I scalp USDJPY mostly between 00:00 and 03:00 London time, so Tokyo morning. The broker I use for London is fine at 08:00, but during Asia its yen pricing gets noticeably worse.
What I've noticed when comparing a few brokers on small size:
Server location matters a little. One broker offers a Tokyo or Singapore server option. Fills were slightly more consistent, but the difference was smaller than I expected.
Spread around the Tokyo fix. Near 09:55 Tokyo time some brokers widen sharply for a minute or two. Worth logging.
Liquidity at the daily rollover. Asia starts right after rollover for brokers on New York close. The first 30–60 minutes can be terrible on all of them.
My shortlist criteria for Asia yen:
Average spread between 00:30 and 03:00 London time, measured myself.
Slippage on stops during the Tokyo fix.
Whether the broker has Asia-based LPs or just quotes from European ones overnight.
I'm not naming brokers yet because my sample is small. If you trade yen in Asia hours, what do you look at, and have you found a meaningful difference between brokers or is it mostly the same pool?
Sleep loss shows up in trading in ways that rarely get mentioned. A short list of the less obvious ones:
- Slower recognition of familiar patterns. The setup is on the screen but takes an extra second or two to register, which on a scalp can be the entire edge.
- Reduced tolerance for being wrong. Small losses feel bigger, so exits get earlier and more emotional.
- Worse estimates of elapsed time and of how much has been won or lost in the session.
- A pull towards stimulating, high-volatility moments, because quiet markets feel unbearable when tired.
- Misclicks and wrong-size orders, which tend to turn up late in a night session when nobody is checking.
Trading the Asian session from a western time zone, or sitting up for the New York close from somewhere much further east, turns this into a regular pattern rather than a one-off. The session itself might suit the strategy fine and still be the wrong session for that particular brain at that hour.
Some people find a hard cutoff rule helps: no new entries after a certain number of hours awake, however good the chart looks.
How do you handle sessions that fall outside normal waking hours, and do you notice a clear drop in quality?
