Fibonacci Retracement Calculator

The number that matters is not the level, it is where you entered

Most Fibonacci calculators hand you nine prices and stop. That is the easy half, and it is not the half that decides whether the trade is any good.

Take the leg in the calculator above. Enter at the 38.2% retracement and your stop — placed beyond the swing that started the leg — is the whole rest of the range away. Wait for the 70.5% and the identical idea, with the identical target, has a stop less than a third of the size. Nothing about the setup changed. The reward-to-risk roughly tripled, because R:R is a property of your entry price, not of your analysis.

That is the entire argument for the deep end of the pullback, and it is why the panel above works the trade out for you rather than leaving nine numbers sitting there. The cost of waiting is the trades that never come back that far — which is a real cost, and one you can measure rather than argue about.

How to anchor a Fibonacci retracement without lying to yourself

The maths is trivial. The anchoring is where every disagreement lives, and it is almost always the reason one trader's fib “works” and another's does not.

Anchor the impulse, not the day. A Fibonacci retracement measures a pullback against the move that caused it. That means the two ends of one clean directional leg — the point it left from and the point it stopped at. Not the session high and low, not the last round number, not whatever makes the level land where you already wanted to buy.

Wicks or bodies — choose once. Both are defensible. Wicks include the stop run, which is what you want if your method is built around liquidity; bodies show where price settled. What is not defensible is switching between them trade to trade, because a fib that can be drawn two ways can be drawn to hit anything.

Direction sets where 0% sits. An up leg ran from the low to the high, so the retracement measures downward from the high. A down leg measures upward from the low. Get this backwards and every level inverts — the picker above exists so you cannot do it silently.

Draw it on the timeframe the leg belongs to. A fib anchored on a 4-hour impulse describes a 4-hour pullback. It will look inert on a 1-minute chart for hours, right up until it does not. Anchoring high and executing low is the standard workflow for a reason.

What the levels are, and which of them are real

23.6%, 38.2%, 61.8% and 78.6% come from the Fibonacci sequence — 61.8% is the golden ratio, and 78.6% is its square root, which is the only reason it appears on a chart at all. 50% is not a Fibonacci ratio.It is Dow's halfway rule, and it survives in every fib tool because it is the equilibrium of the leg and because enough people watch it that it behaves like one.

Equilibrium is worth more attention than the ratio it borrows. Above the 50% level, price is in premium — expensive relative to the leg, which is where a seller wants to be. Below it, discount. That single division does more work in most liquidity-based methods than the whole ladder of ratios above it, and the calculator prints it separately for exactly that reason.

As for whether 61.8% is real: there is no mechanism by which a ratio out of a number sequence should govern an auction between humans and machines. What there is, is an enormous number of traders leaving orders in the same narrow band — which makes the band a genuine cluster of resting liquidity regardless of whether the arithmetic earned it. Treat the level as a place where orders are, not as a law.

The OTE band, and what ICT is actually claiming

The optimal trade entry is the 62% to 79% stretch of the retracement, with 70.5% as the midpoint most rulesets name. Read plainly, the claim is not mystical: it says the best price in a pullback is the one closest to the level that proves you wrong, because that is where the stop is smallest and the target is unchanged.

What makes the band a band rather than a line is the second idea layered on top — that price frequently runs the low that started the leg before reversing, taking out the obvious stops on the way. Entering at 38.2% puts you in front of that run. Entering at 70.5% puts you behind most of it. That is a claim about where liquidity sits, and it is testable: our liquidity sweep reference covers the mechanic, and the liquidity primer covers why the stops cluster there in the first place.

The honest caveat: a deeper entry is a better price on the trades that fill, and no trade at all on the ones that turn at 50%. Whether that trade-off is positive on your instrument, your timeframe and your rules is a counting question, not an opinion — and you settle it by replaying a few hundred legs, not by reading about it.

Extensions: where the leg is going, not where it came from

Retracement measures inside the leg. Extension projects beyond it, in the leg's original direction, which is why the two get confused constantly and why charting packages disagree about the words.

The construction here is the common one: 161.8% means the leg again, multiplied by 0.618, measured past where it ended. 127.2% is the square root of 1.618 and tends to be used for a first partial. 200% is simply the measured move — the leg repeated — and is often the more defensible target of the two because it needs no ratio to justify it.

Extensions are targets, and targets are the part of a strategy most likely to be fitted after the fact. Pick the level in advance, write it into the rules, and let the backtest tell you whether the runner was worth holding for. The rules-writing guide covers how to phrase that so it survives contact with a real sample.

Sizing the trade the levels give you

A fib entry with no position size attached is not a trade. Once the calculator has given you an entry, a stop and a target, three numbers finish the job:

Size. The stop distance decides the position, not habit — a deep OTE entry with a tight stop supports a larger position at the same risk, which is most of the point. The position size calculator turns the stop distance into lots.

The win rate the ratio demands. A 1:4 setup only needs to work 20% of the time. The risk/reward calculator gives the break-even win rate for any ratio, which is the bar your backtested numbers have to clear.

Whether the account survives the streak. Deep entries miss more often, so the losing runs are longer than the win rate suggests in feel. The risk of ruin calculator tells you what size keeps that tolerable.

Test it before you trade it

Fibonacci is unusually easy to fool yourself with, because a level always exists somewhere near where price turned. Scroll back through a chart you have already seen and you will find the fib that worked every single time.

The way out is to draw the level before you can see the outcome. Replay the market one candle at a time, anchor the leg as it forms, mark the band, and log what happens next — a hundred times. What comes out the other side is a hit rate for your anchoring on your instrument, which is the only version of this question worth answering. That is what CRTLAB is for, and how to backtest ICT concepts walks through the method for exactly this kind of rule.

Frequently asked questions