The 4pm London Fix and Why the Window Got Wider

[Shadow Trader]

The WM/Refinitiv benchmark set at 4pm London is a reference rate for a huge amount of portfolio valuation and index-tracking activity. Historically the rate was calculated from a one-minute window of trades and quotes. After manipulation scandals, the calculation window was widened to five minutes, running from two and a half minutes before the hour to two and a half minutes after.

The wider window changed the shape of the flow. Instead of one violent spike, executing banks tend to spread client fix orders across several minutes, so price often shows a steady, one-directional push rather than a sharp wick. A scalper watching one-minute candles can easily mistake this persistent pressure for a genuine breakout.

Asset managers usually place fix orders in advance, which means a bank takes the other side and manages the resulting risk around the window. That hedging can create momentum shortly before 3:57pm and sometimes a partial retracement once the benchmark flow is finished.

Spreads and depth can behave unusually inside the window, and the direction of the move often reflects the net client imbalance, which is invisible on a chart. Treating the period as distorted price discovery rather than a clean signal is the more sensible framing, especially for tight stop placement.

[FXS HongKong Scalper]

Good summary of the change. Widening the calculation window to five minutes either side of 16:00 London spreads the move out, as you say. Anyone wanting to see the effect can pull M1 data for USDCAD and EURUSD on the last business days of a few months and check when the move begins relative to 15:57:30 and whether it fades after 16:02:30; pre-positioning by banks would show up as moves starting before the window opens.

For Asia-based traders, the fix is late evening, so most of us avoid it, but it matters for anyone holding positions into it at month end. The direction of month-end rebalancing flows is sometimes estimated in advance by banks, but those estimates are unreliable.