A Fibonacci extension projects multiples of a chosen price move from a stated reference point. It can help compare exit scenarios, but the formula supplies neither a probability nor a time to reach them. Two lines labeled 161.8% can have different prices because their zero points differ.

How an extension differs from a retracement

A Fibonacci retracement measures a correction relative to an earlier move; a projection lays a multiple of that move off a chosen origin. Neither drawing by itself prescribes an entry or exit. With three anchors, even a level below 100% can lie beyond the earlier extreme, so compare actual prices and reference points rather than tool names alone.

MetaTrader 5’s Fibonacci Expansion uses two successive waves, with the first supplying length and the second endpoint supplying the projection origin. TradingView’s Trend-Based Fib Extension likewise selects the move’s start, end and correction endpoint. MT5’s two-point Retracement tool also lists levels beyond 100%. Record the number of anchors, zero and 100% prices, and direction; Reverse can change the labeling orientation.

How to measure the A-B-C wave

For our linear, non-reversed model, let A start the first move, B end it and C end a later correction. Define signed Δ = B − A and T(k) = C + k(B − A). Upward Δ is positive; downward Δ is negative. Selected A, B and C suffice to calculate the lines before a breakout of B; a breakout can be a separate strategy condition.

k = 1.618 means 161.8% of the AB length projected from C, not a 161.8% rise in the whole price or account return. The multiplier is a rounded approximation to the golden ratio; 1.272 approximates its square root. Grids can also contain 1, 2 or custom multiples. Their labels do not establish a statistical edge or a measured hit frequency.

Record settings with the anchors
Linear modelT(k) = C + k(B − A), with signed length.
Two-point comparisonOur explicit E(k) = A + k(B − A) puts zero at A and 100% at B.
Reversed labelingA line beyond B can have a negative label under another orientation; compare prices.
Logarithmic optionTradingView offers a log-based calculation when enabled. It differs from the linear examples below.

Reaching 161.8% does not mean a 161.8% result on the whole position

Project AB=10 from C=106. A separate rule determines the portion of the hypothetical position closed at each level.

Scroll sideways to see the full diagram

Reaching 161.8% does not mean a 161.8% result on the whole positionHypothetical linear anchors A100, B110 and C106 give T(1)=116 and T(1.618)=122.18. Assume one arbitrary position unit opened at 106 and ideal fills at both specified prices. Closing all at 122.18 gives 16.18 result units. Closing half at 116 and half at 122.18 gives 5+8.09=13.09. No future price path or effectiveness of the levels is assumed.Same grid, two ways to split the position100110A:100B:110C:106100%:116161.8%:122.18All at 122.18:1 ×16.18=16.18Halves:0.5 ×10 +0.5 ×16.18=13.09Amounts in arbitrary result units.
Assumptions: fictional linear instrument; one position unit gains one arbitrary result unit per price point. Both targets are assumed filled; costs, slippage, financing and re-entry are excluded. No comparison of risk or real effectiveness. Projection: T(k)=106+k×10. The 50/50 split is a separate assumed exit rule. TradingView: three projection anchors.
Fill prices and position shares determine the result, not the label on the last line.

A step-by-step worked example

Two independent hypothetical EUR/USD scenarios use one pip = 0.0001 USD per EUR. Upward anchors are A = 1.08000, B = 1.09000, C = 1.08500; downward anchors A = 1.10000, B = 1.09000, C = 1.09500. Each has |AB| = 0.01000, or 100 pips. The following prices assume linear calculation without Reverse, not historical or current quotes.

Three-anchor projections with equal length and opposite directions
kUp: 1.08500 + k × 0.01000Down: 1.09500 − k × 0.01000
01.085001.09500
0.6181.091181.08882
11.095001.08500
1.2721.097721.08228
1.6181.101181.07882
21.105001.07500
2.6181.111181.06882

For the up scenario, T(1.272) = 1.09772: exactly 127.2 pips are added to C, not 127 before rounding. For down, T(1.618) = 1.07882. Retain calculation precision until adapting an order to the instrument’s price step. Upward T(0.618) lies 11.8 pips above B; a selected T(0.382) = 1.08882 lies 11.8 pips below it. Thus not every displayed “extension” line lies beyond the entire earlier wave.

For the same upward A/B, our two-point E(1.618) = 1.09618, while three-point T(1.618) = 1.10118. They differ by C − A = 50 pips for any common k. Moving C to 1.08600 lifts all three-point levels 10 pips but leaves two-point E unchanged. This is geometry rather than evidence of a stronger trend.

Setting the take-profit and trailing the stop

In a hypothetical partial-exit plan, one portion closes at a nearer projection and the remainder at a farther one. Compute the total from each portion’s actual fill price and size, as the diagram shows. Moving the remaining stop to entry does not eliminate loss from costs or slippage, and a projected target does not guarantee a fill.

Execution is a separate specification. In typical MT5 Forex, a long position’s Take Profit condition uses Bid and a short’s uses Ask; exchange instruments can have different triggering rules. A line touched by another price series does not establish a fill. The trade exit strategies lesson discusses separate size, partial-exit and cost rules. In the diagram, both targets fill only because the arithmetic model explicitly assumes that.

Anchor uncertainty and confluence are separate questions

A level near an earlier resistance level or a moving average may define an additional test condition, but those observations often derive from the same prices. Coincidence alone does not prove independent evidence, a high probability or a realizable target. Specify the tolerance and anchors before observing outcomes.

In our own sensitivity model T = C + k(B − A), allow each anchor to vary by at most ε. For positive k, the greatest possible target difference is ε(1 + 2k): C has weight 1, B weight k and A weight −k. The upward extreme raises C/B and lowers A. With each anchor within ±2 pips, k = 1 gives ±6 pips, or 1.09440–1.09560 around the upward baseline 1.09500. k = 1.618 gives ±8.472 pips, or 1.1003328–1.1020272 around 1.10118. These are deterministic bounds on stated anchor shifts, not a price forecast, random-error assumption or confidence interval.

A thin drawn line can therefore hide a wider anchor-selection uncertainty. A touch rule narrower than that uncertainty needs especially precise anchor definitions. Trading harmonic patterns discusses related ratio constructions, and the ForexMechanics technical analysis section covers other tools. Their presence on one chart does not establish an independently confirmed profitable exit.

Practical takeaway: An extension is a projected price under chosen anchors. Reaching the level and executing an exit are separate events.

Preserve when the anchors became available

Record feed, interval, wick or close anchors and the rule for confirming C. If a chosen trough needs two later candles for recognition, its projections become available only after those candles. Plotting that final trough earlier in a backtest uses future data. A nearer or farther line supplies no measured probability by itself.

Moving C after a new low creates a revised scenario. Preserve the prior version and observation time rather than judge an earlier decision with later anchors. Test the defined plan on unused data, retaining unfilled targets and costs. A stop at entry still leaves execution and cost risk, rather than eliminate capital at risk.

What to do tomorrow

  1. Change only C. Keep A = 1.08000 and B = 1.09000 but set C = 1.08300. Three-point 100% becomes 1.09300 and 161.8% becomes 1.09918: both 20 pips lower. Two-point E(1.618) stays 1.09618.
  2. Check the convention. Record zero/100% prices, Reverse, linear or logarithmic calculation and the precise anchors rather than compare percent labels.
  3. Check recognition time. Write when C became available under the selected confirmation rule. Keep every later revision.
  4. Separate execution. For partial exits, specify allocation, fill assumptions, price side and costs. A 50/50 split is optional and does not follow from Fibonacci.
  5. Retain the full sample. Include non-reached targets and canceled or unfilled orders. A few dozen selected examples alone do not establish which level universally “works.”