration visible: USD -3%, EUR +1%, GBP +1%, AUD +1%. If the plan says no currency above 2%, the third trade is declined.

[Fairman]

A hedge shifts the numbers differently. Suppose the third trade is a short EURGBP in place of AUDUSD long. The exposure is now long GBP, short EUR on that trade, plus the first two. EUR nets to zero (long in EURUSD, short in EURGBP), GBP is +2% (long in GBPUSD and long through the EURGBP short), and the dollar is -2%. Total exposure by currency changes, and the plan's 2% limit is met for each. Same count of trades, but it is the currency table, not the trade count, that tells you whether you are diversified.

[Fairman]

The limit can scale with the situation. A normal day might allow 2% per currency. A day with a US release in the afternoon might allow 1%, since all dollar pairs are about to move together. The rule is easy to apply: before entering the third trade, write the net position in each currency and compare it with the limit for the day. It takes a minute, and it stops the common pattern of adding a third and fourth dollar trade because each one looked good on its own chart. Those charts are often showing the same dollar move from different angles.

[Fairman]

Past data gives the real link. Take daily returns of EURUSD, GBPUSD and AUDUSD over the last 100 days and compute the correlation, which a spreadsheet does with CORREL. The values in recent years have often been strongly positive between these pairs, but the number changes by period, so measure your own. A correlation of 0.8 means that when one goes up on a given day, the other is likely to go up too, though not always. Use the figure to adjust the mixture share in the model above, and you will see the combined chance of all three stops rising toward the single trade figure.

[Shadow Trader]

A cross pair is often priced as if two trades were stacked on top of each other. Buying EUR/JPY synthetically means buying EUR/USD and buying USD/JPY, and each leg pays its own spread. So the cross quote frequently lands near the sum of both legs, once they are converted to the same unit, rather than somewhere between them.

That is why a cross can look fine at a pip and a half in calm conditions, then balloon when one leg turns thin. If USD/JPY widens during a quiet Tokyo stretch while EUR/USD stays tight, the cross inherits the messy leg and has nothing of its own to fall back on.

A useful habit is to pull up the two legs beside the cross and watch all three together. A cross much wider than the legs added up points to heavy broker markup or a thin direct market. A cross narrower than the sum suggests real direct liquidity, which is more common in the heavily traded crosses than the obscure ones.

Which crosses have you found trading closest to the sum of their legs, and which ones come in clearly under it?