Is trading during rollover ever worth the spread spike?

[LondonScalper]

Rollover is almost never worth it for my scalp book.

Spread spikes, thin books, and swap oddities turn a tidy M1 idea into a donation. I am flat before the window unless a rare written exception exists, and I cannot remember the last time one did. People who hunt rollover are usually paying entertainment prices.

For anyone who has not looked closely: on most brokers the daily rollover sits around 22:00 London, give or take depending on the season and the server. EURUSD spread can go from 0.1 to 2 or 3 pips for a few minutes, and crosses like GBPNZD go much wider. A stop sitting near price can be taken out by spread alone, without the market actually trading there.

I once had a small EURJPY position stopped out at 22:01 by a spread spike, with price back at entry by 22:10. The lesson cost me a few pips and I am grateful it was cheap.

If a position must be held through it, that decision was made hours earlier with a swing thesis, not a scalp thesis. Different playbook, different account risk. Scalp rules say flatten.

[Shadow Trader]

One edge case worth adding for anyone in London: the rollover is tied to 17:00 New York, not to 22:00 on your own clock. This year the UK clocks go back on 25 October while New York waits until 1 November, and in between the spike lands at 21:00 London.

I learned that the expensive way with a pending order, not a position. A sell stop entry 1.2 pips under the bid got triggered at 21:00 during that mismatch week because the bid dropped with the spread, and the mid price never traded near my level.

The chart below rebuilds that kind of minute as an illustration, with the bid wick printing through the sell stop while the mid barely moves. Spread jumps from 0.2 to 3.0 pips and is back near normal by 21:08.

img_13408_r1.png (64.22 KiB) Viewed 18 times

So my flatten rule now covers resting orders as well as open tickets, from ten minutes before until ten minutes after. I also check the server time on the platform rather than trusting the calendar in my head.

Do you cancel pending orders across the window too, or only close open positions? Most people I ask have only ever thought about the stops.

[Shadow Trader]

LondonScalper, the part I measure now is how long each pair takes to recover after the window, since that decides when evening orders can go back on. On my feed EURUSD is usually back under 1.5 times its normal spread by about 22:12 London, while GBPNZD often needs until 22:40.

That difference matters for anyone who trades the early Asia session. Placing the same entry at 22:15 on both pairs means one of them is still paying a spike price.

Month end is worse than a normal night in my log. The last rollover of the month shows the widest spreads I record, roughly double a mid month night on the crosses.

Wednesday is a separate trap even for a scalp book. Triple swap is booked that night, so a swing position someone forgot to close pays three days of carry on top of the spread spike.

I keep a small table of recovery times per pair and refresh it each quarter. It also flags the clock change weeks on its own, because every time in the table shifts by an hour.

Do you note recovery times per pair, or use one fixed restart time for everything? A single time across all pairs is what got me in trouble.