How to Use Fibonacci Levels for Retracements and Extensions. Practical Crypto Examples.

Fibonacci in trading doesn't predict the future. It shows zones where the probability of a price reaction is above average. It's not number magic, it's crowd-behavior statistics multiplied by the psychology of round percentages.

The tool looks simple on the surface and turns tricky in practice. Anyone can draw the lines in 10 seconds. Understanding why price sometimes stops dead at 61.8% and sometimes blows straight through without even noticing the line is a different matter entirely.

The problem most beginners run into isn't that they can't draw a grid. It's that they believe once a line is drawn, price is somehow obligated to bounce off it. The market owes nobody anything. A Fibonacci level is just a zone of heightened attention among participants, and everything after that comes down to volume, liquidity, and who's more aggressive at that moment, buyers or sellers.

What Fibonacci Levels Are in Trading

Fibonacci levels are horizontal lines drawn between two price extremes (the low and high of a move), dividing that distance according to the ratios of the Fibonacci sequence. Traders use them to find likely stopping points for a correction or targets for a trend's continuation.

The idea didn't originate in trading. The number sequence has been known since the 13th century, and it was 20th-century analysts who brought it into financial markets after noticing that ratios from the sequence kept showing up in retracement sizes on stocks and currencies, and later on crypto.

The tool took hold especially well in the crypto market for one simple reason. Huge numbers of retail traders trade here, and they mass-use the same platforms and the same indicators. When millions of participants are all watching the 61.8% level on Bitcoin at the same time, that level gets a shot at working simply from the concentration of attention and orders, not just from golden-ratio math. In that sense, "Fibonacci crypto" as a combination works even more reliably than on classic exchange-traded assets, precisely because of how dense the retail traffic is.

Fibonacci Numbers and the Golden Ratio

The sequence is built simply: each next number equals the sum of the two before it. 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, and so on. Nothing complicated, grade-school arithmetic.

The magic starts when you divide one number by its neighbor. 21 divided by 34 gives 0.618. 34 divided by 55 also converges toward 0.618. The further you go into the sequence, the more precisely it converges on this number, known as the golden ratio.

This ratio shows up in nature. The arrangement of sunflower seeds, the proportions of a nautilus shell, tree branching, the same proportion keeps peeking through everywhere. Traders carried the same ratios over onto price charts, on the assumption that market psychology follows similar proportions of pullback and recovery.

Sounds elegant, but here's the important part. Fibonacci levels work not because the market "knows math." They work because millions of traders are watching the same lines and taking the same actions near those marks. It's a self-fulfilling prophecy, amplified by mass adoption.

The mechanism behind that self-fulfillment is pretty down-to-earth. Thousands of traders simultaneously place limit buy orders near the same correction level. Those orders stack up into a real density level in the order book. Price arrives, meets that density, and bounces, not because the number is "magic," but because there's physically a stack of orders sitting there. The more popular a level is among market participants, the higher the odds it actually plays out.

Key Levels: 23.6%, 38.2%, 50%, 61.8%, 78.6%

The standard retracement grid consists of five levels.

23.6% is considered a shallow correction. Price pulls back only slightly, which usually signals trend strength, buyers or sellers aren't letting the opposite side grab much profit.

38.2% is already a more noticeable pullback. Short-term traders often take profit around this level, but the main trend keeps going.

50% mathematically isn't part of the Fibonacci sequence. The level got added in practice because price very often reverses right at the midpoint of a move. Psychologically, it's the point where the crowd finally loses confidence that the pullback will continue.

61.8%, the golden ratio, is the most revered level in the trading community. This is where corrections of strong trends most often end.

78.6% is a deep pullback, and behind it usually sits the question: is this still a correction, or has the trend already changed? If price pushes through this zone too, the odds of a full reversal of the entire move jump sharply.

The 100% level deserves a separate mention. Technically it's not a correction at all, it's a full return to the starting point of the move. If price gets there and doesn't find a buyer (in an uptrend), there's a high chance you're no longer looking at a correction but at a full change in market structure, a trend reversal rather than a temporary pullback.

If limit orders and the order book are still new territory for you, there's a deeper breakdown in the free lesson of our trading-from-scratch course, part of the "Trading From Scratch" playlist.

Below is a summary table of each level's strength and typical price reaction.

Fibonacci Retracement: How to Build It

Building the grid by hand isn't hard once you remember the sequence of steps and don't mix up the direction.

Step by Step: From Low to High

For an uptrend, you draw the grid from the low to the high of the move. The "Fibonacci Retracement" tool in most terminals and charting platforms is activated by a hotkey or through the drawing panel.

• Find the last significant low where the upward move started.
• Find the last significant high where the move stopped (at least temporarily).
• Click on the low, drag the line to the high, release.
• The terminal automatically places horizontal lines at the 0%, 23.6%, 38.2%, 50%, 61.8%, 78.6%, and 100% levels.

For a downtrend, the logic mirrors this, you drag from high to low. In this case, the retracement levels show how far price might bounce upward before the decline continues.

Beginners often mix up the direction and end up with an inverted grid. It's easy to check yourself: if the trend is rising, the 0% level should sit at the bottom and 100% at the top.

For precisely hitting the extreme, it's convenient to use the magnet function on the chart (in Secret Terminal it's triggered by holding LeftCtrl). The line automatically snaps to the candle's High or Low, so the grid is built from the actual extreme point instead of a spot a few pixels off to the left or right where your mouse cursor happened to land. A difference of a couple points on BTC looks trivial, but when you're building a grid on a large move, it shifts every retracement level by tens of dollars at once.

How to Identify Points A and B

Point A is the start of the move (the extreme the impulse launched from). Point B is the end of the current impulse (the last high for an uptrend, or the last low for a downtrend).

The problem is that a "significant" extreme has no strict mathematical definition. I usually take the last impulse that's obviously visible on the chart without zooming, the one that jumps out immediately. If you have to squint and look closely to find the point, the move probably isn't clean enough for a quality grid.

The practical rule is simple. The stronger and cleaner the impulse between A and B (minimal pullbacks inside the move itself), the more reliably the grid built from it will work.

A useful habit is checking your chosen points on a higher timeframe. A local low on the 15-minute chart might just be noise inside a larger move on the 4-hour chart. If the extreme holds up on both, the grid comes out more stable, and levels "break" less often when you switch to a different analysis scale.

Examples on BTC and ETH

Let's break it down with real data. In March 2024, Bitcoin moved from a local low around $60,500 to a high around $73,800 over a few weeks of powerful growth. That's a clean impulsive move with no sharp counter-trend candles.

Drawing the grid from $60,500 to $73,800, the 61.8% level landed at roughly $65,580. After the local peak, price pulled back exactly into that area, consolidated there for a few days, and then continued moving up with fresh strength. A textbook golden-pocket play.

The second example, ETH in early 2024. A move from $2,150 to $2,715 put the 38.2% level around $2,500. Here the correction turned out shallower. Price never even reached the middle of the grid, pulling back only a third of the move, which pointed to strong lingering demand below the market.

I've checked this kind of setup on BTC/USDT dozens of times, and noticed a pattern. The stronger the volume on the A-B impulse (visible through cluster analysis), the more often the pullback stays contained within the 38.2%–50% range instead of going deeper.

A third example shows how the grid works with a choppier move. In October 2023, Bitcoin moved from $26,900 to $35,200 amid expectations of spot ETF approval. The impulse wasn't perfectly clean, it had two 4-5% corrections inside it. Even so, the grid built from the extreme low to the extreme high put the 61.8% level around $30,070. That's exactly where price returned three weeks later before the next leg up, the correction took nearly a month, but the level held.

The takeaway from these three examples is simple. The cleaner and more powerful the original impulse, the more precisely the grid plays out. On choppy, stretched-out moves, the levels still work, but accuracy suffers and it takes longer to play out.

One more thing worth noting from practice. On high-cap assets, BTC, ETH, large top-20 altcoins by market cap, the grid behaves more predictably than on young coins with thin markets. The reason isn't the math itself, it's order book depth. The more independent participants trading an instrument, the closer real liquidity actually pools near the Fibonacci level.

Fibonacci Trading: Strategies

Theory without application is useless. Here's how you actually build the grid into trades.

Entry From the 61.8% Level (Golden Pocket)

The golden pocket is the narrow zone between 61.8% and 65%, where statistically most correction reversals happen. The strategy is simple on paper. Wait for the correction to reach this zone, look for reversal confirmation, and enter in the direction of the main trend.

Confirmation is the key word. A bare touch of the level with no additional signals is a coin flip. You need reversal candlestick patterns (pin bar, engulfing), declining sell volume as price approaches the level in an uptrend, or divergence on oscillators like RSI.

In my experience, this pattern fails around 25-30% of the time even with good confirmation. That's why the stop always goes beyond the 78.6% level or beyond the structural low/high of the correction, no exceptions.

Let's run the actual arithmetic of a trade. Say ETH rises from $2,900 to $3,400, then starts pulling back. The 61.8% level for this move sits at $3,091, the 78.6% level at $3,007. Entry is planned in the $3,091-$3,070 range (the golden pocket itself plus a small buffer), the stop goes at $2,995, just below the 78.6% level. Risk per trade comes out to around $76-96 per unit of the asset.

The first take-profit target is usually set at the 100% level, i.e., a return to $3,400. The risk-to-reward ratio in this layout comes out to roughly 1 to 3.5, which on its own already justifies the trade even at a win rate around 40%. It's math like this, not faith in a line, that makes the strategy work over the long run.

Fibonacci Extension: Profit Targets

Extension works differently from retracement. It projects where price might go after the pullback finishes, using the same impulse as the basis for the calculation.

Standard extension levels are 127.2%, 161.8%, 200%, and 261.8%. The grid is built through three points: the start of the impulse, the end of the impulse, the end of the correction. The terminal automatically draws zones beyond the original move.

Practice shows the 161.8% level most often becomes the first serious target for locking in part of a position. Not the whole position, part of it. Further movement to 261.8% happens less often, and usually only on very strong trending impulses with a steady inflow of volume.

Extension is especially useful for anyone holding a position with the trend who doesn't know where to take profit. Instead of guessing "maybe it'll go up a bit more," you get an actual number on the chart. I typically close a third of the position at 127.2%, another third at 161.8%, and trail the rest with a trailing stop to 261.8% or until a clear reversal signal shows up in the order book.

Fibonacci + Support/Resistance Levels

A single Fibonacci line is a hypothesis. A Fibonacci line lining up with a horizontal support or resistance level is already a strong argument. For more on how these levels work, see the article "Support and Resistance Levels in Trading".

For example, if the 61.8% retracement level lands exactly on a historical low that price has already tested before, the odds of a reaction go up considerably. Coincidences like this are called a level cluster, and they're where the most reliable trend trades get built.

The weak side of this approach is that Fibonacci clusters with horizontal levels don't happen every time. Sometimes you have to wait weeks for price to form a suitable configuration.

Another effective pairing is Fibonacci plus RSI. If price approaches the 61.8% level while RSI is showing bullish divergence at the same time (price makes a new correction low, but RSI's low is higher than the previous one), the signal doubles in strength. There's a detailed breakdown of this indicator in the article "RSI: How to Read Divergences". The combination works similarly with classic technical-analysis patterns: if a double bottom or a hammer candle forms at the golden pocket, three independent tools are confirming the same hypothesis.

Fibonacci + Order Book (Confirmation by Density)

This is where things get genuinely interesting for an active trader. A Fibonacci level says "a reaction might happen here." The order book says "there's a real big player sitting here, or there isn't."

The logic is simple. Price approaches the 61.8% zone. You check the order book, and if there's a density level sitting right there (a large limit order, not an empty order book without a single meaningful order), that strengthens the signal. If the density level sits for a while, it means the interest is real, not spoofing that gets pulled seconds before price arrives.

It's important to judge density not in absolute numbers, but relative to the instrument's normal volume. If the average cluster volume on this coin runs around $5 million, say, and the order book density at the golden pocket level sits at $12-15 million, that's already a meaningful order. There's a good chance the market can't push through that volume in one move, and it's from zones like this that a reaction lines up with the mathematical Fibonacci level more often.

Next, you check the tape. Price approaches the level, and if a buyer kicks in right at that moment (large spot buys or aggressive market orders on futures showing up on the tape), that's a third confirmation stacked on top of the mathematical grid and the horizontal level.

This is exactly how a trade gets built in Secret Terminal out of three layers of information. The chart with the Fibonacci grid shows the zone of interest, the density map and order book show whether real liquidity is actually sitting there, and the tape and clusters answer who's more aggressive right now at that level. Three signals at once is a completely different level of confidence compared to a bare line on a chart.

The terminal additionally shows a lifespan timer for a large order right in the order book. If the density at the 61.8% level has been sitting there for five to ten minutes and doesn't vanish as price approaches, that's different from an order that appeared a second before the touch and disappeared instantly. A long-lived density level usually signals genuine interest from a big player, rather than an attempt to move price with someone else's hands through spoofing.

Working with futures deserves separate attention. If a golden-pocket trade is planned to run for several hours, it's worth checking the time until the next funding rate settlement (the periodic fee between longs and shorts, charged every few hours). A position that's technically up 1% can have part of that profit eaten by funding working against you, if you hold through a settlement moment without accounting for it.

Mistakes When Working With Fibonacci

For scalping, the Fibonacci grid gets used less often than for medium-term trading, but it still comes up. How to build a strategy on short timeframes is covered in detail in the article "Crypto Scalping: Strategies for Intraday Trading". On the 1-minute and 5-minute charts, retracement levels get built inside local impulses spanning a handful of candles, rather than on daily moves. Accuracy here is lower due to market noise, so scalpers almost never use a bare Fibonacci level as their only entry trigger, they always add confirmation from the tape.

The first and most common mistake is overloading the chart. A trader builds grids on five different impulses at once, and the chart turns into a mess of lines. In the end, any price ends up "near some level or other," which cancels out the whole tool's value.

The second mistake is picking the wrong points A and B. Grab the wrong low, the wrong high, and the entire grid shifts, showing false zones. Check your construction on a higher timeframe if you're unsure about your choice of extremes.

Third, trading off a bare touch of a level with no confirmation from volume or the order book. A Fibonacci level is a probability zone, not a "buy" button. Without confirmation, entry win rates drop dramatically. From what I've observed, the difference in results between entering on a bare touch versus entering with order-book and tape confirmation reaches 15-20 percentage points of win rate over a stretch of a few dozen trades.

The fourth mistake is ignoring trend structure. Building a Fibonacci retracement in a sideways range, where there's no clear directional impulse, is pointless. The tool only works where there's a clear impulsive move between two points.

And fifth, a stop set too tight right behind the 61.8% level. Price frequently pokes through a level by a few tenths of a percent before reversing (this is called a false-breakout wick), and a stop that tight gets taken out prematurely. It's smarter to place protection beyond the 78.6% level or the nearest structural extreme.

A sixth mistake shows up among traders who moved over from stocks or forex into crypto and don't account for the difference in liquidity. On BTC and ETH, volume is sufficient for Fibonacci lines to play out similarly to traditional markets. On an altcoin with a $20-30 million market cap, one large holder can push price straight through three levels in a row without stopping, simply because there's physically nobody there to stand on the opposite side of the order book. Fibonacci and low-cap crypto is a combination that trips up beginners more often than it helps them.

How professionals read the market through the order book and clusters is covered in lesson of the same free course.

FAQ

• How many Fibonacci levels should you use in trading?

  Five basic retracement levels are enough: 23.6%, 38.2%, 50%, 61.8%, and 78.6%. For profit targets, extensions of 127.2%, 161.8%, and 261.8% get added. More lines on the chart don't make the analysis more precise, they just make price harder to read.

• Why isn't the 50% level part of the Fibonacci numbers?

  50% mathematically isn't part of the Fibonacci sequence, it's simply half of the move. The level stayed in the grid by tradition, because markets often reverse right at the midpoint of a move.

• What is the golden pocket in trading?

  The golden pocket is the zone between the 61.8% and 65% levels, where correction reversals happen most often. The name came from traders noticing a heightened concentration of reactions specifically within this narrow range.

• Do Fibonacci levels work on crypto?

  Yes, on BTC and ETH the grid plays out regularly, especially on timeframes of 4 hours and higher. On low-liquidity altcoins, accuracy drops, price gets pushed past a level by one large order more easily there.

• How do you tell a real bounce from a random coincidence?

  Look for confirmation in the order book and on the tape. If density shows up at the 61.8% level while a buyer is actively building volume on the tape, the odds of it playing out are higher than with a bare touch of the line.

• Can you trade using only Fibonacci levels, with no other tools?

  Technically, yes, but it's risky. The grid shows likely zones of interest, not guaranteed signals. Most traders combine it with horizontal levels and order book data before entering.

• Which timeframe is best for building the grid?

  For medium-term targets, traders use the daily and 4-hour chart; for intraday trading, the 1-hour and 15-minute. The higher the timeframe, the more reliable the level, but the less often it gets tested.

Fibonacci doesn't replace reading the market, it complements it. The level shows you where to look. What's actually happening in that zone gets shown only by the order book, the tape, and clusters.

An empty order book as price approaches a level means there's no real interest there, and the line most likely won't play out. A density level in the order book that's alive and doesn't vanish as price gets closer is a completely different story. Confirm levels with order book data in Secret Terminal before putting on a trade based on a bare line in the grid.