Fibonacci Retracements: Drawing Direction, Where 0 Goes, When to Redraw, and Which Levels Actually Hold
The conventions are simpler than they look — here is what the rules actually are, and what backtesting Fibonacci-based systems across 903 assets revealed.
Where 0 Goes, and Why the Debate Exists
The confusion starts with platform defaults. Some charting tools place 0 at the start of your drag and 100 at the end. Others reverse it. Neither is wrong — they are just expressing the same math differently. What matters is which end of your swing each number labels, not which direction you dragged the tool.
The convention that most traders and textbooks use: in a retracement after an upswing, 0 sits at the swing high and 100 sits at the swing low. Price pulls back from 0 toward 100. The 0.618 level is 61.8% of the way down from the high to the low. In a retracement after a downswing, the convention flips: 0 at the swing low, 100 at the swing high, and a bounce retraces upward toward the higher numbers.
If your platform defaults to the opposite labeling, the math is identical — just relabeled. The 0.382 retracement from the high is the same price as the 0.618 retracement from the low. Before assuming your drawing is wrong, check which end of your swing is labeled 0.
Drawing Direction: Anchor to Significance, Not to Time
The rule most sources agree on: drag from the more significant swing point to the less significant one. In an uptrend, anchor at the prior swing low and extend to the most recent swing high. In a downtrend, anchor at the prior swing high and extend to the most recent swing low.
Why significance rather than chronological order? Because the retracement levels describe how far price has given back relative to the full move — and the full move is defined by its extreme points, not by which came first in time. On most platforms the drag direction determines which end gets labeled 0 vs 100, so consistency in drag direction keeps your labels consistent.
If you drag low-to-high in a downtrend by mistake, your levels still mark the same prices. You will just be reading 0.618 where the convention would label 0.382. The price reaction, if any, does not care what your label says — but the labeling inconsistency will slow you down when comparing charts.
Why Price Stops at 0.618 More Than 0.382 — and When It Does Not
The honest answer is that neither level is reliably predictive on its own. A retracement stopping near 0.382 tends to occur in strong trends where momentum is intact and sellers are light. A retracement reaching 0.618 tends to occur when the original move's momentum has partially faded. Which level holds depends heavily on context: trend strength, volume, and what other structure sits near that price.
Fibonacci levels function best as confluence zones, not standalone triggers. When the 0.618 aligns with a prior resistance level that has turned into support, and volume contracts into it, the probability of a reaction improves — not because of the number itself, but because of the structural confluence at that price. A Fibonacci level sitting in empty space, with no confirming structure, carries much weaker evidence.
Our backtests covered 1-Hour, 4-Hour, Daily, and Weekly timeframes across 903 assets. We did not find that any single Fibonacci level operates as a universal reversal point across all markets and conditions. What we did find is that Fibonacci Pivot systems — which apply Fibonacci ratios to prior-period highs, lows, and closes to project support and resistance zones mechanically — ranked among the top five indicators for stocks across 22 assets and for crypto across 4 assets. That suggests Fibonacci-derived levels carry genuine structural information in those asset classes, more so than in others.
When to Redraw to a New Swing
The clearest redraw signal: price closes beyond your current 0 level — meaning a new confirmed swing high in an uptrend, or a new confirmed swing low in a downtrend. A single wick through the level is not a redraw trigger. A close beyond it usually is.
Practically, you redraw when your current grid stops describing the active move. If price has made a new significant high, your existing grid's 0 is stale — it is now anchored to a lower high that no longer represents the move's ceiling. Redraw with 0 at the new high and 100 at the most recent significant low.
The common mistake is redrawing prematurely: adjusting the grid every few candles to make the levels fit whatever price is doing. This is retrofitting, not analysis. A useful rule: redraw only when there is a confirmed new swing — a point where price has reversed, moved a meaningful distance in the other direction, and produced a clear structural high or low on your chosen timeframe. If you are redrawing more than once every several sessions, you are probably chasing noise.
What the Backtests Show — and What They Do Not
Across 660,005 backtests on 903 assets, only 26% of indicator-and-asset combinations beat a simple buy-and-hold strategy. Fibonacci-based systems were not exempt from that pressure. The Fibonacci Pivot strategy — which mechanically enters and exits at Fibonacci-derived levels — ranked in the top five for stocks (22 assets) and crypto (4 assets), making it one of the better-performing systematic approaches in those categories. It did not appear at the top of leaderboards for commodities, ETFs, or indexes, where different indicator families led.
That distinction matters. The backtests evaluated systematic rules applied to Fibonacci levels: enter when price touches level X, exit at level Y, same rule every time across every asset. Systems that performed well did so through consistency, not through discretion. When a trader redraws freely, shifts anchor points by eye, and decides which touches count after the fact, the results can diverge substantially from a rule-based baseline.
All results on IndicatorEdge are hypothetical backtests run on historical data with simulated transaction costs. They are not a record of real trading, do not reflect real fills or slippage, and are not financial advice. Past backtest performance does not guarantee future returns. Fibonacci levels are one input among many; nothing here should be treated as a signal to buy or sell any asset.
Questions, answered
Should I draw Fibonacci from low to high or high to low?
For a retracement after an upswing, draw from the swing low to the swing high so that 0 appears at the high and 100 at the low — the standard convention for labeling a pullback. For a retracement after a downswing, reverse it: draw from the swing high to the swing low. The direction of your drag on most platforms determines which end gets labeled 0, so match your drag direction to your platform's convention and stay consistent across all your charts.
Why does price stop at 0.618 sometimes and at 0.382 other times?
Shallower retracements stopping near 0.382 tend to occur in stronger trends where the original momentum is largely intact. Deeper retracements reaching 0.618 tend to occur when momentum has partially stalled. Neither level is universally more reliable — the level where price actually reverses depends on what other structure aligns near that price, not the Fibonacci ratio alone. Our backtests across 903 assets did not find a single level that consistently outperforms the others across all asset classes and timeframes.
When should I redraw my Fibonacci to a new swing?
Redraw when price closes beyond your current 0 level, confirming a new swing high or swing low on your timeframe. A wick through the level does not require a redraw. Avoid adjusting the grid frequently to fit current price action — the purpose of Fibonacci is to project levels in advance, not to explain what has already happened. If you find yourself redrawing multiple times per week, treat that as a signal that you are retrofitting rather than analyzing.
Are these backtest results real trading performance?
No. All results on IndicatorEdge come from hypothetical backtests run on historical data with simulated trading costs. They reflect what a mechanical rule-based strategy would have returned in the past — not real money, not real execution, and not a prediction of future returns. This is not financial advice. Indicator performance varies by asset, timeframe, and market conditions, and no result shown here should be the sole basis for any trading decision.
Every figure here comes from our own out-of-sample backtests, costs included — not a course or a guess. Educational information only — not investment advice. Hypothetical backtested results; past performance does not guarantee future results. Trading involves risk of loss.
Keep reading
Get the weekly edge report
The best-performing indicator per asset, what changed this week, and the honest caveats — straight to your inbox.