The wide-open stretch after the NY close
Posted: Fri Oct 30, 2026 11:17 am
The wide-open stretch after the NY close
The late hours of the American day have an odd structure. Equity markets close at 4 p.m., banks wind down, and the foreign exchange market loses a big piece of its volume. What remains is a mix of Asian-Pacific early birds and algorithmic flows.
Volatility metrics sometimes drop to their lowest of the day. Spreads, though, rise. The result is an unfavourable ratio of cost to opportunity.
Trading in this window demands more discipline than at the peak. Small targets become hard to hit, because the typical range is narrower than the spread plus slippage. The strategy that looked reliable at 14:00 GMT falls apart at 21:00.
Backtests that ignore time-of-day cost differences make this window look as good as any other. It is not.
The simplest correction is to exclude the hours with the worst cost-to-range ratio. A strategy with a smaller number of trades often shows better net performance.
How many trades in the sample were taken in the lowest-quality hours, and what happened to the results when those were removed?
The late hours of the American day have an odd structure. Equity markets close at 4 p.m., banks wind down, and the foreign exchange market loses a big piece of its volume. What remains is a mix of Asian-Pacific early birds and algorithmic flows.
Volatility metrics sometimes drop to their lowest of the day. Spreads, though, rise. The result is an unfavourable ratio of cost to opportunity.
Trading in this window demands more discipline than at the peak. Small targets become hard to hit, because the typical range is narrower than the spread plus slippage. The strategy that looked reliable at 14:00 GMT falls apart at 21:00.
Backtests that ignore time-of-day cost differences make this window look as good as any other. It is not.
The simplest correction is to exclude the hours with the worst cost-to-range ratio. A strategy with a smaller number of trades often shows better net performance.
How many trades in the sample were taken in the lowest-quality hours, and what happened to the results when those were removed?