An ATR stop answers a different question from a structure stop. Practise them as rivals on the same entries, not as synonyms you mix when one feels tight.
Average True Range, as used in technical analysis since Welles Wilder's work, is a volatility measure. A multiple of ATR is a distance. A structure stop is a place — beyond a swing, a box, a session extreme. Distance and place will not agree. Your job in practice is to see what that disagreement does to your process, your hypothetical size, and your urge to move the stop.
This is not a recommendation to use ATR live. It is a how-to for a two-stop ledger on historical NSE charts.
Two questions, two ledgers
Structure ledger: "Where is the idea dead?" — beyond the swing that defines the trade.
ATR ledger: "How much room does recent volatility suggest?" — entry plus or minus k × ATR, with k and the ATR lookback frozen.
If you only ever place the wider of the two, you are not testing ATR. You are using ATR as permission to loosen structure. If you only ever place the tighter, you are using ATR as permission to sit inside noise.
The experiment: every practice entry gets both prices logged. You still take one live-in-replay stop — chosen by a rule you write before the week — but you record where the other would have been, and which one was hit first.
Freeze the arithmetic before the week
Write, then do not touch:
- ATR length (example: 14 on your practice timeframe — pick one)
- Multiple k for the ATR stop (example: 1.5 or 2 — pick one per week, not per trade)
- Which ATR value: at last completed bar, not a live updating excuse mid-bar
- Structure rule: swing of record, box opposite, or support/resistance beyond the trigger — one family
Intraday versus swing: do not compute a daily ATR and drop it onto a five-minute stop without writing that you are doing so. Mixing timeframes is a third method. Third methods wait their own week.
Position size in practice should follow position-sizing rules from the stop you actually honour, not from the stop you wish you had. If ATR is much wider than structure, the same rupee risk implies smaller size. Log that. It is often the whole lesson.
The Two-Stop Ledger (same trade)
For each of 20 entries in a frozen setup (pullback, breakout, fade — one setup):
| Field | What you write at entry |
|---|---|
entry |
Price you would have |
struct_stop |
Structure death price |
atr_stop |
Entry ± k × ATR |
honour |
struct or atr (pre-chosen for the week) |
R_struct |
Distance entry→struct (for risk-reward design) |
R_atr |
Distance entry→ATR |
first_hit |
After reveal: struct / atr / neither (target or time first) |
urge |
Did you want to widen after a wick? Y/N |
Week A: honour structure, ATR is shadow.
Week B: honour ATR, structure is shadow.
Same setup, same session windows, same liquidity tier. Change only which stop is real.
Decision criteria: when ATR is even in the conversation
Use ATR as a candidate only if you can say yes to at least one:
- Structure stop sits inside a noise zone you can define (opening spread, typical wick of this segment).
- Structure stop is so wide that your size becomes theatrical for the account you imagine.
- You have no clean swing because you are trading a break, not a pullback.
If none of those is true, skip ATR for that trade and log atr_not_needed. Forcing a volatility stop onto a clean swing is how traders complicate a good death point.
Never use ATR to justify staying in after structure is gone. That is ledger fraud: the place died, the distance is still open, and you are arguing with the original idea.
What the comparison is for
You are not hunting a winning k. You are answering process questions:
- In 09:15–10:00, did ATR stops get hit by ordinary open wicks while structure survived — or the reverse?
- In midday chop, did tight structure stops lottery you out of ideas that ATR would have kept — and was that "kept" actually a different, worse idea?
- Did honouring ATR make you skip trades because size became too small — a valid skip, not a missed fortune?
- Did
urgeto widen cluster on one ledger? That cluster is temperament, not proof.
Do not average these into a universal "ATR works on NSE." Session segment and setup will dominate. Keep the tags.
Indian-market mechanics (practice, then humility)
NSE cash trades 09:15–15:30. ATR computed on five-minute bars at 09:20 is a short, violent sample; ATR at 14:30 includes the whole day. If your lookback spans the open, the number is an open-volatility number. Tag atr_includes_open when that is true.
Gaps: a stop placed from yesterday's ATR does not know this morning's gap. Intraday practice should recalculate from bars that exist in the session you are in, or you should write a gap rule. Do not silently keep yesterday's distance.
Live trading adds brokerage, taxes, and slippage that practice will not fully simulate. Risk management without live capital still trains the habit of placing a stop at decision time. It does not train fill quality. Keep that humility in the ledger notes.
Failure modes
- k-shopping: 1.5× after a loss, 2.5× after a wick. Frozen k or the week is void.
- Hybrid mid-trade: start on structure, switch to ATR after pain. Tag
hybridand score zero. - Target copied from the other stop: honouring ATR but taking profit at structure because it is nearer — write targets separately.
- Indicator clutter: adding a second volatility band "for context" mid-block. Context is how k-shopping enters.
A 30-day replay block at ₹999 is enough to run Week A and Week B on one setup. Twenty plus twenty shadow comparisons will not certify a stop method. They will show which death rule you can obey when the date is hidden and the next bar can wick you.
Run the Two-Stop Ledger on historical NSE entries before you treat ATR as a personality. Join the waitlist.
Related reading
- Risk-Reward Practice: Designing Trades Before Entry
- Position Sizing Practice Without Live Capital
- Risk Management Practice Without Live Capital
- Support and Resistance: How to Practise Level Reading
Educational note
Educational practice content only. Not investment advice. No guaranteed returns. Practice ≠ live profitability.