After a sudden spike of 15 pips in a minute, price often retraces some of it. It's one of the most commonly cited tendencies in intraday trading, and a favourite basis for scalping strategies.
The logic is sound. Extreme moves in thin conditions may overshoot fair value, and liquidity providers step in to fade them. But the tendency isn't universal. Moves driven by genuine information, like news, may continue.
A good study separates spikes by cause. Spikes without news, in quiet hours, show stronger reversion. Spikes after data releases show weaker reversion, and sometimes continuation.
The definition of "extreme" matters. Using a multiple of recent ATR or standard deviation adapts to current volatility. A fixed pip threshold doesn't.
Execution is the other problem. Fading a spike means entering as price is moving fast, with wide spreads and high slippage. The retracement may be profitable on a mid-price basis and unprofitable after real costs.
Testing with tick data and realistic execution assumptions is crucial before trusting a reversion strategy.
A sanity check is looking at how many spikes occurred in total, and how the profitable ones are distributed. If most of the profit comes from a handful of spikes in a few unusual sessions, the strategy depends on rare events and may not be reliable.
How is an extreme move defined, and what's the measured reversion rate after costs?
Mean reversion after extreme short-term moves
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
-
FXS Prop Desk
- Posts: 1689
- Joined: Wed Oct 07, 2026 9:52 am
Re: Mean reversion after extreme short-term moves
Fifteen pips in a minute in a thin market and fifteen pips in a minute after a data release are different animals. The first tends to retrace as liquidity returns; the second is new information and often continues. Splitting the mean-reversion results by "release in the last five minutes: yes/no" usually shows the edge lives almost entirely in the first group.
FXS Official Team | The Idea Lab & Strategy
-
Amanhasnoname
- Posts: 2060
- Joined: Fri Oct 30, 2026 9:45 am
Re: Mean reversion after extreme short-term moves
I ran something close to this on EURUSD and GBPUSD for most of last winter, and the split by cause made all the difference. My trigger was a one minute range bigger than three times the average of the previous thirty minutes, and I logged every case in a spreadsheet with a column saying whether anything was on the calendar within ten minutes either side. The quiet hour spikes, mostly in the gap between New York closing and Tokyo getting going, gave back about half the move fairly often. The ones around data mostly kept running or chopped sideways.
The cost side was what nearly killed it. On paper the fade looked lovely from the mid price, but my fills in those moments were often a pip or more worse than the screen, and the spread was double its usual width. Once I put my real fills into the sheet, a lot of the apparent edge simply vanished.
What survived was a much narrower version. I only fade quiet hour spikes, I wait for the first candle after the spike to close inside its range before entering, and I accept fewer trades in exchange for not chasing a moving price.
The cost side was what nearly killed it. On paper the fade looked lovely from the mid price, but my fills in those moments were often a pip or more worse than the screen, and the spread was double its usual width. Once I put my real fills into the sheet, a lot of the apparent edge simply vanished.
What survived was a much narrower version. I only fade quiet hour spikes, I wait for the first candle after the spike to close inside its range before entering, and I accept fewer trades in exchange for not chasing a moving price.