Consider a hypothetical trader who enters the whole planned size before a pullback hits the initial stop. Entering half the size would halve the price loss on that initial tranche, but adding later changes both the average entry and later risk. Pyramiding defines those additions and their combined stop exposure in advance; it does not guarantee that a later trend can be captured.

What pyramiding is

A plan defines the number, size and trigger of additions. In the hypothetical 50–30–20 split below, the first tranche is half the target, followed by 30% and 20%. These fractions describe an exposure path, not a proven trend or a ranking of technical risk. Other plans may use equal units.

The principle is adding to winners. Each new tranche is allowed only when the existing position is already in profit and the trend structure remains intact. Adding to a losing position is averaging-down.

Classic scaling schemes

Illustrative splits and a documented historical convention
50–30–20A hypothetical target-size allocation: 50%, then 30%, then 20%. No popularity or performance ranking is asserted.
60–25–15Another hypothetical allocation with more exposure at the first entry. The result depends on actual entry and exit prices.
Historical Turtle unitsThe 2003 Original Turtle Trading Rules describe additions at half-N intervals from the previous actual fill, with at most four units per market. N is the system’s smoothed daily True Range measure. This is a historical futures-system convention, not a retail-CFD rule.
60–40A two-stage hypothetical allocation. Each resulting size still needs to meet its own minimum, step and loss-budget checks.

The 50-30-20 split on a concrete example

Use a hypothetical EUR 25,000 account and a 100,000-EUR EUR/USD contract per lot, with no costs. Three long entries are 0.50 lot at 1.0850, 0.30 at 1.0890 and 0.20 at 1.0930. The first stop is 1.0800, then proposed common stops move to 1.0840 and 1.0880 after confirmed changes. Entry triggers are assumptions, not proof of a future trend. The weighted entry is 0.50 × 1.0850 + 0.30 × 1.0890 + 0.20 × 1.0930 = 1.0878. The first tranche’s exact stop-price loss is USD 250, or EUR 231.48 if converted at the assumed stop rate 1.0800 USD per EUR. The conversion convention is part of this model; an actual provider may apply its own rates and costs.

At an assumed full exit of 1.1000, the pyramid gains 122 lot-weighted pips, or USD 1,220. Converting at that exit rate gives EUR 1,109.09. One full lot bought at 1.0850 would gain USD 1,500, or EUR 1,363.64 on the same exit. In this particular path, later entries reduce the gross profit by EUR 254.55. If the initial stop at 1.0800 executes before any addition, half a lot loses USD 250 (EUR 231.48 at that rate), versus USD 500 (EUR 462.96) for a full lot. These comparisons require the specified path, sizes, conversion and exact fills; they do not establish that scaling always improves a strategy.

Same final pyramid, different risk while adding

Our model has two10,000 EUR legs entered at1.1000,1.1020 with SL 1.0980. Add a third at1.1040 and aim to move all stops to1.1000. Adding before confirmed stop changes creates greater intermediate model loss than reversing the order, although the final state matches.

Scroll sideways to see the full diagram

Intermediate pyramid riskAdd first:model entry-based losses60,120,60 USD. Confirm stops first:60,20,60 USD. Maxima120 and60 USD.Entry-based model loss[USD]1206020StartOperation 1Operation 2Add→stop: 60→120→60Stop→add: 60→20→60
Steps show states after confirmed operations, not prices or exit probabilities.

Old stops give−20−40=−60 USD. Adding before modification adds−60, total−120. Moving stops first gives0−20=−20;then adding at the new stop adds−40, total−60. A final snapshot misses the largest intermediate burden. Do not assume multiple instructions execute atomically.

Assumptions: Hypothetical prices and ordering; each leg USD 1 per pip, ideal stop fill. No costs, slippage or price movement between operations. Loss is measured from entries, not current equity. Stop changes for all existing legs and the new leg are confirmed. Legs analytically divide shared exposure. This model does not guarantee separate-position handling across account modes; check accounting and stop application for the actual account. Formula or data reference: MetaTrader5 — order execution and stop loss.

Risk cap and a combined stop loss

Distinguish the combined result measured from entries from the decline measured from current equity. A common stop above the weighted entry can show a gross profit in an exact-fill model while still giving back substantial floating profit. In a separate USD model, two 10,000-EUR legs at 1.1000 and 1.1020 have USD 60 floating profit at 1.1040. A common stop at 1.1000 gives USD −20: the decline from now is USD 80. If the chosen additional equity-decline budget is USD 100, only USD 20 remains. A third leg at 1.1040 loses USD 0.0040 per EUR to that stop, so the model permits 20/0.0040 = 5,000 EUR, or 0.05 lot. It gives total stop P/L USD −40 and an equity decline of USD 100. Adding 0.10 lot instead gives P/L USD −60 and decline USD 120. No costs, spread, price movement between operations or worse fills are included; the budget is not guaranteed.

  • Verify each modification before treating it as the account state. A proposed stop change and an executed addition need not be atomic. Calculate exposure and model loss in the intermediate states as well as the final one.
  • Recalculate the combined position after every addition. Sum each tranche’s profit or loss at the proposed stop, including costs and account-currency conversion. A stop above the weighted average entry can indicate a planned gross profit, but gaps and slippage prevent calling earlier tranches risk-free.
  • State the stop rule rather than a universal ATR buffer. An ATR multiple is a chosen price-distance input, not a guaranteed pullback range. In the historical Turtle document, ordinary unit stops and the alternative Whipsaw rule differ; equal units do not inherently require an unusually fast universal trailing rule.

Add-on triggers in practice

Define the addition trigger so the rule can be tested: for example, a pullback to a 20-period moving average, a specified ATR-distance move, a consolidation breakout or a close beyond a stated higher-timeframe level. These are illustrative conditions, not verified ways to reduce loss or certify trend continuation. Record their units, timing and permitted size in the plan.

The most common mistakes

  • Confusing a favourable-move addition with averaging down. Both increase exposure, but they use a different price-path condition. Adding to a loser is not a mathematical guarantee of bankruptcy; its result depends on the added size, exit and later path. It still needs an explicit aggregate-loss calculation.
  • Failing to recalculate after an add. If the example’s stop remained 1.0800, its first two tranches would lose 25 + 27 = 52 lot-weighted pips, or USD 520. Converted at that assumed stop rate, this is EUR 481.48, 1.9259% of initial EUR 25,000. All three give 25 + 27 + 26 = 78, or USD 780 = EUR 722.22, 2.8889% of initial capital. The reference base here is initial capital, not current equity. Fees, conversion spreads and worse fills are excluded.
  • Adding more exposure than the plan permits. Some pyramiding systems use equal units, including the Turtle example above. The key control is the combined exposure and stop risk, not a claim that equal additions cease to be pyramiding.
  • Ignoring shared currency exposure. Long EUR/USD, GBP/USD and AUD/USD all include short USD exposure, but their other currencies and position sizes differ. Assess their combined USD amounts and changing correlations instead of assuming the risk is exactly three times one position.
  • Calling a late signal proof of continuation or reversal. A later tranche enters at a different price and may lose even while an earlier tranche remains profitable. No highest correction probability follows simply from being the last addition. Test the stated trigger and track the combined outcome.
  • Discretionary overrides of the scheme. “It is going so well, I’ll add more than planned” is the moment when a mechanical technique mutates into gambling. The split is binding even when euphoria says otherwise.

When pyramiding does not make sense

Pyramiding changes the entry and exposure path, so test it against a defined single-entry baseline. Ranges and news can challenge particular addition rules, but there is no universal timeframe at which scaling fails or fixed two-to-threefold news-volatility multiplier. Positive baseline expectancy does not prove the scaled variant is profitable.

What to do tomorrow — your first steps

  1. Specify the addition trigger, timing, size and exit conditions before using them. Record changes to those rules so a later comparison describes the actual procedure.
  2. Set the number of additions and size allocation as explicit plan inputs. Calculate each leg’s volume, minimum, step, nominal exposure and margin; no universal limit of three or four follows from the term pyramiding.
  3. In a hypothetical or demo comparison, use the same declared capital, price path and cost assumptions for the scaled plan and its single-entry baseline. A month of simulation does not validate future live profitability.
  4. Record every entry and proposed exit, combined P/L from entries, floating P/L now, and the resulting decline from current equity. Include intermediate confirmed states, costs and conversion rather than relying only on the weighted average entry.
  5. Document any review or stopping condition as a chosen rule, with a reason. Three losses in a week or a four-week pause is not a universal evidence-based threshold; a circuit breaker does not guarantee the final loss amount.

Related reading: the anti-martingale system as the foundation of sizing; trend-following systems as the context for pyramiding; ATR-based trailing stop for calibrating stops in changing volatility. For the wider context, see the risk management section on forex-basics.com.