FXS Prop Desk wrote: No more than two fades in the same direction per session.
That rule probably does more work than any of the entry filters. The expensive version of fading is rarely the first trade, it's the third and fourth into a move that has quietly turned into a trend day, each one with the stop beyond a new extreme. One more filter that's easy to check in real time: if the session range is already bigger than its typical size for that window, say the 20 day average, by the time you want to fade, the flat higher timeframe condition probably isn't true anymore even if the H1 EMA hasn't caught up. EMAs lag by design, so on the morning a trend starts they can still look flat for a while.
The late New York and early Asia window has one trap worth writing into the rules. Spreads on EURUSD and GBPUSD widen for several minutes at the 22:00 London rollover, and a fade entered just before it can be stopped by spread alone.
I stop fading from about 21:45 to 22:15 London, and a little longer on GBP pairs. Range touches in that period also tend to be spread spikes rather than real tests, so they do not count toward the two touches either.
On the target, I take profit at the middle of the range rather than the far side. The far side gets reached often enough to tempt you, but the middle gets reached far more, and that win rate is what keeps a run of small stops survivable.
Fairman's session range point is the one I would make a hard filter. Once the range so far passes about 80 percent of its usual size for those hours, I stop fading for the day.
The filter that has saved me most is simpler than any of these. If the second fade in a direction is stopped, the session is done for fades in either direction, because whatever is moving the market is not finished.
Numbers on the middle of range target show how tight the entry must be. In a 12 pip Asian range on EURUSD the middle is 6 pips from the edge, so with the stop 3 pips beyond the extreme, an entry 2 pips inside the edge already drops the trade to about 0.8R.
That is why I only fade with a limit order resting at the edge itself. Chasing a fade after the turn has started looks fine in hindsight but bleeds over a month.
On FXS Prop Desk's measures of extended, I have had more luck with distance from session VWAP than with a deviation band. My threshold is 1.5 times M5 ATR from VWAP, and below that I treat a touch of the range edge as unfinished.
The two touches each side condition is the one I bend for Asia. By the time a quiet range has printed four clean touches it is often past 02:00 London, and the remaining hours rarely have enough movement left to reach the middle twice.
So I accept one touch per side if the range is already at least 10 pips wide and the H1 slope is flat. Does anyone else loosen that rule late in the session, or is the second touch worth the missed trades?
Mean-reverting series pull back toward an average, and the speed of that return can be quantified. The half-life estimates how long it takes for a deviation to shrink by half. For intraday forex, it provides a guide to holding periods and parameter choices.
Estimation commonly uses a regression of price changes on the lagged deviation from a moving mean. The coefficient implies a decay rate, from which the half-life follows. Results vary by pair, session, and period, and many series show little or no reversion at certain scales.
A half-life of a few minutes suggests that short holding times fit, while a half-life of hours suggests that scalping-style exits would cut trades before the reversion completes. If the estimate is unstable across periods, the reversion effect may not be dependable. Testing different lookback windows and verifying out-of-sample consistency avoids relying on statistical artifacts. The measure helps match strategy design to the actual time scale at which price behaves.