What news still surprises you on EURUSD?

[LondonScalper]

After two decades I like to think the calendar holds few surprises for me. Then a Wednesday like 9 September happens and I'm reminded that it isn't the scheduled data that catches me out, it's everything around it.

So I made a list of what still surprises me on EURUSD, from my own notes over the past year.

Speeches by ECB board members that aren't on my calendar, or that are listed as a minor event. A comment about the inflation outlook at a conference in some smaller European city can move the pair 15 pips while I'm looking at a quiet chart.

Revisions. The headline US number comes in close to forecast, the pair goes nowhere for ten seconds, then someone reads the revision to the previous month and it moves 25 pips.

Month-end flows around the 16:00 London fix. Not news exactly, but the moves at 15:55 on the last trading day of some months have no obvious cause and can be bigger than the morning's data.

German state CPI releases on the morning of the national flash. Small releases one after another, and by the time the national number arrives the market has already priced most of it.

What I've stopped being surprised by is NFP. It moves, I'm flat, and I come back after 14:00.

[Shadow Trader]

Revisions are the one on your list that still gets me as well. The ten seconds of nothing followed by the real move is almost worse than an outright miss, because by then you've relaxed.

I'd add "sources" stories to the list. After an ECB meeting, headlines quoting unnamed officials can land at any point in the afternoon, and I've seen EURUSD move 20 pips on one at 15:40 with nothing on any calendar.

US Treasury auctions are another one. They are scheduled, so technically not news, but a weak 10 or 30 year auction around 18:00 London can push the dollar more than the morning data did, and most FX calendars barely flag them.

On the German state CPI point, my experience is a bit different. Some months the states point one way and the national flash still surprises, because the states that report first aren't always representative.

Speeches are the hardest for me to plan around. Do you keep a separate list of who is speaking each week, or do you just accept that a few of them will catch you?

[Fairman]

Month-end flows around the 16:00 London fix.

On the month-end one, the moves are less causeless than they look, and the direction can sometimes be estimated in advance. A lot of it is rebalancing by funds that hedge their foreign equity holdings. If US stocks have outperformed over the month, a European fund holding hedged US shares now has more dollar exposure than its hedge covers and needs to sell dollars to top it up, and the reverse when US stocks lagged. That rebalancing is done around the WM/Reuters 16:00 London fix, which is why the move sits at 15:55 to 16:05. Several banks publish month-end estimates for exactly this. It isn't reliable enough to trade blind, but comparing the month's S&P and Euro Stoxx returns in the last week at least tells you which way a surprise is more likely to go.

[Shadow Trader]

Fairman, your month-end rebalancing note is the bit that made the 15:55 moves stop feeling random. Comparing the month's S&P and Euro Stoxx returns in the last week is crude, but it at least points which way a surprise is more likely to lean before the WM fix.

I still do not trade the fix itself. What I do now is treat the last trading day as a blackout from 15:45 to 16:15 whenever equity performance that month was clearly skewed, which is most months.

Bank month-end estimates help as a sanity check, not a signal. If my S&P versus Stoxx read and the bank note disagree, I just stay flat and let the window pass.

LondonScalper, on the speech side I keep a short Executive Board list only, two or three names a week, and ignore the rest. Anything wider and I stop respecting it.

Curious whether the bank estimate has ever been opposite your own equity skew read in a way that cost you if you leaned one side. I have only used the disagreement as a flat signal so far.