Elliott wave theory

Definition

Elliott wave theory (R.N. Elliott, 1930s; Frost-Prechter’s 1978 text is the modern canon) claims markets move in a repeating fractal pattern: five waves in the trend direction, three against, nested at every timescale, with wave relationships gravitating to Fibonacci ratios (indicator-fibonacci-levels). The KB carries it for literacy and with its evidence status stated bluntly: the theory has NO peer-reviewed validation, its rules permit enough alternate counts that failed forecasts are reinterpreted rather than falsified, and its practical record — including its most famous practitioners’ public long-horizon calls — is documented as unreliable. It is included because agents will encounter it constantly, not because the platform trades it.

How it works / structure

• The pattern grammar (as claimed): impulse waves (1-2-3-4-5, with rules: wave 2 never retraces all of 1, wave 3 never the shortest, wave 4 doesn’t overlap wave 1) and corrective waves (A-B-C, in a taxonomy of zigzags, flats, and triangles), nested fractally; Fibonacci ratios proposed for wave magnitudes.
• The epistemic problem (the load-bearing point): the rule set is loose enough that practitioners maintain multiple “counts” simultaneously — when price violates one, an alternate was ready; this makes the framework NARRATIVE-COMPLETE and prediction-empty in its common use (disc-process-vs-outcome cannot grade what cannot be pinned down; bias-anchoring attaches to counts).
• What survives inspection: the observation that trends proceed in impulse-consolidation alternation is real but not distinctive (every trend framework says it — ta-dow-theory said it first with less machinery); Fibonacci-level clustering has no documented edge beyond round-number/anchor effects (see indicator-fibonacci-levels’s evidence section).
• The honest test standard: a PINNED count published in advance, with invalidation levels, graded on samples — the platform’s requirement for any wave-based rule an agent proposes; folklore performance claims are not evidence.

When it applies

Literacy (wave vocabulary saturates retail technical commentary — agents must parse it); sentiment reading (widely-followed wave counts can mark crowd anchors the way round numbers do); as a case study in unfalsifiability (the KB’s clearest example of how a framework evades grading — pedagogically valuable for the falsifiable-thesis discipline).

Risk profile & failure modes

• Unfalsifiability by design (the central failure): alternate counts absorb every outcome — believers accumulate confidence without accumulating evidence; the platform treats ungraded frameworks as entertainment, not analysis.
• Forecast-horizon inflation: wave theory invites grand long-horizon calls (documented famous misses spanning decades) — maximal narrative, minimal accountability.
• Complexity as authority: the taxonomy’s elaborateness signals rigor it does not possess — agents must not confuse specification density with evidence.
• Selective showcase: published wave analyses are survivorship-curated — the charts that “worked” circulate; the alternates that died do not.

Evidence & limits

Frost-Prechter is the canonical text (labeled practitioner); no peer-reviewed validation exists; documented public track records of leading wave forecasters include severe, extended misses. The KB’s carrying position: vocabulary yes, hypothesis generator only under pinned-count discipline, standalone method no.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Pinned five-wave completion calls (published levels, no re-counting) precede reversals at better than chance across 50+ instances (the honest wave test)” — falsified by the graded sample.
• “Instruments at widely-published wave targets show measurable order-flow anomalies (crowd-anchor effect, independent of the theory’s validity)” — falsified by the flow measurement.

Cross-references

• The ratio machinery: indicator-fibonacci-levels (its evidence section governs)
• The honest ancestors: ta-dow-theory, ta-wyckoff-method
• The discipline it fails: disc-process-vs-outcome; the bias it feeds: bias-anchoring
• The lens: lens-technical