Education · Reference · Indicators
Fibonacci retracements: a grid, and what it is for
This one attracts more certainty, in both directions, than anything else in this section. One camp treats the ratios as a law of markets; the other calls the whole thing numerology. Both are answering a question that has not been separated properly, and separating it is most of what this page does.
Mak · 7 min read · Reference · Indicators, 8 of 8
The mechanics take one sentence. Pick a move — a low to a high, or the reverse. Draw horizontal lines at fixed percentages of that distance. When price pulls back, those lines are where you watch.
Where the numbers come from
The Fibonacci sequence is the one where each term is the sum of the two before it: 1, 1, 2, 3, 5, 8, 13, 21. Divide any term by the next and the answer settles down as you go further out, approaching 0.618. The rest of the grid is built from that one number:
| Level | Where it comes from |
|---|---|
| 61.8% | The ratio the sequence converges on |
| 38.2% | What is left over: 1 − 0.618 |
| 23.6% | 0.618 applied twice, then taken from 1 |
| 78.6% | The square root of 0.618 |
| 50% | Nothing. It is not a Fibonacci number. It is «half way back», which long predates any of this and was inherited from older market writing |
That last row is worth sitting with, because the 50% line ships turned on in every piece of charting software and is discussed alongside the others as though it belonged to the same family. It does not. It is there because halfway is an obvious place to look, which is a perfectly decent reason and a completely different one.
Two questions, usually asked as one
- Is the arithmetic sound?
- Entirely. The ratios are exact, the sequence is a real object, and the lines land where the maths says they land. Nothing to argue with here, which is why the topic feels more rigorous than it is.
- Do markets respect those ratios?
- That is a claim about behaviour, and it is not established. A ratio being true of a number sequence says nothing about order flow. The sequence turning up in some natural growth patterns says nothing about futures either — the inference from one to the other is decorative, and it is doing a lot of the persuading.
The honest reason to have them on a chart
A great many people draw these lines. Orders get placed at them, stops go behind them, and software puts them there by default. A level can become a reference point purely because enough participants are treating it as one — and that is a fact about the crowd, not about mathematics.
That reason is weaker than «the market obeys the golden ratio» and much sturdier, because it can be checked: it predicts that these levels matter most where the most people are looking — liquid instruments, obvious swings, common timeframes — and matter least on obscure charts and awkward moves. Which is what you tend to find.
It also sets the correct expectation. A shared reference point is a place where reactions are more likely to be visible, not a place where price stops. The useful question at one of these lines is never «will it hold», it is what happens when price gets there — and answering that needs something other than the line.
The anchor problem, again
There is one more difficulty and it is the same one that runs through the VWAP page: you chose the endpoints.
This is what makes the tool so easy to fit to the past. With several candidate highs and lows and five levels each, something will have landed near almost any turn, and the version that fits is the one that ends up in the screenshot. If you cannot say which two points you are using before the pullback starts, you are not measuring anything.
You will also meet the band between roughly 62% and 79% given a name and treated as a special entry zone. That naming belongs to a particular school of analysis with its own vocabulary — the abbreviations page marks those terms so you can read other people's material without being lost. It is deeper than the levels above it. It is not a separate discovery.
Where it is genuinely worth the space
The defensible use is as one input among several that are measured differently. A retracement level that happens to sit on a price where real business was done — a high volume node, the previous session's low, the edge of the value area — is interesting for the second reason, with the first one adding company. A retracement level alone in an empty part of the chart is a line you drew.
That is the argument the article on confluence makes at length, and it is the right note to end this section on: the tools in these eight pages are at their most useful when they are asking different questions of the same market, and at their most misleading when several of them are asking the same question in different costumes and you mistake that for agreement.