Why the first minute of the hour feels different

[Fairman]

Why the first minute of the hour feels different

Look at a one-minute chart for a few weeks and mark the first candle of each hour. Something shows up: larger range, higher tick count, more false signals.

The reason is a cluster of mechanical orders. Algorithms scheduled on the hour. Option expiries cut at round times. Rebalancing programs that wake up at 00 minutes. News headlines timed for the top of the hour. A burst of orders hits the same minute, and price jumps around.

Scalpers using indicators on short timeframes often notice that signals around the top of the hour have worse results than those in the middle of the hour. The indicators are not wrong. The market is simply noisier.

One approach is to blank out the first two minutes of every hour from the strategy and see what changes. The answer depends on the style. Mean reversion setups usually improve. Breakout setups sometimes get worse, because some of the best moves start there.

Backtests with a minute-of-hour filter are quick to run and sometimes surprising.

Does performance differ between the first minutes of the hour and the rest, and in which direction?

[Shadow Trader]

Spread logs sometimes show small bumps at :00 and :30 every hour. Not as dramatic as rollover, but visible once someone plots spread against minute-of-hour across a few weeks.

Part of the cause is scheduled data. Many economic releases are timed at the hour or half hour, and liquidity providers pull back a little in advance. Part of it is behavior: algorithmic traders and fixing-related flows cluster around round times. And part is simple staffing patterns on trading desks.

For a scalper entering at 9:59:55, the spread at the entry may be normal, while the spread at the exit at 10:00:10 is noticeably worse. The trade looks fine on the chart and loses money on the statement.

One easy experiment: log the spread every second for a week on the pair of interest and average by minute-of-hour. A pattern usually emerges.

Strategies with short holding times, say under thirty seconds, can end up in these windows more often than expected, especially when the trade frequency is high.

Another source of top-of-hour bumps is the platform's own clock. Brokers whose servers run on a minute-aligned schedule may update markups at :00, so a spread jump can be a pricing-engine change rather than a market reaction. Overlaying the bump on a plot of true market volatility shows whether it coincides with real activity or with the broker's housekeeping.

Which minutes of the hour show the biggest spread bump on your pair?