A Session Average Anchored At Midnight Measures The Wrong Session

A volume weighted average that resets once a day is meant to describe one session: the range of prices that session's own participants have traded, weighted by how much volume each price attracted. That description only holds if the reset happens where the session itself starts. On most instruments, resetting at server midnight instead does not do that. Two different clocks A broker publishes a trading-hours schedule for each symbol, one entry per weekday, because a session boundary is a market fact, not a clock reading: it is set by when trading in that instrument actually opens and closes. Server midnight is something else entirely, the point where a server's own internal day counter rolls over. On most brokers the two moments do not coincide, sometimes by minutes, often by hours. Any calculation that resets once a day has to pick one of the two. Server midnight is the boundary every terminal already tracks by default, which makes it the boundary that gets picked when nobody has specifically arranged to read the symbol's own schedule instead. What the mismatch does to the number Where the reset lands partway through what the broker still counts as the previous session, the bucket that opens at midnight starts by inheriting whatever volume printed between midnight and that session's true close. Those prints belong to the session that just ended, not the one that is starting. They get folded into the new average anyway. In the early part of the new day, before much volume has traded under the new session, that inherited tail can make up most of what the number is built from, a session average that is mostly measuring a session that already finished. The size of the distortion follows the size of the gap between the two clocks. A symbol whose session happens to open close to midnight is barely affected. One whose broker sets the session boundary several hours away from midnight can carry a meaningful stretch of the wrong session into every reading taken early in the new day. Checking it against your own broker The check itself needs nothing beyond what a broker already publishes: its trading-hours schedule per weekday for the symbol in question. Look up where that schedule places the session open for the weekday you trade, and compare it to midnight on the server clock. A gap of a few minutes means a midnight anchor is a close enough stand-in. A gap of an hour or more means the two are marking different boundaries, and any average that resets at midnight is carrying part of the session before it into the first hours of the new one. The exception worth keeping in mind Not every symbol has a published session schedule to check against. Where none exists, server midnight is not an approximation standing in for something better, it is the only boundary the broker has given a terminal to work with, and falling back to it is the correct choice rather than a shortcut. The distortion above only applies where a narrower, real session boundary exists and a calculation ignores it in favor of the clock that happens to be easiest to read. A free indicator, Session Value Desk, reads each symbol's own trading hours before drawing its session line, and only falls back to the server day where no schedule is published for that symbol, so this choice is made once, from the broker's own data, instead of left for a reader to check by hand.