Price pulled back exactly to 61.8%, turned around, and on the finished chart it all looks almost inevitable. The problem is that before the reversal there were also 23.6%, 38.2% and 50% nearby — a very handy set for explaining the past. Fibonacci levels become more useful when you choose the wave and the reaction rule in advance. We’ll show how to calculate the grid and how to turn it into a testable scenario on an intrade.bar demo account, without crediting the lines with the power to see the future.
What a Fibonacci grid actually measures
The grid shows what share of the chosen move price has given back during a pullback. If a rise started at 100 and ended at 120, a drop to 110 gives back 10 of the 20 units, or 50%. That is an exact description of the correction’s depth. On its own, it says nothing about whether price will bounce off 110 or keep falling.
The common levels are 23.6%, 38.2%, 50%, 61.8% and 78.6%. Most of them come from ratios used in analyzing the Fibonacci sequence. The 50% level is added as a convenient midpoint of the move; it does not become a classic Fibonacci ratio just because it sits in the same toolbar.
The practical point of the grid is to organize your observation. Instead of “the pullback is already pretty deep,” you get a specific range and a retracement share. Then you test how price behaves near the chosen area. Buyers and sellers are under no obligation to reverse the market out of respect for mathematics, however neatly the line is labeled.
How to calculate levels on a rise
Take a hypothetical rise from 100 to 120 on a linear scale. The height of the move is 20. For a retracement of share r, the price is 120 − 20 × r. At 38.2% you get 112.36; at 61.8%, 107.64. In this example, the zero retracement is at the 120 peak, and a full retracement is back at the original 100.
| Share of the rise given back | Price in the example |
|---|---|
| 23.6% | 115.28 |
| 38.2% | 112.36 |
| 50% | 110.00 |
| 61.8% | 107.64 |
| 78.6% | 104.28 |
In some terminals the labels run the other way. Before analyzing, check the tool’s orientation: 50% should land in the middle of the linear range, at 110. In your notes, give the percentage together with the price. That way you won’t confuse “a 38.2% retracement from the peak” with a different orientation of the same lines.
Reaching 112.36 means only that 38.2% of the chosen rise has been given back. It is not the probability of a successful trade. Nor does a value of 61.8% translate into 61.8% winning entries: retracement-depth percentages and outcome statistics measure different things, even though they look equally convincing.
How to draw the grid on a decline
If price fell from 120 to 100, the recovery is calculated upward from the low. For 38.2% you get 100 + 20 × 0.382 = 107.64; for 50%, 110. First determine the direction of the original wave, then check the labels. Don’t automatically copy the levels over from the rising example.
A recovery after a fall may turn out to be a correction, the start of a reversal or part of a sideways move. The grid does not choose between these options. If your idea assumes the decline will continue, you need additional observations: lower highs holding, a reaction at a zone marked in advance, a specific entry condition.
Save a snapshot of your markup before the market reacts. It should show the start and end of the wave, the timeframe and the chosen levels. Such a simple archive is more useful than a dozen images where the grid has already been fitted to a bounce that has played out.
Why choosing the points matters more than pretty percentages
On one chart you can see a rise from 100 to 120 and its last leg from 110 to 120. The midpoint of the big wave is at 110, of the small one at 115. Both calculations are correct, but they answer different questions. If you always pick the wave that price has already bounced off nicely, the tool will seem infallible.
Fix the scale: for example, the last completed impulse on the five-minute chart between confirmed extremes. Specify whether you use wicks or closes and how the end of the wave is confirmed. Confirmation takes time, so in a test you cannot assume the end point was known before your rule was actually met.
A small shift of a point changes every level. If you move the start of the rise from 100 to 102 with the same 120 peak, the midpoint becomes 111, and 38.2% becomes 113.124 instead of 112.36. This shows why two similar markups sometimes lead to different decisions. Precision to the third decimal place does not remove the ambiguity of the original choice.
When a new high appears, decide whether you update the grid and from what moment. Both versions should stay in your history. Otherwise the new peak will move the old levels after the fact and create the illusion that you knew the right reference points in advance. For a reproducible test, it is especially important to save the state of the chart exactly as it was before the entry.
Level, zone and confirmation
Price is rarely obliged to stop at a single mathematical coordinate. Different quote feeds, the price step and the choice of extremes all shift the grid slightly. You can study a small zone around a level, but its width should be set in advance. If you widen it after every miss, eventually the whole chart ends up inside the zone.
A line coinciding with an earlier support level helps you pick out an area to watch. However, two lines drawn from the same wave do not necessarily give two independent confirmations. The more grids and indicators you stack, the easier it is to find a pretty intersection by chance. One clear level and a clear reaction are more useful than a collection of coincidences with no entry rule.
Describe the reaction in terms of observable events: a candle touched the zone and closed above its upper edge; the next one did not make a new low; or some other event chosen in advance occurred. That can be tested. The phrase “price respects the level” sounds expressive, but different people can read completely different signals into it.
How the chart scale affects the calculation
A linear scale shows equal absolute changes as equal distances. A logarithmic one shows equal relative changes that way. On a large historical rise, the difference becomes noticeable. So two charts of the same wave may have different calculated levels if the tools use different methods.
Don’t confuse the visual scale with the grid’s calculation setting. In a charting tool, you should check both parameters. For a first training experiment, it is simpler to take a small range and a linear calculation, as in the 100–120 example. Switching modes once you already know about a successful reversal only improves your illustration of the past.
A scenario on intrade.bar: a pullback after an upward impulse
On the intrade.bar demo, choose one available currency pair and the five-minute chart. Find a rise with higher lows and a wave completed according to the rule you wrote down in advance. If you analyze in a separate TradingView, match the quotes with the FXCM feed that intrade.bar uses for currency instruments. Markup based on a different source may differ slightly.
For this training experiment, use only 38.2% and 50%. Suppose the wave has hypothetical coordinates of 100–120; the zone to watch lies between 112.36 and 110. Fix it before the first touch. A context requirement: the chosen area must contain a previously marked support level, not one found after a successful bounce.
Watch the first return into the zone. The training confirmation is a five-minute candle closing above 112.36 after visiting the area, with no close below 110. Only after that candle has finished do you consider the “Call” direction. This approach deliberately separates the touch from the reaction; it is not a proven profitable strategy.
Choose a 10-minute Sprint expiry for the test in advance and the same small training amount every time. Before confirming on intrade.bar, check the current quote, the expiry time and the payout percentage. After the trade opens, save the actual opening price. The trade starts from your entry price, not from the Fibonacci line. If the confirming candle has already risen sharply, part of any future bounce may have happened without you.
Skip the setup if price closed below the lower edge, a new peak appeared and the original grid lost its relevance, or you had to move the anchors to keep the idea alive. These skips have an explanation you can check. “I had a feeling it would bounce a bit more” is far too loose for one and the same method.
Why expiry needs a separate decision
Fibonacci levels measure price, not time. A bounce off support half an hour later does not mean a ten-minute trade ended in profit. If you change the expiry each time after seeing the result, you are no longer testing the method but your own ability to find a lucky moment in history.
intrade.bar has no early closing. So the expiry, the amount and the acceptable conditions matter before you enter. A known payout percentage makes the money side clear, but it does not speed up the market’s reaction to your grid. At a hypothetical 85%, an amount of 100 units gives 85 in profit if successful and a loss of 100 if not; use the actual percentage on the trade form, not the number from a training example.
How to check whether you need this grid
In a separate record, save the anchors, the time they were confirmed, the direction of the wave, the chosen zone, the reaction rule and your decision. After the trade, add the entry price, the expiry and the result from your intrade.bar history. The CSV export helps you reconcile trades, while the screenshots and explanations of your markup stay in your notes.
Compare bounces off support with and without the requirement to be in the Fibonacci zone. It may be that the support itself is doing the useful work and the grid adds nothing. Or it may help rule out some awkward entries. The answer comes from identical rules and new observations, not from the most striking picture.
In the next period, apply the same markup without retuning it. Count first touches and repeat touches separately if you have decided to study both. Start on the intrade.bar demo with one wave and one complete review: drawing before the move, waiting for confirmation, recording the skip or entry, checking the expiry. After that, the tool becomes a working scale rather than a chart decoration.
Questions and answers
How do I choose the right Fibonacci points?
Take the extremes of the wave you are studying, according to a rule chosen in advance. Write down the timeframe, wicks or closes, and the moment the end of the wave was confirmed. Don’t move the points to fit a bounce you have already seen.
Does 61.8% mean the probability of a bounce?
No. It is a share of the chosen range. The probability of a specific trading outcome is estimated separately, from your entry rules, the expiry and suitable data.
Can I enter as soon as price touches the level?
You can study such a rule separately, but a touch does not confirm a reversal. The training scenario in this article uses a closed candle with a reaction. Don’t mix these approaches in the same statistics.
How can I use the grid for digital options?
Set the price zone, the entry condition and the expiry separately. On intrade.bar, test the scenario on the demo, recording the actual opening price and the percentage. The grid does not choose the trade’s expiry for you.
Place your first practice trade
A demo account opens right after registration
