3 January 2019: USDJPY drops to about 104.9 while Tokyo is shut, and what a 0.5 lot long would have lost

[Fairman]

3 January 2019: USDJPY drops to about 104.9 while Tokyo is shut, and what a 0.5 lot long would have lost

Apple cut its revenue guidance after the US close on 2 January 2019. In the early hours of 3 January, with Tokyo's markets closed for the New Year holidays, the yen jumped against almost everything. USDJPY fell to around 104.9 within minutes, AUDUSD to roughly 0.6745, and most of it came back within hours.

Run the sums on an illustrative account: $5,000, long 0.5 lot USDJPY at 108.00 with a 20 pip stop. A pip on 0.5 lot is 500 yen, about $4.60 at that price, so the planned loss is roughly $92, or 1.8% of the account.

Now suppose the stop fills near 105.00 because there were no quotes in between. That's 300 pips, 150,000 yen, about $1,430 at 105. Not 1.8%. It's 28.6% of the account, from a trade sized at under 2%.

The exact fill depends on the broker and the moment, and plenty of stops filled at levels better or worse than the lowest print, so this is an order of magnitude and not a claim about any specific account.

A size that survives it: if a 300 pip gap should cost no more than 5% of $5,000, that's $250. A pip on one lot at 105 is about $9.50 (1,000 yen divided by 105), so 300 pips costs $2,857 per lot, and $250 buys 0.087 lots. Round down to 0.08.

[Fairman]

In a gap like that, different order types behave differently. A stop becomes a market order at the first available price, so it fills far from where it was placed. A buy limit resting below the market fills at its price or better, so anyone with a buy limit around 106 got in close to the lows and watched price come back. Limits protect the price but not the outcome. The trade is then open in a market that is still falling, with a stop of its own that can be skipped the same way. And a limit that sat below the gap never fills at all. Whichever order type you use, the exposure lives in the position size.

[Fairman]

Margin adds another layer. One lot of USDJPY at 105 needs about $3,333 of margin at 1:30, which is 100,000 divided by 30. The same 300 pip move costs $2,857, so a $5,000 account falls to about $2,143 of equity and a margin level of roughly 64%. The usual retail stop out level in Europe is 50%, so that's close. A second small position would have pushed it over, and a stop out in a market with no quotes happens at whatever price is available. So the question isn't only whether the stop loss survives the gap. It's also how close the margin level gets to forced closing.

[Fairman]

Thin days are predictable. Tokyo's exchange is closed from 31 December to 3 January, and Golden Week runs from 29 April into early May, with the exact days shifting by year. The yen is traded by desks around the world, so a Tokyo holiday doesn't stop it, but local liquidity thins and the global desks carry the risk. Christmas and Boxing Day, and US holidays, thin out other pairs the same way. Marking those dates on a calendar and treating JPY crosses as no overnight positions on those days costs very little. It's the combination of a holiday and a surprise that does the damage, and the surprise can't be scheduled.

[Fairman]

The size formula generalises. Position size equals the amount you accept losing in a gap divided by the assumed gap in pips times the pip value per lot. EURUSD on $5,000 with $250 accepted and an assumed 150 pip gap: 250 divided by 1,500, so 0.17 lots. GBPJPY with a 300 pip gap, where a pip is 1,000 yen or about $6.60 at 152: 250 divided by 1,974, so 0.13 lots. Each pair needs its own gap assumption, and stop distances don't enter into it. Sizes from a tight stop alone ignore the case where the stop doesn't work. Taking the smaller of the two sizes gives you protection in both situations.

[FXS HongKong Scalper]

That move came at a time when Tokyo was closed for the New Year, and many Japanese banks were not quoting. Stops in USDJPY and AUDUSD filled far from their levels. One lesson from that day is to check the holiday calendar for Japan and other Asian centres before holding positions overnight. Golden Week, Obon and the New Year break all have thinner trading, and flash moves are more likely in those periods.