EURUSD spikes to 1.0920 in a thin ten minute burst, more than double the pair's normal ten minute range, then gives almost the entire move back within twenty minutes. That kind of overextension, price traveling further and faster than the market can actually sustain, is the entire premise behind mean reversion: fading it, betting the snap back continues, is one of the higher probability trades available on a range day. It is also one of the fastest ways to blow an account on a trend day, which is why the conditions matter more than the trade itself.
Mean reversion is not a single setup so much as a filter applied to an extreme: is price far enough from its recent average, at a real level, in a market that is not actually trending. Get any one of those three wrong and the fade turns into catching a falling market with both hands.
When fading actually works
Fades belong in range conditions, where price has spent recent sessions oscillating between two levels rather than making new highs or lows. A market like that has a proven ceiling and floor, and an extreme move toward either edge is more likely a stretch than the start of a breakout. In a genuine trend, the same extreme reading is often just the trend continuing, and fading it means selling strength in an uptrend or buying weakness in a downtrend, a losing habit dressed up as a strategy. Confirming range conditions first is not optional here.
Defining extreme with actual tools
Extreme needs a number, not a feeling. Distance from a moving average measured in ATR is one honest gauge: price trading three or more average true ranges away from its 20 period moving average on the working timeframe has stretched further than its typical rhythm allows. RSI adds a second read, with readings above 75 or below 25 flagging momentum that has outrun itself, though RSI alone gives too many early signals to trade without support. Bollinger Bands add a third: a close outside the outer band, especially the first such close in a while, marks a statistically unusual move on that instrument's own recent volatility. The strongest fades stack two or three of these with an actual support or resistance level nearby, the same confluence logic that strengthens any level based trade, rather than firing off one reading in isolation.
The net: risk rules for fading
Fades need smaller size and tighter invalidation than trend trades, because the trade bets against the most recent momentum, and being wrong about direction here tends to be wrong immediately and sharply rather than slowly. A sensible net: risk half your normal size on a fade, place the stop just beyond the extreme itself rather than an arbitrary distance, and treat any new extreme beyond your entry as invalidation rather than an invitation to add. Most fades that fail do so quickly and visibly, and a trader who exits fast on that signal loses far less than one who holds hoping the snap back is merely running late.
A worked fade on Nasdaq
Nasdaq has ranged between 19,150 and 19,350 for four sessions, no new highs or lows, a textbook range day. Price spikes to 19,362, twelve points above the range high, on a thin one minute burst with no news behind it, closing outside the upper Bollinger Band and printing an RSI of 81. The trigger is a rejection candle back inside 19,350 within the next two candles. The stop sits at 19,370, twenty points above entry near 19,350, just beyond the spike's own extreme. The target is the range's midpoint near 19,250, a hundred points of reward against a twenty point risk, 5 to 1, typical of how a good fade's reward to risk looks precisely because the entry sits so close to the actual extreme. On $10,000 risking 1 percent, $100 divided by 20 points times $1 gives 5 contracts.