Rejections cluster around the same minutes
Order rejections do not spread evenly through the day. Pull the failed-order log for a month and plot them by minute. Spikes show up at the top of the hour, at the 30-minute marks, and a few seconds after scheduled data releases.
Why the top of the hour? Banks and algorithms rebalance on round times. Quote streams get busier. Some liquidity providers widen or pull quotes in the seconds around fixings, and the broker's bridge starts rejecting orders it cannot price safely.
The effect is small in count and large in damage. A scalper who has a signal at 14:00:03 and gets rejected twice has missed the best part of the move, and the third attempt fills ten ticks worse.
Knowing the spikes, the response is not to avoid them but to price them in. Either skip entries in those few seconds, or use limit orders that do not care about rejection, or accept the cost and test whether the strategy survives it.
None of it requires a fancy tool. A spreadsheet with a timestamp column does the whole job.
Do rejection spikes line up with the same clock times on a second broker, or are they unique to one bridge?
Rejections cluster around the same minutes
Rejections cluster around the same minutes
It’s Fairman 