A fixed ATR threshold is not portable
ATR measures the average size of recent bars, and it reports that size in the instrument's own price units. That one property decides how you should build with it.
An ATR of 1.5 means:
| On | 1.5 ATR is |
|---|---|
| EURUSD quoted to 5 decimals | an enormous move |
| a stock priced at 40 | a normal daily range |
| an index at 18,000 | almost nothing |
So a filter written as "only trade when ATR is above 1.5" is not a volatility filter. It is a filter that happens to select almost every bar on one instrument and almost none on another, and it cannot be moved between symbols at all.
The fix: divide ATR by something with the same units
A ratio of two quantities in the same units is dimensionless, which makes it mean the same thing everywhere. Two constructions, both available in a form:
ATR against its own average — "ATR above its 50-period average" says volatility is expanding relative to its own recent normal. That is a genuine regime statement, and it is identical in meaning on EURUSD and on an index.
ATR as a fraction of price — ATR divided by close gives you a percentage range. Useful when you want an absolute notion of "quiet" rather than a relative one.
Both are portable. A fixed ATR level is not. This is the same reasoning that makes a CCI threshold unportable while an RSI threshold largely is — why CCI has no ceiling.
Why this matters beyond tidiness
Testing on a second instrument is the single best check that you built a strategy rather than a description of one market — it is on the evidence list in what to hold before risking money.
You cannot run that test at all if your filter is calibrated in the price units of the first instrument. So making the filter dimensionless is not a style preference; it is what makes the most important test possible.
Building the regime filter with a trend entry
| Field | Value | Role |
|---|---|---|
| MA fast / slow | 20 / 50 | the entry signal |
| ATR length | 14 | |
| Regime condition | ATR above its own 50-period average | only trade when volatility is expanding |
| Stop | 2 × ATR | scales with the instrument automatically |
| Target | 3 × ATR |
The ATR-based stop is the other reason ATR belongs in almost every strategy: a stop in ticks has to be re-tuned per symbol, and a stop in ATR does not. The trade-off is real though — an ATR stop widens exactly when the market turns violent, which is when you may most want to be out. It is a choice, not a free improvement.
And the honest caveat
A volatility filter tells you the size of recent moves. It says nothing about direction, and nothing about whether the expansion will continue. Combining it with a trend entry is sensible; expecting it to predict anything on its own is not. Confirm it earns its place by counting how many trades it removes — the count test.