Do Fibonacci Levels Help Predict a Bounce?
Version 1.0 · report 2026-09-10 · study recorded 2026-07-08

QUESTION AND FINDING
In this test, Fibonacci lines did not give us a better clue about where prices would bounce. Moving the lines to different percentages produced almost the same recovery rate. The familiar numbers alone were not a good reason to expect a rebound.

METHOD
Study universe: A selected group of 50 stocks and funds traded on stock exchanges. Failed or delisted businesses were not fully included.
Chart and anchors: Monthly bars. Earlier major monthly lows within a ten-year window, plus recent major weekly lows, were paired with the running high. Anchor classifications used full-history information, which limits predictive claims.
Comparison: Drew comparison levels from the same earlier lows and highs, using different percentages. The follow-up note says these levels could identify different months. We omit a separate test using random lines on the same months: those months also share the same future prices.
Measured outcome: Counted whether price rose from the touch month's low by at least twice the recent average monthly range within the next six months. The average uses the latest 12 months.
Time split: Before January 2016 versus January 2016 onward. Exact endpoint dates and comparison-group counts are not retained in the saved summary.
Available evidence: Dated internal result table and research script, reviewed for this report. Original observation rows were not located for this edition.

SOURCE
Source: TradingPal's research notes from July 8, 2026. The download contains the summary figures saved in those notes, not a list of every price movement we studied.

LIMITATIONS
- These are exploratory historical comparisons, not papers reviewed by independent academic experts or a forecast of the next trade. They do not measure a complete trading strategy's return.
- The study used a selected group of 50 stocks and funds. It was not a random sample of the whole market, and failed or delisted businesses were not fully represented. This can change the picture.
- Several observations can come from the same security or from overlapping periods. They are not independent coin flips. The saved summaries do not provide a confidence interval that accounts for that overlap, so we do not claim statistical significance.
- The July 8, 2026 research notes preserve the reported percentages and selected group counts. This edition does not include the original observation-by-observation dataset or an independently repeated analysis. Exact first and last observation dates are not preserved in those summaries.
- The historical anchor labels used information from the full price history. A low being several bars old does not remove that hindsight. This is a major reason not to present the result as a clean test of a rule a trader could follow at the time.
- This retracement result covers monthly bars only. The saved note says weekly replication had not yet been run. It does not answer questions about daily or intraday trades, every Fibonacci ratio, or every way people use the tool.

CITATION
TradingPal Research (2026-09-10). Do Fibonacci Levels Help Predict a Bounce? Version 1.0. Historical study recorded 2026-07-08. https://tradingpal.io/learn/research/fibonacci-retracement-bounce-study