Fibonacci Levels in Trading: How They Are Drawn, and 835 SPY Swings Tested

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-05 · Updated 2026-11-02 · Reviewed 2026-11-01

▶ Watch: How to Draw Fibonacci Retracement Correctly

Covered on this page: TradingView.

Fibonacci levels are five fixed fractions of one price swing, 23.6%, 38.2%, 50%, 61.8% and 78.6%, drawn as lines where a pullback might pause. We tested them on 835 SPY swings since 1993, rising and falling, and the odds of the trend carrying on slid down smoothly with depth, with no jump at any of the lines.

How it works: drawing the levels

Pick a swing low and a swing high. In September 2026, SPY’s daily low was $749.60 on 16 September and its high $775.14 on 22 September, a distance of $25.54.

Then mark fixed fractions of that distance below the high:

That table is the entire calculation. Charting software does it the moment you drag from one point to the other: from the low to the high in a rising market, and from the high to the low in a falling one.

Where the ratios come from, and why 50% is the odd one

Three of the five come straight from the Fibonacci sequence (1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89…). Divide any term by the next and the answer settles on 0.618: 55 ÷ 89 = 0.618. Square it and you get 0.382; cube it and you get 0.236. The 78.6% line is the square root of 0.618.

50% has no Fibonacci origin at all. It is the midpoint of the move, and traders were watching halfway retracements long before anyone drew a ratio from a thirteenth-century sequence onto a price chart. It stays on the tool because people expect it there.

A worked example: the September 2026 rise

The pullback from the 22 September high bottomed at $758.79 on 1 October. That is a retracement of 64.02% of the swing, 57 cents below the 61.8% line. SPY traded above $775.14 again on 5 October and reached $781.62 on 6 October.

Note what that hides. “The 61.8% level held” is a sentence you can only write once the low is in. While SPY was falling through $765.38 on the way down, three more lines sat below it and nothing on the chart said which one would be the floor.

The swing stops existing below its low. A trade under $749.60 would have ended the move being measured, which is why traders who use the tool put the invalidation there rather than at a line.

Fibonacci in a downtrend

Flip the tool and the levels become ceilings. Measure from a swing high down to the swing low, and each line is the low plus a fraction of the fall. The question becomes how far a bounce gets before the selling resumes.

The same weeks give a real one. SPY fell from $775.14 on 22 September to $758.79 on 1 October, a $16.35 drop. Its 38.2% line sat at $765.04, its 61.8% at $768.89 and its 78.6% at $771.64.

SPY closed 1 October at $763.99, already above the 23.6% line, then opened 2 October at $770.58 and traded as high as $772.65, 84.77% of the way back and through every line. On 5 October it took out the $775.14 high. Read as a downtrend, the pattern failed outright; read as a pullback inside the bigger rise, it was a normal low. Same bars, two swings, two verdicts.

The real weakness: the swing is a choice

Everything above depends on two points you chose. SPY’s low of $757.97 on 18 September was lower than the sessions either side of it, so it is a defensible swing low too. Measured from it, the 61.8% level is $764.53 instead of $759.36, and the same 1 October low becomes a 95.22% pullback, close to failure rather than a clean hold.

Nothing was miscalculated. The same tool on the same chart told two different stories.

The defense is procedural: draw it before you have a position, use the most obvious swing on the chart, and write down which two points you used. If you find yourself re-picking the swing, you are negotiating, not measuring, the same failure as on the trend lines page.

The golden pocket and the extensions

Most traders ignore three of the five lines and watch the band between 50% and 61.8%, here $762.37 to $759.36. That turns the tool into a zone rather than a line, which is more honest: price reacts to areas, and expecting a turn at an exact cent is expecting more precision than any of this has.

Extensions project the swing past its own high: low plus the distance times 127.2% or 161.8%. From the September swing that is $782.09 and $790.92.

SPY reached $781.62 on 6 October and turned, 47 cents short of the first. An extension sits where price has not traded yet, so the only thing at that price is arithmetic; the Fibonacci extension page covers them.

The original data

What does it mean when price comes back to a level? We took SPY’s daily bars from Yahoo Finance, 8,493 sessions from 29 January 1993 to 26 October 2026, at the prices traded.

The swing rule. A swing high is a daily high above the five sessions on each side, a swing low the mirror, and when two of a kind come in a row the more extreme one is kept. That gives 418 rising swings (swing low to the next swing high) and 417 falling ones (swing high to the next swing low).

For each depth, we kept the swings whose later prices came back that far before the swing extended, then asked which came first: the trend carrying on (a new high after a rise, a new low after a fall) or the start of the swing being broken.

Both columns fall smoothly. Reaching 61.8% on a rise left 28.42% making a new high first, against 28.67% at 60%, a depth nobody draws.

On falls, 61.8% gave 21.91% against 22.70% at 60%. Eleven more depths from 10% to 90% sit on the same two curves, from 52.28% down to 10.36% for rises and from 27.34% down to 7.34% for falls. One rising swing at 60%, and one at 20%, was left out because a single bar touched the depth and made a new high, so the order is unknown.

Falls were undone far more often than rises. 303 of the 417 falling swings saw SPY trade above the swing high before any new low, against 199 of 418 rising swings that broke their low first. That fits a fund that rose over 33 years; it says nothing special about Fibonacci, and every one of those 303 passed all five lines on the way.

The median rise in the sample was 5.49% over 9 sessions, the median fall 4.18% over 6.

The swing rule changes the numbers, not the shape. With three sessions on each side there are 644 rising and 643 falling swings; 61.8% gives 29.82% (rises) and 21.16% (falls) against 30.44% and 21.56% at 60%.

With ten sessions, 186 and 185 swings, 35.00% and 19.41% against 35.54% and 19.88%. Every rising swing and its outcome at each level is in the swing list.

Checked against Nasdaq. On Nasdaq’s daily bars from 26 October 2016, the same rule finds the same 116 falling swings on the same dates as Yahoo’s bars over that window, with identical counts at 50%, 61.8% and 78.6% (18, 15 and 8 new lows first, 95 highs broken).

The rising swings matched the same way, 117 of 117. Where pullbacks tend to end, rather than what came after, is measured on how to use Fibonacci retracement.

It is also one of the most crowded topics here: 381 of the 24,971 trading videos in this site’s corpus have Fibonacci or fib in the title, at a median of 8,105 views.

When it fails

Price goes through every level

All five lines, and then the start of the swing. That is how 199 of the 418 rising SPY swings ended, and 303 of the 417 falling ones. A retracement past 78.6% is barely a retracement any more, and once the start of the swing is broken the move you measured is over. There is no sixth line to fall back on.

There is always a level nearby

Five lines across one swing means no point between the 23.6% and 78.6% lines is more than 8.4% of the swing away from one of them. So “it reacted at a Fibonacci level” is very hard to say after the fact and be wrong.

The test that survives: did you name the level before, and did anything other than the line agree with it, such as a level price already reacted at or an old swing low?

The swing was chosen to fit

A different low gave a different 61.8. In September 2026 the choice moved it by $5.17. If the level moves whenever the trade needs it to, nothing is being measured.

Questions people ask

What is the most important Fibonacci level?

None stood out on SPY. Among 418 rising swings since 1993, the share making a new high first fell from 50.13% at 23.6% to 19.76% at 78.6%, and 61.8% sat on the same smooth curve as 60%, which is not a level.

Do Fibonacci levels work better in uptrends or downtrends?

On SPY the curves had the same shape both ways, with no step at any line. Falls were simply retraced in full more often: after reaching 61.8%, 21.91% of falling swings made a new low first, against 28.42% of rising swings making a new high.

Which two points should you draw Fibonacci from?

The most obvious swing low and high on the chart, chosen before you have a position. Pick a different point and the lines move: a later, higher September 2026 low moved 61.8% from $759.36 to $764.53.

Can Fibonacci levels be used as stop losses?

A line is a poor stop on its own, because 199 of 418 rising SPY swings went through all five. The point that ends the idea is the start of the swing, $749.60 in the September example.

What is the golden pocket?

The band between the 50% and 61.8% lines, treated as one zone. On the September 2026 swing it ran from $762.37 down to $759.36.

Related

Pullback is what these levels are measuring, and when one stops being a pullback at all.

Support and resistance is what turns a drawn line into a real one, the confluence test above.

And swing highs and lows decides which two points you measure from, which is the choice everything here rests on.

This page is educational, not financial advice. Test every idea on your own charts before risking money. Some links on this page earn a commission if you buy through them. It costs you nothing and it does not decide what appears here or in what order — how these pages are made is set out in our methodology.