Elliott Wave Theory: The Three Rules, the Wave Pattern and Why Counts Differ

Elliott wave theory, or the Elliott wave principle, holds that market prices move in repeating patterns driven by crowd psychology: five waves in the direction of the main trend, then three waves against it. Ralph Nelson Elliott (1871 to 1948) developed the idea in the 1930s and published it in The Wave Principle in 1938. A. J. Frost and Robert Prechter later set out the rules in Elliott Wave Principle, first published in 1978.

The wave pattern: five waves up, three waves down

In a rising market, waves 1, 3 and 5 go up and are called motive or impulse waves; waves 2 and 4 pull back. The impulse is followed by a correction in three waves, labelled A, B and C. In a falling market the pattern is reversed. Each wave splits into smaller waves of the same shape, so the pattern repeats on monthly, daily and hourly charts. Elliott analysts call each of these levels of scale a degree.

How each wave usually looks, as Elliott wave references describe it; these are the method's descriptions, not results we measured:

• Waves 1, 3 and 5 advance with the trend, and each splits into five smaller waves.
• Waves 2 and 4 pull back, each in three smaller waves.
• Wave 3 is often the long one. One advance usually extends far beyond the other two, and LuxAlgo notes that Elliott literature treats third-wave extensions as the most common in stock indices. When wave 3 extends, StockCharts writes that wave 5 tends to be about as long as wave 1.
• Wave 4 often ends near the level of the previous, smaller wave 4, according to StockCharts' guidelines.
• Wave 5 often makes its new high on weaker momentum than wave 3, the divergence that the Elliott Wave Oscillator is meant to show.

The three rules of a valid count

LuxAlgo, citing Frost and Prechter, gives three hard rules for an impulse, and the counts on this page require all three:

1. Wave 2 never retraces more than 100% of wave 1. If price falls below the start of wave 1, the count is wrong.
2. Wave 3 is never the shortest of the three impulse waves 1, 3 and 5.
3. Wave 4 never enters the price territory of wave 1. In a rising impulse, the low of wave 4 stays above the high of wave 1.

Everything else is a guideline, not a rule. The best known is alternation: if wave 2 is a sharp move, wave 4 tends to be a sideways one, and the reverse. Another is that wave 3 often reaches 1.618 times the length of wave 1, a ratio tendency that LuxAlgo lists as not binding; we check it below. The rules have one recognized exception: diagonals, wedge-shaped five-wave moves, allow wave 4 to overlap wave 1.

One possible count on weekly Bitcoin

To avoid placing waves by hand, the chart below applies a stated rule. It finds the swing points with 10 weekly bars on each side, looks at every run of six alternating swing points, keeps those that obey the three rules, and labels the run with the largest overall move.

The three rules hold. Wave 2 ended at 49,000, above the start at 24,901. Wave 3, at 60,588, is the longest of the three. Wave 4 ended at 74,508, only 731 above the top of wave 1 at 73,777, so the count survives rule 3 by a thin margin. Wave 3 is 1.24 times wave 1, not 1.618, and wave 4 took back 57.9% of wave 3.

What followed: after the wave 5 high, Bitcoin fell to 57,800.19 in the week of June 29, 2026, a drop of 54.2%, below both the end of wave 4 and the top of wave 1. The last weekly close of the window, in the week of October 19, 2026, was 80,154.09. Whether that decline is an A-B-C correction, or whether the count itself was wrong, is a matter of interpretation.

A second valid count of the same chart

Change one setting, swing points with 5 weekly bars on each side instead of 10, and the same rule finds a different count that also obeys the three rules:

Here wave 1 runs from 24,800 in the week of June 12, 2023 to 31,804.20 in the week of July 10, wave 2 ends at 24,901 in the week of September 11, only 101 above the start, and wave 3 is the rise to 73,777 in the week of March 11, 2024. Waves 4 and 5 end at 56,552.82 and 71,997.02 in the weeks of April 29 and June 3, 2024. In this count, wave 3 is 6.98 times wave 1. Wave 5 ends below the top of wave 3, a shortened fifth that the three rules allow.

Does wave 3 extend wave 1 by 1.618?

We measured the guideline on the daily S&P 500 index from January 1970 to October 2026. Method: find every swing point with 5 bars on each side (then 10), take every run of six alternating swing points, keep those that obey the three rules, and compute wave 3 divided by wave 1. The runs overlap, so a long trend can appear in several of them.

With 5-bar swings, 88 of 1,347 runs of six points obey the rules; the median of wave 3 divided by wave 1 is 1.72. Wave 3 was within 0.1 of 1.618 times wave 1 in 8 counts (9.1%), within 0.1 of 2.0 in 7 (8.0%) and within 0.1 of 1.0 in 4 (4.5%). With 10-bar swings, 50 of 665 runs obey the rules, with a median of 1.54: 4 counts (8.0%) near 1.618, 4 (8.0%) near 2.0 and 6 (12.0%) near 1.0.

Two points stand out. Few runs of six swing points obey the three rules, so a mechanical count rarely finds a clean impulse. And among those that do, wave 3 was close to 1.618 times wave 1 about as often as close to 2.0. The ratio varies widely; 1.618 is not a special value in this sample. The sample is small and the swing rule is ours, so this is a description, not a test of Elliott analysis as practiced.

Criticism of Elliott wave theory

The main objection is the one the Bitcoin counts illustrate: the theory can fit almost any chart after the fact. In “Magic numbers in the Dow” (Cass Business School, September 2006, archived copy of September 18, 2009), Roy Batchelor and Richard Ramyar tested whether the ratios of successive trends in the Dow Jones Industrial Average cluster around round fractions or Fibonacci ratios, and concluded: “A few significant ratios appear, but no more than would be expected by chance given the large number of tests we conduct.”

How traders use Elliott waves

• Context. A count tells a trader where the market might be in a larger move, for example in a wave 3 rather than a wave 5.
• Targets. Fibonacci extensions are often drawn from waves 1 and 2 to estimate the end of wave 3.
• Invalidation. The three rules give a price that cancels the count, which can serve as a stop.

For an older account of trend phases, see Dow theory; for the broader idea of repeating phases, see market cycles. Our bench tests chart patterns detected by fixed rules; it does not test wave counts. More guides are in technical analysis.

Elliott wave courses

Video courses on Elliott wave analysis on Udemy (affiliate links):

• Technical Analysis Elliott Wave Theory for Financial Trading
• Predict the Market with Harmonic Elliott Wave Analysis
• Trading Harmonic Elliott Waves like a PRO
• Elliott Wave, Forex Trading With The Elliott Wave Theory
• Elliott Wave for Beginners
• Elliot Wave Principle, Key to the Market Behaviour