Fundamental Analysis vs Elliott Wave

Michael Whitman · Trading since 2017, full-time since 2020 · Published 2026-09-09 · Updated 2026-10-19 · Reviewed 2026-10-05

These have almost nothing in common, which is itself the interesting part. One is entirely about a particular business and cannot be applied anywhere else. The other applies to everything with a price and contains no information about any of it. The sharpest difference is how each one finds out it was wrong.

What each one is

Fundamental analysis reads a specific company’s financial statements to judge what the business is worth. It exists only where accounts exist. Fundamental analysis covers it.

Elliott wave counts five-wave and three-wave structures claimed to appear in every market at every scale. Elliott wave covers the model, and technical analysis is the chart-reading school it comes from.

One is specific and the other universal. A valuation is about one company and transfers nowhere; a wave count is applied identically to a currency pair, a wheat future and a technology share.

Where they differ

Where each can be used. Fundamentals have nothing to say about a currency pair, which has no earnings and no balance sheet. Elliott applies there exactly as it does to a company, which is genuine breadth.

What that breadth costs. A framework that works the same everywhere is not using anything particular about the instrument. That is a strength when there is nothing particular to use and a weakness when there is: a company’s accounts are information, and the count ignores them.

What timescale each operates on. A valuation resolves over years as earnings arrive. A count operates at whatever degree you choose, which sounds flexible and means the same reading can be produced at any timeframe.

How each finds out it was wrong. A fundamental forecast meets a filed number once a year, and the number either arrived or did not. A count meets only more price, and a count that is contradicted can be relabelled as a different degree or an extended wave, which is the framework’s characteristic escape.

What a valuation and a wave count share

Both are attempts to anticipate rather than to react, which is what separates them from a plain trend-following rule.

Either can be bent to fit what happened: a valuation reverse-engineered from a preference, a count assigned after the move.

And acting on either costs the same round trip. A value purchase and a wave-count entry pay the same spread and commission.

Which one to use: Fundamental Analysis or Elliott Wave?

Use fundamentals when the instrument has accounts and your horizon is years. Over that period what the business earns drives much of the result, and it is knowable from public documents.

Use a chart framework when there is no business to analyze. Currencies, commodities and index futures have no statements, so any method that requires them does not apply.

Use Elliott only with a written count and invalidation. That turns it into a claim, and it is what the framework asks for rather than what it usually receives.

And do not use a count to override an accounts-based view on a company. They operate on different horizons, so a disagreement is usually two true statements about different periods.

Who keeps the score

The accounts do it for you. Every year a company files an annual report, and the earnings figure in it either grew or shrank. A forecast that the business would keep growing meets a public, dated number that nobody can recount.

A wave count has only its own rules. Elliott’s model, as summarized on Wikipedia, sets three: wave 2 cannot fall back past the start of wave 1; of the three driving waves, the third cannot be the shortest; and wave 4 cannot overlap the price range of wave 1. If those rules carried information about markets, real prices should obey them more often than prices with the order of their days scrambled.

Do the three rules tell a market from noise?

Where to test it. Fundamentals cannot be applied to bitcoin or to a currency pair, so on those two markets a wave count is the whole analysis. That makes them the fair place to ask whether obeying the rules says anything about the market.

The test. Swings were marked mechanically on daily closes: a swing ends when the close reverses from its extreme by a fixed threshold, roughly five times the market’s median daily move, then again at roughly ten times for a larger degree; the first close of the sample starts the first swing. Every run of six consecutive swing points, five legs, was checked against the three rules in the direction of its first leg. The same was then done on 200 copies of each market with its daily changes shuffled into random order, which keeps every move and removes any structure.

Bitcoin since 2014. At a 7.35% reversal, 29 of 293 runs passed (9.90%), against a shuffled average of 9.45%, and 71 of the 200 shuffled copies scored higher than the real market. At 15.25%, 11 of 107 passed (10.28%), against 8.98%.

The euro against the dollar since 2003. At a 1.57% reversal, 32 of 347 runs passed (9.22%), against 8.68% shuffled. At 3.16%, 9 of 125 passed (7.20%), against 7.97%, and 119 of the 200 shuffled copies scored higher.

What that means. Roughly one run in ten passes, in real and scrambled prices alike, and every real figure sits inside the range the shuffles produced. A count that obeys the rules is therefore not evidence that a wave structure exists. The rule broken most often was the fourth-wave rule: 244 of the 293 smaller bitcoin runs had wave 4 reach back into wave 1’s range.

A run that passed: bitcoin, October to December 2025

At the smaller degree, the latest passing run was a decline. It started at the 6 October 2025 close of $124,752.53. Wave 1 fell to $106,467.79 on 17 October, wave 2 rallied to $114,472.45 on 26 October, below the start, wave 3 fell to $84,648.36 on 22 November, wave 4 bounced to $93,527.80 on 3 December, still under wave 1’s low, and wave 5 ended at $85,462.51 on 18 December. Wave 3 was the longest leg, and all three rules held. As the shuffle test shows, that alone does not make it a wave, and no annual report exists that could later say it was wrong.

Why universality is a trade rather than a feature

Because a method that ignores the instrument cannot use anything about it. The same count logic on a currency and on a company is not evidence that both behave alike. It is a consequence of the model having no input that could tell them apart.

And because a pattern that appears everywhere may be loose rather than universal. Price alternates between up and down swings on every timeframe, which produces alternating shapes. With a fixed swing size the rules pass about as often in scrambled data as in a real market, and a counter who may also choose the degree has far more room than that.

The original data

The prices. Bitcoin closes from Yahoo Finance, a UTC day each, mid-September 2014 to 4 October 2026: 4,401 closes with a median daily move of 1.419%. The euro against the dollar, daily closes from 1 December 2003 to 2 October 2026: 5,926 closes, median daily move 0.311%. Only closes are used, because the highs and lows in Yahoo’s currency history are not reliable.

The swings and the rules. The threshold is the median absolute daily log change times five (or ten), turned back into a percentage. A swing point is the highest close of an advance once the close falls by the threshold from it, and the lowest close of a decline once the close rises by the threshold. Each run of six swing points is tested in the direction of its first leg: point 2 must stay beyond point 0, leg 3 must not be shorter than both legs 1 and 5, and point 4 must stay beyond point 1. At the smaller bitcoin degree the first rule failed 148 times, the second 98 and the third 244.

The shuffle. For each market and degree, the daily log changes were put in a random order 200 times (seed fixed so the result repeats), rebuilt into a price path from the first close and run through the same code. The middle 95% of shuffled results, 190 of the 200, ran from 6.27% to 12.37% for the smaller bitcoin degree and from 6.16% to 11.38% for the smaller euro degree.

What this does not measure. It tests the three rules on mechanical swings, not a human counter’s judgment, and it says nothing about what followed a run. Every run, its six dates and prices and the result are in the Elliott rules file.

When it fails

The typical failure here is using a count to time an investment thesis. Somebody holds a company on the basis of its accounts, sees a wave structure suggesting a decline, and sells, abandoning a years-long argument on a reading that operates over weeks and cannot see the earnings at all. The position is then re-entered higher, or not at all, and the fundamental work that produced the original holding contributed nothing to the outcome. Mixing the horizons is the error rather than either framework being wrong.

A second failure is the retrospective recount, which preserves the framework and removes the lesson.

A third is ignoring a bad annual report because the thesis was long-term. A long horizon is a reason to wait for earnings, not a reason to disregard them once they are reported.

A fourth is applying fundamentals to instruments with no accounts, where there is nothing to read.

And a fifth is treating a count that passes the rules as confirmed. Random prices pass them too, so only an invalidation level written down beforehand turns the count into a claim.

Related

Fundamental analysis covers reading a specific company’s accounts. Elliott wave covers the wave model and its three rules. And technical analysis covers the chart-reading tradition the count belongs to.

This page is educational, not financial advice. Test every idea on your own charts before risking money.