Session Overlap Trading: Why Most of Your Volatility Happens in a Few Hours

A 24-hour market doesn't move at a constant rate for 24 hours. Most currency pairs do the bulk of their daily range in a two-to-four hour window, and are comparatively flat for the rest. A strategy's statistics computed across the whole day are really an average of two different markets — a fast, liquid one and a slow, thin one — and averaging them together hides which one the strategy actually depends on.

Where the volatility actually comes from

Forex trades through overlapping regional sessions — Tokyo, London, New York — and each one opens with a different set of participants active. Volume and volatility rise when two sessions are open at once, because that's when the largest number of banks, funds and retail participants are pricing the same pairs simultaneously. The London–New York overlap, roughly 13:00–16:00 UTC, is the single busiest window for most majors — it's common for a pair to cover a third or more of its entire daily range inside those few hours. The Tokyo session and the early London session, by contrast, are usually thinner and choppier, with wider effective spreads relative to the moves available.

This isn't a minor scheduling detail. It means the market a strategy is trading changes shape several times a day — different liquidity, different spread cost relative to typical move size, different tendency to trend versus chop — while the strategy's rules stay the same.

Why a strategy's numbers can hide this

A backtest or a journal summary that reports one win rate and one expectancy for a strategy is implicitly assuming the strategy performs the same way regardless of when it's traded. In practice it's common for a strategy to be strongly profitable inside the overlap and roughly breakeven or negative outside it — two very different edges blended into one number that looks moderately positive overall. Traded only in its good window, the strategy might be excellent. Traded around the clock because the setup “still looks valid,” it's carrying a chunk of trades with a different, weaker expectancy than the summary statistic suggests.

The same split shows up in cost, not just in edge. Spread and slippage are close to fixed in pip terms but the average move available to absorb them isn't — during a thin session the same spread eats a much larger share of the typical range, which quietly lowers expectancy on every trade taken there even before the setup quality is considered.

What to check in your own numbers

• Win rate and expectancy split by hour of entry (UTC), not just totalled across the day — the overlap window and the quiet windows should be looked at as separate samples.
• Average realized spread cost as a share of the stop distance, by session — a stop that's comfortably wide relative to spread during the overlap can be tight relative to spread during the Asia session.
• Trade frequency by hour — a strategy that takes as many trades in a quiet session as in the overlap is very likely taking marginal setups just to stay active, not because the setups are equally good.
• Whether losing streaks cluster in a specific session— a strategy failing consistently outside its best hours looks, from the trade log alone, indistinguishable from a strategy that's simply having a bad run.

None of this requires a different strategy. It requires knowing which hours the current one actually works in, and either restricting trading to them or holding the outside-window trades to a stricter standard.

Why this is easy to miss by hand

Splitting a trade log by entry hour and recomputing expectancy for each slice is exactly the kind of check that's simple in principle and rarely done by hand, because it means re-running the same math several times instead of once. getALPHA syncs closed trades directly from MT5 with their entry timestamps intact, so that split is a filter, not a recalculation, and the AI coachcan surface it in a process review — whether a strategy's losses are concentrated outside its usual overlap-hour trades — rather than leaving the pattern to be noticed by feel after enough of it has already happened. The session briefcovers the other half: what's scheduled today and where the calendar itself points to elevated volatility, before the session starts.