Fibonacci Retracements in Futures Trading: How to Draw and Test Them on ES and NQ

Draw a Fibonacci retracement from almost any two points on a chart and one of the lines will probably sit near a turning point. That's the problem with the tool. It's flexible enough to explain almost any pullback after the fact. Before you trade it, you need rules that fix the swing points in advance and a way to count the misses.

Where the ratios come from

The Fibonacci sequence starts 1, 1, 2, 3, 5, 8, 13, 21, and each number is the sum of the previous two. Divide a number by the one after it and you approach 0.618. Divide by the number two places later and you approach 0.382. Three places later gives about 0.236. The 78.6% level is the square root of 0.618. The 50% line isn't a Fibonacci ratio at all. It's the midpoint of the swing, included by most platforms because traders watch it.

Nothing in that arithmetic says price has to respect these numbers. The usual argument is that many traders watch the same lines and place orders there. That's a hypothesis about behavior, and it can fade in conditions where few people are looking. I'd treat it as testable, not as a market law.

Choosing swing points

The retracement is only as good as the two points it's drawn from. Define the swing in writing before you look at results. For example: the most recent swing high and low on the 30-minute chart that are separated by at least a set number of points, or the highest high and lowest low since the prior session's close. If you choose swings after seeing where price reacted, every level will look like it worked.

When two reasonable swing choices give levels a few points apart, you can't tell which one the market is watching. That's a sign to pass on the trade or to require something else, such as a prior-session high, VWAP, or a fixed reference price, to agree.

A worked ES example (hypothetical)

ES rallies from 4780 to 4840, a 60-point swing. You plan to buy a pullback and want the 61.8% and 78.6% retracements.

Say you buy at 4803.00 with a stop at 4792.50, just below the 78.6% level, and a target at the prior high of 4840.00. Risk is 10.50 points, or $525 per ES contract ($52.50 per MES contract). Reward to the target is 37.00 points, or $1,850 per ES contract. That's about 3.5 to 1 on paper. On paper is the operative phrase: you have to get filled at 4803, stops can slip, and commissions come off both sides. If price blows through 4803 and never returns, you have no trade. If it reaches 4803, spends a minute there and runs to your stop, the geometry didn't protect you.

Where retracements get ignored

• Strong momentum moves. Price can pass through several levels in a single bar. Buying each level on the way down is a way to build a large losing position.
• Thin trade. Quiet periods give you fewer participants and noisier ticks. Check whether your own data shows a difference between periods.
• Ambiguous swings. Small, arguable swing points create arbitrary lines.
• Overlapping grids. If you draw from three different swings and lines are everywhere, you've created a zone, not a level.

Pairing with other reference prices

A retracement that lands near another reference price, such as a prior-session extreme, VWAP or a KPL, gives you two arguments for the same area. KPLs are fixed reference prices anchored to the RTH opening print at 9:30 AM ET, so they aren't available before the open and don't move during the session. Overlap is useful for deciding which levels deserve attention. It doesn't prove a higher probability of success, and I wouldn't claim that without test results.

Drawing it in NinjaTrader 8

NinjaTrader 8 has a Fibonacci Retracements drawing tool in the chart's drawing tools. Click the swing low then the swing high for an up move, and reverse for a down move. Open the tool's properties to see which ratios are plotted and adjust them if you want to match your written rules. Save a drawing template once you settle on a configuration. The NinjaTrader 8 help guide documents drawing tool properties.

Testing before trading

1. Write your swing definition, the levels you'll trade, entry, stop and target.
2. Apply it to a block of historical charts in order, logging every qualifying setup, including the ones where price didn't reach the level and the ones where it blew through.
3. Record R-multiples after commissions and a slippage assumption.
4. Hold a later period aside and check the same rules there without changes.

Thirty trades won't establish an edge, and fifty sim trades aren't live proof. A simulated fill also doesn't tell you how your order would have queued at 4803. If the numbers are flat once costs are in, the tool is a visual aid and nothing more, which is a perfectly fine thing to learn.