Most traders who use Fair Value Gaps treat them the same way regardless of the time of day. Find the gap, enter the trade, manage the exit. The session is an afterthought.
After backtesting 4,476 FVG trades on the US30 across six and a half years of M5 data, I can tell you the session is almost everything.
The idea behind this analysis was straightforward: test the FVG retest strategy — waiting for price to return and fill the gap rather than chasing the initial impulse — across three high-probability institutional windows. London open, where European money moves the market at the start of the day. The New York open, where the highest volume and most volatility converge. And the late US session between 7pm and 9pm GMT+2, when the afternoon institutional cleanup trade tends to play out quietly in the background.
The results were not what I expected.
The New York open, the session most traders associate with the biggest moves and the clearest setups, produced a profit factor of 0.99 across 1,615 trades. Breakeven before costs. The noise at the open overwhelms the signal — FVGs form but the directional momentum of the session crushes the retest play before it has a chance to develop.
London was marginal. A profit factor of 1.08 over 1,325 trades is positive but thin. Not enough edge to size confidently.
The late US session is a different story entirely. 1,536 trades, 41.6% win rate, profit factor 1.22, and +9,674 cumulative points over the test period. The structural reason makes sense once you see it: the 7–9pm window is where market makers are filling residual institutional orders after the US afternoon close, and price regularly reverts to value levels — including the FVGs left behind by the earlier session's impulse moves.
The filters that made the difference were a minimum gap size of 8 points, an impulse body requirement above 50% of the candle's full range, and alignment with the higher-timeframe EMA50 bias. Nothing exotic. The retest entry — placing a limit order at the gap boundary rather than entering at market after the FVG forms — was the single biggest variable. Without it, the same signals produced a profit factor of 0.81 and a total loss of over 17,000 points.
Below is the full breakdown, with equity curves across all three sessions and a month-by-month P&L for the late US setup.
Cumulative Points — 1 Contract · FVG Retest Entry · Jan 2020–May 2026
Late US is the edge
19:00–21:00 GMT+2 produces +9,674 pts over 6 years at PF 1.22. Institutional reversion after the US afternoon session — the cleanest order flow window.
NY Open FVGs don't work
The NY open (14:00–16:30) is too noisy. FVGs form but don't follow through — too many conflicting flows at the open. PF 0.99 = breakeven before costs.
Retest entry is everything
Immediate entry after FVG forms: PF 0.81, −17,799 pts. Waiting for price to re-enter the gap: PF 1.09–1.22. Same signal, completely different outcome.
Monthly P&L — Late US Session (pts)
Annual Late US P&L
Filter Stack (Late US)
What is a Fair Value Gap?
A 3-candle pattern where rapid institutional movement leaves an unfilled price gap between the high of candle 1 and the low of candle 3 (bullish FVG) or vice versa.
Price tends to return to fill these gaps, offering a defined entry with a tight stop at the gap boundary.
ENTRY: Retest of gap top/bottom
STOP: Far side of the gap
TRAIL: Activates at 1R, trails 2R behind peak