News days versus non news days in my results, a split I'd never looked at

[Fairman]

News days versus non news days in my results, a split I'd never looked at

I tagged every trading day in my journal as a news day or a quiet day, depending on whether there was a major release for the pairs I trade. After a couple of months, I split my results to see what it showed.

I assumed I'd do worse on news days because of the volatility, and I was partly right. But the more interesting thing was how I behaved. On news days, I had more rule breaks and more trades outside the plan. I got excited, or I got caught out by a spike. On quiet days, I was calmer and my results were steadier.

The money result was lower on news days, even though I'd planned to stand aside during the releases. The damage was mostly done in the hours after, when I was trying to catch the move, or before, when I was nervous.

That's led to a change. On major news days, I now trade a smaller size and a shorter window, or only in the hours before the release. On very big days, I may not trade at all. I'll keep tracking to see if it helps.

[Fairman]

I did the same split and found the same thing. The damage was not during the release, it was the hours around it when I was chasing the reaction or jittery in the lead up. What helped was making news days lighter by design: fewer trades, smaller size and a firm stop time. On quiet days I trade my normal plan. Seeing the two groups separately showed that the problem was behaviour, not the market. Planning the lighter version of those days ahead of time is what made the difference for me.

[Fairman]

Worth adding a tag for the type of news as well. For me, central bank days and jobs data behaved very differently from smaller releases. On the big ones I am now flat for most of the day, and on the smaller ones I just avoid the half hour around the release. Splitting it that way showed that most of the damage came from a handful of big events, not news days in general. Knowing that, I can plan the few heavy days in the month instead of treating every release the same way.

[Shadow Trader]

Splitting by day type is useful, but the split hides a confound. Payrolls land on Fridays and FOMC on Wednesday evenings in London terms, so a news day bucket is partly a day of week bucket, and Friday afternoon behaviour is its own problem.

A cleaner cut is by hours rather than whole days. Tag each trade by its distance from the nearest Tier-1 release, for example more than 2 hours, 30 to 120 minutes, and inside 30 minutes, then compare the three groups.

Counting rule breaks per day type is the other number worth having. Say you review 40 sessions with 12 news days and find 9 off plan trades on those 12 days against 6 across the other 28. That is about 0.75 per news day versus 0.21 on quiet days, and it says more than the money result does.

The lighter news day plan works best when the stop time is a clock time. A rule like done by 12:00 GMT is easy to obey, while done once the move settles never really ends.

Do you count the morning after a big release as a news day too? A lot of the restless trading seems to spill into the next session.

[FXS Prop Desk]

What did the quiet days show? Many traders find that they do worse on quiet days than they expect, because they overtrade when nothing is happening.

If both types of days have weaknesses, different rules for each can help. For example, fewer trades and smaller targets on quiet days, and longer waiting times after releases on news days.

[Shadow Trader]

On the quiet day question above, one number shows overtrading quickly. Divide the day's range by trades taken, so a quiet 45 pip day with 4 trades is roughly one trade per 11 pips of movement, while a 90 pip news day with the same 4 is one per 22.

If quiet days show twice the trades per pip, the plan is being forced onto a market that has less to give. A cap of 2 trades when the Asian range is under a set size handles that without a separate rulebook.

The release tags need care over the next two weeks because the clocks change on different dates. A US release at 8:30am New York is 1:30pm in both Lagos and London today, still 1:30pm Lagos but 12:30pm London from Monday 26 October, then 2:30pm Lagos and 1:30pm London from Monday 2 November.

Anyone tagging trades by distance from the release with a fixed local time will mislabel that whole week. Taking the release time from the calendar in UTC and converting it day by day avoids the problem.

Is the news versus quiet split also run by session? A quiet day in London can still turn into a news afternoon once New York data lands.