The pound on 7 October 2016: 1.2600 to 1.1491 in 40 seconds, back at 1.24 by 00:17

[Fairman]

The pound on 7 October 2016: 1.2600 to 1.1491 in 40 seconds, back at 1.24 by 00:17

A thin market at midnight can do things a London morning never would. Here is the sterling flash event, using the figures from the BIS and Bank of England reports.

At about 00:07 UK time on Friday 7 October 2016, during early Asian trading, one sided selling hit GBPUSD. Between 00:07:03 and 00:07:11 the pair fell from 1.2600 to 1.2494, about 8 seconds and 106 pips, with the spread barely changed. At 00:07:15 the CME paused sterling futures for ten seconds, and spot spreads widened sharply. By 00:07:34 it had reached 1.2000, and on Reuters it printed 1.1491 at 00:07:41. That is a fall of nearly 9.7% in about 40 seconds. By 00:17 the pair was back at 1.24.

The sums: from 1.2600 to 1.1491 is 1,109 pips. One lot of GBPUSD moves $10 a pip, so that range is $11,090 per lot.

Now a long position. A 0.5 lot long entered at 1.2610 with a stop at 1.2590, 20 pips and $100 at risk. When the stop triggers the market has gapped, and the fill comes at whatever bids exist. At 1.2400 the loss is 210 pips, or $1,050. At 1.2200 it's 410 pips, or $2,050.

The pair had been under pressure all week on hard Brexit worries, but nothing on the calendar was due at 00:07.

[FXS HongKong Scalper]

Good to see the BIS figures used. The time is worth emphasising: just after midnight in London means early morning in Asia, when the main sterling liquidity is not there. Many Asian banks had wide sterling quotes, and some retail stops filled far below 1.20. It remains one of the best arguments for not holding tight stops on GBP pairs through the hours between New York close and London open, unless the position is sized for a gap.

[Fairman]

The mechanics were laid out in the BIS paper in three stages. First, an orderly but sharp depreciation lasting seconds, with about £252 million of GBPUSD traded in 8 seconds, mostly aggressive sales. Second, less than 10 minutes of extreme dysfunction, when resting bids on trading platforms were exhausted several times and gapping between trades became visible below 1.22. Third, a slower recovery as liquidity came back. The Bank of England adds that a second CME limit at 00:07:29 halted futures trading for two minutes. The point for a retail trader is that the price you see is only as good as the quantity behind it, and at midnight the quantity is small.

[Fairman]

Different feeds printed different lows, which is why your chart may not match someone else's. Reuters, the BIS source, shows 1.1491. Other platforms reported even lower prints, and an earlier figure of 1.1378 was later revised after an outlying trade was cancelled. The BIS adds that some trades during those minutes were torn up or repriced under bilateral arrangements, and that there's no single methodology for the low in FX. A broker's candle low for that minute therefore depends on its feed, and so does whether your stop filled. If you hold a statement from that night, the fill price is the evidence, not any chart.

[Fairman]

Look at the size of the position against the account. A 0.5 lot long with 20 pips of intended risk is $100. On a $5,000 account that is 2%. A fill at 1.2200 costs $2,050, which is 41% of the account, from a trade planned as a 2% risk. A 0.1 lot version would lose $410, about 8%, with an intended risk of $20. Sizing for the stop you plan is one number. Sizing for the worst fill you'd survive is another, and for anything held through a thin session it's the one that matters, since the position doesn't know which kind of session it's in.

[Fairman]

Midnight in London is an awkward hour. New York has closed, Sydney and Tokyo are only just trading, and liquidity providers are thin. The 21:00 UTC rollover had passed two hours earlier. Reuters noted that sterling had been heading for its worst week since January 2009, so participants were already nervous about holding it. Some brokers widen spreads or restrict new orders at that hour for exactly this reason. Looking at your own symbol's trading hours and typical spread for 23:00 to 01:00 UTC is a better guide than anything about 2016 specifically: if you hold positions then, the exposure is there every night.