Stop-loss sizing on Deriv synthetic indices is best treated as a volatility measurement problem, not a guess. Because synthetic indices are generated by a defined stochastic process rather than by underlying cash markets, their short-horizon volatility is comparatively stable within a session but still varies across instruments and volatility tiers. This article shows how to compute an adaptive stop distance from tick data using Average True Range (ATR) and a volatility-scaled rule, then how to translate that distance into parameters your Deriv Bot strategy can act on. The method is mechanical, auditable, and reproducible from the data feed alone.
Trading involves significant risk and you can lose your entire stake. Test every strategy on a demo account first.
Why fixed stop distances fail on synthetic indices
A fixed stop distance in points is the most common default in community bots, and it is the most common source of silent strategy drift. Synthetic indices are offered across volatility tiers (for example, the Volatility 10, 25, 50, 75 and 100 indices on the synthetic indices page), and a 20-point stop that is generous on a low-volatility index is effectively a coin flip on a high-volatility one. The same fixed distance therefore produces wildly different hit rates depending on the instrument, and the strategy’s expectancy changes without any change to the code.
A second problem is that fixed distances interact badly with Deriv’s contract mechanics. On Rise/Fall and Digits contracts, the stake is the maximum loss; there is no separate stop-loss order. On Multipliers, a stop-loss by price exists, but the distance you choose still determines how much of your stake is exposed to adverse movement. In both cases the correct quantity to control is the distance relative to current volatility, not a fixed number of points. ATR gives you that distance in a way that adapts to the instrument and the session.
A third problem is regime change. Synthetic indices can shift between quiet and active phases within a session. A stop sized on last week’s ATR may be too tight today, or too loose. Recomputing ATR from recent ticks on every evaluation keeps the stop aligned with the current regime, at the cost of some noise in the estimate. That trade-off is manageable if you smooth the estimate and cap it.
Computing an adaptive stop distance from tick data
The ATR is the average of the True Range over a lookback window. For tick data, define the True Range of tick i as the maximum of three quantities: the absolute difference between the current tick price and the previous tick price, the absolute difference between the current price and the previous window’s high, and the absolute difference between the current price and the previous window’s low. In practice, with a single price series, the first term dominates and the other two reduce to the same value, so a simplified tick-ATR is the average absolute tick-to-tick move over N ticks. This is a defensible simplification for synthetic indices, which have no gaps in the tick feed.
Two parameters matter. The lookback N controls responsiveness: small N reacts quickly but is noisy; large N is stable but slow. The multiplier k converts the ATR into a stop distance. A common starting point is k = 2 to 3, meaning the stop sits two to three average tick ranges away from entry. Both parameters should be chosen before trading and recorded in your strategy notes, per the ORSTAC rule that rules are written before trading and changes follow a process (hypothesis, test, version).
The stop distance in price units is then:
- `stopdistance = k × ATRN`
Convert that to the units your contract uses. For Multipliers, this is the price distance for the stop-loss level. For fixed-stake contracts, the stop distance informs your duration and stake choices rather than a price order: a wider stop distance implies a longer duration or a smaller stake to keep the maximum loss within your per-contract limit. This distinction is important and is where many bots go wrong — they apply a price-based stop rule to a contract that has no price stop.
A minimal pseudo-code loop for a DBot-style strategy is:
atr = mean(abs(price[i] - price[i-1]) for i in last N ticks)
stop_distance = k * atr
if contract_type == "MULTIPLIER":
stop_loss_level = entry_price - stop_distance # for long
assert stake * (stop_distance / entry_price) <= 0.01 * balance
else:
# fixed-stake: size stake so that stake <= 0.01 * balance
stake = min(stake, 0.01 * balance)
duration = choose_duration(stop_distance)
The assertion in the Multiplier branch is the important one: it enforces the ORSTAC rule that risk per contract is at most 1% of balance (ceiling 2%). On fixed-stake contracts the stake is the maximum loss, so the same 1% ceiling applies directly to the stake.
Worked example
Suppose a synthetic index has a recent average absolute tick-to-tick move of 0.50 price units over the last 200 ticks, and you choose k = 2.4. The stop distance is 2.4 × 0.50 = 1.20 price units. On a long Multiplier position entered at 100.00, the stop-loss level is 100.00 − 1.20 = 98.80.
Now consider a balance of 1,000 and a 1% per-contract risk ceiling, which is 10.00. The maximum stake that keeps the stop-loss exposure within that ceiling is stake × (stopdistance / entryprice) ≤ 10.00, so stake ≤ 10.00 / (1.20 / 100.00) = 10.00 / 0.012 = 833.33. Round down to 833. If you instead set the stake at 1,000, the exposure at the stop would be 1,000 × 0.012 = 12.00, which exceeds the 1% ceiling and violates the rule.
For a fixed-stake Rise/Fall contract, the same 1% ceiling means the stake is capped at 10.00 regardless of ATR. The ATR instead informs duration: if the stop distance is 1.20 and the contract pays a fixed return, you should check whether the duration is long enough for the price to plausibly move that far. That check is a filter, not a guarantee — it does not predict the outcome, it only avoids durations that are structurally too short for the volatility you measured.
Encoding the rules in the bot
The stop distance is only useful if the bot enforces it without human intervention. Per ORSTAC premise 5, the bot must carry hard automatic limits: a daily loss limit, a maximum drawdown limit, and a consecutive-loss limit, each of which stops the bot by itself. These limits are set in currency terms, not in ATR terms, and they are checked before every new contract is placed.
A minimal rule set for a DBot-style strategy:
- Compute ATR over the last N ticks; if fewer than N ticks are available, do not trade.
- Compute `stopdistance = k × ATR`; if `stopdistance` exceeds a pre-set cap (for example, twice the median ATR over the last 24 hours), do not trade — the regime is outside the tested range.
- If the contract is a Multiplier, set the stop-loss level at `entry ± stop_distance` and assert that the resulting exposure is within 1% of balance.
- If the contract is fixed-stake, cap the stake at 1% of balance and choose duration from the stop distance using a pre-tested mapping.
- Before placing, check daily loss, drawdown, and consecutive-loss counters; if any limit is hit, stop.
- Log the ATR, stop distance, stake, and outcome for every contract, so the demo sample can be reviewed later.
Step 6 is not optional. The ORSTAC premises require a recorded demo sample before real money, and a sample is only reviewable if it is logged. A backtest is a filter, not proof: report the number of configurations tried, use out-of-sample data, and treat the demo sample as a separate, forward-looking check. No data, no claim.
Limitations
This approach does not solve the fundamental problem that ATR measures recent volatility, not future volatility. A stop sized on the last 200 ticks can still be hit by a regime shift that the lookback has not yet absorbed. The method also does not address payout structure: on fixed-stake contracts, a wider stop distance may require a longer duration, and longer durations change the payout and the break-even rate. Win rate without payout is meaningless, so the stop distance must be evaluated alongside expectancy and drawdown, not in isolation. Finally, the method assumes a clean, gap-free tick feed; if the feed is interrupted, the ATR estimate is stale and the bot should refuse to trade until it refreshes. Martingale-style progressions are not made safe by ATR sizing — ORSTAC recommends against martingale, and any progression needs a hard cap on steps and stake plus a ruin test before going live.
Frequently Asked Questions
Does an ATR-based stop improve win rate? Not necessarily, and win rate is the wrong metric. ATR sizing aims to keep the stop distance proportional to current volatility so that the strategy’s expectancy is not silently changed by the instrument’s volatility tier; judge the result by expectancy, break-even rate and drawdown, not by win rate alone.
Can I use the same k and N on every synthetic index? No. The appropriate k and N depend on the instrument’s tick frequency and volatility tier, and should be chosen per instrument and recorded before trading. Applying one setting across all indices is a configuration error, not a simplification.
Does this replace the bot’s daily loss and drawdown limits? No. The stop distance controls per-contract exposure; the daily loss, drawdown and consecutive-loss limits control session-level exposure and must stop the bot automatically. Both are required, and neither substitutes for the other.
Key Takeaways
- Size stops from measured tick volatility (ATR) rather than a fixed point distance, and record k and N before trading.
- On Multipliers, convert the stop distance into a price level and assert that exposure stays within 1% of balance; on fixed-stake contracts, the stake is the maximum loss and is capped at 1% directly.
- Enforce daily loss, drawdown and consecutive-loss limits in the bot so it stops by itself; log every contract for the demo sample.
- Treat ATR as a filter, not a prediction: report configurations tried, test out-of-sample, and judge by expectancy and drawdown.
- ORSTAC recommends against martingale; any progression needs a hard cap and a ruin test before going live.
The ORSTAC repository hosts open-source bots for Deriv that implement these patterns, and the Deriv platform offers a free demo account to test them before any real-money use. The Deriv API documentation and the official Deriv API libraries cover the tick feed and contract calls referenced above, and the Deriv Bot page describes the block-based environment. Discussion of the sizing rules and their edge cases is welcome in the ORSTAC community channels linked from the repository.
Trading involves significant risk and you can lose your entire stake. Test every strategy on a demo account first.
