A spread to ATR filter that quietly removes bad trades

[Fairman]

A spread to ATR filter that quietly removes bad trades

A simple rule that helps many scalpers: compare the current spread with the recent average true range on the timeframe being traded. If the spread takes up more than a certain fraction, say 10 to 15 percent of the typical one minute range, skip the trade.

The reasoning is plain. When the market is moving 6 pips a minute, a half pip spread is a minor cost. When it is moving 1.5 pips a minute, the same spread is a third of the potential move.

Setting the threshold requires testing. Too strict and the strategy rarely trades. Too loose and the filter does nothing.

A nice feature is that the filter adapts automatically. It lets the strategy trade freely when the market is active and pulls back when conditions are poor, without any clock-based rules.

It also protects against rollover, news spreads, and thin sessions without needing a list of times.

A starting point many traders use is a ratio of the spread to the average one minute true range over the last twenty bars. If the ratio exceeds roughly one tenth, skip. Testing values from 0.05 to 0.2 on a few months of data shows where the knee of the curve is for the specific pair and style.

What ratio of spread to range has worked as a cutoff?

[Fairman]

My version: calculate the 14 period ATR on the 1 minute chart, and the current spread. If the spread is more than 15 percent of the ATR, skip the trade. On EUR/USD in the London session, ATR is around 2 pips and spread around 0.2, so the ratio is about 10 percent, and trades go through. Around rollover, ATR might be 1 pip and spread 2 pips, ratio 200 percent, and trades are blocked. The filter removed about 8 percent of my trades and improved the profit factor from 1.18 to 1.27.

[Amanhasnoname]

I use a version of this and my reason for adopting it was a very ordinary week of losses. I went through 200 trades and found that the ones taken when the spread was above about a fifth of the one minute ATR had a profit factor of 0.8, while the rest sat close to 1.3. The filter did not need to be clever, it only needed to be applied every time.

My setting is a ratio cutoff of 18 percent, with the ATR taken over the last 20 bars on M1. On GBPUSD, in the London morning, the ratio sits at around 9 percent, so nothing is blocked. Right after 5pm New York it often reads 40 or more, and the blocked trades there were the ones I used to regret.

One thing I added is a cooldown. If the ratio fails, I wait for three consecutive bars with a passing value before I allow entries again, because a single good reading during a spike misled me twice.