Petko AleksandrovRisk warning. Leveraged currency products carry substantial risk to your capital. Nothing below predicts price movement, and a level reached once tells you nothing about the next occasion. Practise on a virtual account before committing money.
The Short Answer
Draw the Fibonacci retracement from the first confirmed low of an uptrend to the first qualifying swing high. Price then usually pulls back. When it retraces into the 38.2 to 61.8 band, that band becomes your zone of interest, and a break of the counter-trend line drawn across the pullback acts as confirmation. Targets sit at the 1.272 and 1.618 extensions.
Then comes the part that makes this method distinctive. Once price touches the second target, you redraw: anchor a fresh Fibonacci from the next high back down to the lowest point of the retracement, and repeat. The drawing rolls forward with the move rather than sitting static.
If price falls below the zero anchor of your current drawing, everything resets, and you start again from the new low.
That’s the whole method in four paragraphs. Everything below turns it into rules precise enough that two people reading the same chart would place the same lines, which the original version of this article honestly did not achieve.
Where This Came From
Years ago in London, I spent time with an older trader who drew Fibonacci differently from anyone I’d seen. Not the textbook single measurement, but a rolling sequence that advanced with the market.
That conversation shaped how I’ve marked charts ever since, and I’ve used the approach for a long stretch now. Worth saying plainly, though: I’ve never put it through a formal test with recorded statistics. Personal experience is not evidence, and I’ll return to that gap near the end rather than pretending otherwise.
Retracement Against Extension
Within forex technical analysis, these are two different measurements taken from one drawing, and conflating them causes real confusion.
Fibonacci retracement levels sit inside the move you measured. They mark how far price has pulled back against the direction of the impulse. The 38.2, 50, and 61.8 readings all fall here.
Extension levels sit beyond the end of the move. They project forward, estimating where continuation might reach. The 1.272 and 1.618 readings live out there.
One measures a correction. The other guesses at a destination. Same tool, opposite purposes.
Which Levels to Keep
Most platforms load the Fibonacci tool with more lines than anyone needs, and a cluttered chart hides more than it shows.
| Level | Keep or remove | Purpose |
| 38.2 | Keep | Upper edge of the retracement band |
| 50 | Keep | Midpoint reference, not a true Fibonacci ratio |
| 61.8 | Keep | Lower edge of the retracement band |
| 78.6 | Remove | Too deep; setups reaching here usually fail anyway |
| 1.272 | Keep, colour it | First target |
| 1.618 | Keep, colour it | Second target, and the trigger for redrawing |
| 2.618 | Remove | Rarely reached before conditions change |
| 3.618 | Remove | Same |
I colour both extensions yellow or green so they read as targets at a glance rather than blending into the retracement lines. Small thing. Saves squinting.
Worth noting that 50 isn’t derived from the Fibonacci sequence at all. It survives on convention and on the observation that markets frequently retrace about half a move, which is not the same as mathematical justification.
Defining the Swing Points Objectively
Here is where the original article fell short, and where most explanations of this topic fall short too.
Saying “use the highs that are really visual and ignore the small ones” describes what an experienced eye does. It does not tell a newcomer anything reproducible, and two traders following that instruction will anchor differently and reach different targets.
Pick one objective definition and apply it consistently:
| Method | Rule | Suits |
| Fractal | High with two lower highs either side | Anyone; built into MetaTrader |
| Swing count | High with at least three lower highs either side | Filtering more noise |
| ZigZag threshold | Only reversals exceeding a set percentage | Cleaner charts, fewer signals |
| ATR distance | Swing must span at least 1.5 times average range | Adapts to changing volatility |
| Percentage move | Minimum move of, say, 0.5 percent on the pair | Simple, though fixed |
My preference is the fractal rule with a volatility filter layered on top, since fractals alone produce plenty of insignificant points during quiet sessions. Whichever you choose, write it down and stop overriding it when a chart looks tempting. Discretion applied inconsistently is how a method stops being a method.
The Long Setup, as Rules
| Component | Rule |
| Direction filter | Higher highs and higher lows on the working timeframe |
| Anchor point one | First confirmed swing low ending the previous downward move |
| Anchor point two | First qualifying swing high after that low |
| Zone of interest | 38.2 to 61.8 retracement band |
| Confirmation | Close beyond the counter-trend line drawn across the retracement highs |
| Timing | Enter at the open of the candle following confirmation |
| Protective exit | Below the retracement low, or below the zero anchor for a wider version |
| First target | 1.272 extension |
| Second target | 1.618 extension |
| Redraw signal | Price touches the 1.618 level |
| Invalidation | Price closes below the zero anchor |
| Sizing | Fixed percentage of account risked per position |
Notice what changed from the original description. Entry timing, the closing-basis requirement on the line break, protective placement, and sizing were all absent before, and without them the method is a drawing exercise rather than something tradeable.
Working Through One Long Example
Say the market has been falling, then puts in a low and turns.
- Identify that low. It marks the end of the downward phase and the start of the upward one, the same candle serving both roles.
- Find the first qualifying high after it, using whichever swing definition you settled on. Anchor the Fibonacci from low to high.
- Read the levels. The band between 38.2 and 61.8 is where you want price to return. Above it, the 1.272 and 1.618 extensions are marked in your target colour.
- Wait. Price returns into the band, or it doesn’t. Nothing obliges you to act.
- As the correction develops, draw a line connecting its descending highs. That’s the counter-trend line, and a close above it while price sits inside the band is your confirmation.
- Open the position at the next candle’s open. Protective exit below the correction low. Size the position so the distance to that exit costs you a fixed, predetermined fraction of the account.
- Manage toward 1.272, then 1.618.
That’s one complete cycle. What follows is where this method separates from ordinary Fibonacci usage.
The Rolling Continuation
When price reaches the 1.618 extension, the current drawing has done its work. Rather than abandoning the chart, you advance the tool.
- Wait for price to touch the second target.
- Find the first qualifying high after that touch.
- Find the lowest point of the retracement that followed.
- Draw a fresh Fibonacci between those two points.
- Read the new zone of interest and the new targets.
- Repeat from step one.
Direction of drawing doesn’t matter mathematically. Anchoring high-to-low or low-to-high produces identical levels, so I draw whichever way keeps the lines compact and stops successive drawings from covering each other. Purely cosmetic, though on a chart with five or six overlapping measurements it stops being cosmetic quickly.
Remove old drawings once they’re superseded. Keep the current one and perhaps the previous. Beyond that you’re looking at clutter, not analysis.
One structural limitation, and it matters: you cannot start this sequence from the middle of an established move. The measurement runs forward with the market from a genuine turning point. Trying to reverse-engineer it backward from where price sits today produces levels that look meaningful and aren’t.
When to Reset, and What Reset Means
Price closing below the zero anchor of the current drawing means that drawing is finished.
The original article said to restart without explaining what restarting affects. Three separate things need distinguishing:
| Situation | What happens |
| No position open | Delete the drawing, wait for a new confirmed low, begin again |
| Position open, protective exit already hit | Nothing to manage; redraw from the new low |
| Position open, exit not yet hit | The setup’s premise has failed; close manually rather than hoping |
That third row is the one costing people money. A break below the anchor invalidates the reasoning behind the position, and holding past invalidation because the protective level hasn’t technically triggered is how a small loss becomes a large one.
Expect several resets in a row sometimes. During choppy conditions the sequence starts, fails, starts again, fails again. Frustrating, and normal.
The Short Setup
Everything mirrors. Nothing about the logic changes.
| Element | Long | Short |
| Market condition | Higher highs, higher lows | Lower lows, lower highs |
| Drawing direction | Swing low to swing high | Swing high to swing low |
| Retracement direction | Price moves down | Price moves up |
| Zone of interest | 38.2 to 61.8 | 38.2 to 61.8 |
| Counter-trend line | Across descending correction highs | Across ascending correction lows |
| Confirmation | Close above the line | Close below the line |
| Protective exit | Below the correction low | Above the correction high |
| Targets | 1.272 and 1.618 above | 1.272 and 1.618 below |
| Reset condition | Close above the zero anchor | Close below the zero anchor |
The original article showed only the upward version, which left half the method undocumented. Both sides trade identically once you accept that “up” and “down” simply swap places.
Protective Exit Placement
Four workable options, and choosing one matters more than which one you choose:
- Below the retracement low: Tightest, best reward ratio, most likely to be triggered by ordinary noise.
- Below the zero anchor: Widest, survives deeper corrections, costs considerably more when wrong.
- A fixed distance beyond 61.8: Predictable, ignores structure.
- An ATR multiple below entry: Adapts to conditions, needs recalculating each time.
I lean toward the retracement low with a small buffer, though I’d accept that the anchor-based version is more logically consistent, since that’s the level whose breach invalidates the whole premise. Reasonable people differ here, and I’ve changed my own mind about it more than once.
Sizing the Position
This part receives almost no attention in most Fibonacci explanations, which is odd given that it determines survival more than entry precision does.
- Decide the maximum fraction of your account you’ll risk on any single position. Something modest.
- Measure the distance in pips from your entry to your protective exit.
- Divide your risk amount by that distance to get position volume.
- Adjust for the pip value of the pair you’re trading.
Consequence worth understanding: wider protective distances mean smaller positions, not bigger risk. Many beginners do the opposite, keeping volume constant and letting risk float with whatever the chart happens to offer.
Targets and Partial Exits
Two levels, two decisions.
Taking everything at 1.272 produces a higher hit rate and a smaller average result. Holding for 1.618 does the reverse. Closing part of the position at the first level and moving the protective exit to breakeven for the remainder sits between the two, which is what I do, though I won’t claim it’s mathematically superior.
What I would say: pick one approach and apply it consistently. Switching between them based on how confident a particular chart feels reintroduces exactly the discretion this framework exists to remove.
What This Method Has Not Demonstrated
Honest section, and the one the earlier version needed most.
I’ve used this approach for years and find it useful. That statement is worth very little as evidence, because personal impressions are shaped by memorable wins and forgotten losses, and because no statistics were ever recorded.
Nothing here has been tested with:
| Missing evidence | Why it matters |
| Defined test period | Conditions differ enormously between years |
| Pairs tested | Behaviour varies across instruments |
| Number of positions | Fewer than several hundred proves nothing |
| Win rate | Absent |
| Average reward against risk | Determines whether a low hit rate still works |
| Profit factor | Absent |
| Maximum drawdown | The figure deciding whether you could hold on |
| Spread and commission assumptions | Costs matter, particularly on shorter timeframes |
| Out-of-sample segment | Separates a real edge from a fitted one |
| Forward-test record | Nothing published |
So treat this page as a documented method rather than a validated one. If you test it properly, the table above lists exactly what to record, and your results will be worth more than my recollection.
The subjectivity problem compounds this. Until swing points are defined by a fixed rule, no two backtests of “this method” test quite the same thing, which is another reason the definitions section matters.
Common Mistakes
- Anchoring to the wrong low: Deeper points hiding inside a spike get missed constantly. Zoom in before committing.
- Ignoring a reset: The sequence broke; you kept the drawing anyway because you liked the target.
- Drawing backwards from current price: Produces levels with no predictive basis whatsoever.
- Treating a level as a guarantee: Price reaching 1.618 exactly on one occasion is an anecdote, not a property of markets.
- Keeping every drawing on screen: Six overlapping measurements produce visual noise that hides the current one.
- Using this in isolation: Levels work better alongside structure, momentum context, and awareness of scheduled news.
Does This Belong Alongside Other Tools?
Short answer: yes, and treating these levels as a standalone system is probably the most common error I see.
Fibonacci levels mark areas where orders tend to cluster. They say nothing about whether the broader market wants to continue in that direction, whether a scheduled release is about to arrive, or whether volatility has collapsed to the point where targets sit unreachably far away. Traders combining these zones with market structure, session timing, and basic awareness of the economic calendar generally report steadier results than those reading levels in isolation, though I’d stress that’s a common impression rather than a measured finding.
What I’d avoid is stacking confirmations until a setup requires five conditions aligning. Each addition reduces your sample size, and a method producing three signals a year cannot be assessed meaningfully however good those three look.
Timeframes and Pairs
Higher timeframes produce fewer forex setups with cleaner swing points. H4 and daily charts suit this approach best in my experience, since the qualifying highs and lows are unambiguous enough that the subjectivity problem shrinks.
Below H1 the swing definitions start firing on noise, and spread costs eat a larger share of each move. Possible, just harder.
Liquid majors behave most predictably here. Exotic crosses with wide spreads and erratic gaps make both the levels and the costs less reliable.
Frequently Asked Questions
Why do these particular ratios get watched?
The numbers derive from the Fibonacci sequence, where each figure is the sum of the two preceding it, producing ratios near 0.618 and 1.618 as the series extends. Whether markets respect them for mathematical reasons or simply because enough participants watch the same lines remains debated, and the self-fulfilling explanation is at least as plausible as any natural-order argument. Either way, the practical effect is that orders cluster around these areas, which is what makes them worth marking.
Does this work on stocks and indices too?
The drawing method transfers to any instrument with visible swing structure, including indices, metals, and shares. Adjustments are usually needed, since equities gap overnight in ways forex pairs generally don’t, which can jump straight past a protective level. Session hours differ as well, affecting where swings form. Test each market separately rather than assuming settings carry across, because volatility characteristics vary considerably between asset classes.
Can I automate this method?
Partly. Level calculation and target projection code easily, and swing detection works once you’ve committed to a fixed definition such as fractals or a ZigZag threshold. Harder to automate is the counter-trend line, since drawing one requires choosing which highs to connect, and different reasonable choices produce different break points. Most people who automate this end up simplifying confirmation to something like a candle close above a short moving average.
How long should I wait for price to reach the zone?
No fixed limit exists in the rules, though a retracement that stalls without reaching the band and then resumes the original direction simply means no setup occurred. Some traders add a time filter, cancelling any setup where price hasn’t reached the zone within a set number of candles. Reasonable idea, and one worth testing rather than adopting on instinct. Waiting indefinitely ties up attention on a chart that has moved on without you.
What if price reaches the first target then reverses hard?
Common, and precisely why partial exits appeal to many people. Without one, a position that reached 1.272 and then collapsed to your protective level converts an unrealised gain into a realised loss, which is psychologically brutal and financially worse. Moving the protective exit to breakeven once the first target prints removes that outcome at the cost of being stopped out flat more frequently. Neither choice is free.
Should I combine this with indicators?
Many traders do, most commonly with momentum readings or moving averages confirming direction before they act on a zone. Adding filters reduces the number of setups, which is usually the point. Be careful about stacking several confirmations, since each one you add makes the method harder to test and easier to rationalise around. Two well-understood filters beat five you’ve never examined separately.
Is the 50 percent level actually Fibonacci?
No, and it’s worth knowing. The genuine ratios derive from the sequence itself, producing 38.2, 61.8, and 78.6 among others. The halfway mark got adopted separately, partly through Dow Theory’s observation that corrections often retrace around half a move, and partly because it sits conveniently between the two ratios traders watch most. Useful as a reference point regardless, though nobody should claim mathematical authority for it.
How many resets are normal before a setup works?
Highly variable, and genuinely unpredictable. Directional markets can run through several cycles without a single reset, while sideways conditions produce failure after failure as price crosses back below each anchor. Anyone finding themselves resetting repeatedly is probably looking at a market unsuited to this approach right now, and stepping back beats forcing setups onto a chart that isn’t offering any. Patience costs nothing here.
Disclosure: Educational content only, not a recommendation to trade. Some material referenced here relates to courses the author sells. Consider your circumstances and, where appropriate, seek guidance from a regulated professional before risking capital.
