One 40 pip gap bar in ATR(14): what each platform does with it afterwards

[Fairman]

One 40 pip gap bar in ATR(14): what each platform does with it afterwards

Suppose M5 true range is 4 pips bar after bar, and then Monday's first bar opens on a gap and has a true range of 40.

A simple average of the last 14 true ranges, which is how I understand the stock MT4 and MT5 ATR works, jumps to 4 plus 36/14, which is 6.57, on that bar. It stays there for 13 more bars, and on the 15th bar after the gap it drops back to exactly 4.00 in one step.

Wilder smoothing, the RMA that TradingView's ta.atr uses, jumps to the same 6.57 and then decays gradually: 5.78 after 5 bars, 5.23 after 10, 4.91 after 14, and 4.85 after 15. It approaches 4 and never reaches it.

Effect on a stop of 1.5 times ATR. Normal: 6 pips. Right after the gap: 9.9 pips, 64% wider. After 14 bars, the simple average version is back to 6 pips (it fell off the cliff), while the Wilder version is still at 7.4.

So the same strategy on two platforms is using a 6 pip stop on one and a 7.4 pip stop on the other at that moment. If you backtest on one and trade on the other, the stops differ after every gap. To see it on a live chart, read ATR(14) on an M5 chart across a Monday open at the gap bar and at 1, 14, 15 and 16 bars after it, and compare with the figures above. Verify which one your platform uses, since I may be wrong about a specific build.

[FXS HongKong Scalper]

Very clear comparison. The cliff edge drop on the 15th bar with the simple average is the part that catches EAs out. If stops are set from ATR, a trade opened 14 bars after the gap gets a much wider stop than one opened a bar later, purely from the averaging method. I prefer Wilder smoothing for that reason, or capping each true range at a multiple of the median before it enters the average, so one gap bar cannot distort the next hour.

[Fairman]

A direct way to see which version you have. Take a chart where a large gap bar happened long ago, or feed a test series with 4 pip ranges and one 40 pip bar, then look at the ATR value 15 bars after it. If it's exactly back to the baseline of 4, it's a simple average of true ranges. If it's still above, around 4.85, it's Wilder style smoothing. You can do it on a spreadsheet in five minutes, and then compare the number with what your platform shows on the same data. It's better than relying on what the documentation or a forum post says.

[Fairman]

Volatility filters react differently. Suppose a rule says trade only when ATR(14) is below 6 pips. With the simple average, the gap bar keeps ATR at 6.57 for 14 bars, so the filter blocks trading for 14 bars, about 70 minutes on M5. With Wilder smoothing, ATR falls under 6 on the fourth bar after the gap (6.06 on the third, 5.91 on the fourth), so the filter blocks for only about 4 bars, 20 minutes. Same rule, same data, 70 minutes against 20 minutes of sitting out. The Monday open is exactly when many scalpers would like to know which of those they're getting.

[Fairman]

A median avoids the problem. The median of the last 14 true ranges stays at 4 with a single gap bar, since one outlier doesn't move the middle value. It's a less smooth indicator and it moves in steps, but it ignores one-off events like a weekend gap or a spike on a news release. A second option is to leave out the first bar of the week from the calculation. Both are easy in Python or in a custom indicator. The cost is that a genuine volatility increase takes longer to register, since several large bars are needed to move the median, so it's a trade off.

[Fairman]

Position sizing from the stop makes the difference concrete. With $100 at risk and a pip valued at $10 per lot, a 6 pip stop gives 1.67 lots and a 9.9 pip stop gives 1.01 lots. So the bar after a gap sizes the position 40% smaller than normal. Over the next 14 bars, the Wilder version shrinks that reduction gradually, while the simple average version removes it all at once. If your sizing formula depends on ATR, you'll see the lot size jump on the day the gap bar leaves the window. That's a real jump in exposure, caused entirely by a bar that happened 70 minutes earlier.

[Shadow Trader]

The one-minute ATR is a quick reality check that gets skipped a lot. If the 14-period ATR on the one-minute chart reads around 4 pips, a 10 pip target needs roughly two and a half candles of movement in one direction with nothing in the way. That can happen, but it needs a push, not drift.

At the other end, a 3 pip target against that same ATR is a small fraction of one candle, which sounds easy until spread and slippage eat a third of it.

The useful idea is to express targets as ATR multiples instead of pips. A target of one to two times ATR is reachable in normal conditions. Beyond three times usually needs a trend leg or news. The multiple stays valid when volatility changes, while a fixed pip number quietly becomes unrealistic as the market goes quiet.

It also helps to remember that ATR on a one-minute chart includes wicks, so it overstates how far price travels in a clean direction. Net displacement over several candles is typically smaller.

When the target and the ATR disagree, the target is the thing that should move.

How does anyone here set targets on the one-minute chart, fixed pips or an ATR multiple, and which has held up better?