Predictive Frameworks (and Their Failures)

What you'll learn

• The Allure and Danger of Prediction
• Stock-to-Flow: The Most Famous Failure
• Pi Cycle Top, Mayer Multiple, MVRV Z-Score
• Realized Cap, Realized Price, and Cohort Models
• Why Technical Analysis Often Fails in Crypto
• What Survives Honest Scrutiny
• For Deeper Reading

Key terms

Stock-to-Flow (S2F)

Price model published by PlanB March 2019 using stock/flow ratio. Predicted $98K Nov 2021 (actual: $57K) and $100K+ throughout 2022 (actual: $20K bottom). Effectively discredited by 2022. Textbook example of overfitting.

Pi Cycle Top Indicator

Created by Philip Swift 2019. Triggers when 111-day MA crosses 2x the 350-day MA. Accurate at the 2017 top (within 3 days), accurate at April 2021 local top, but did NOT trigger at the actual November 2021 cycle top.

MVRV Z-Score

MVRV normalized by historical standard deviation. Readings below 0 = bear-market lows; readings above 7 = cycle tops historically. The 2025 cycle peaked at ~4.5, below the historical 7-threshold (ETF participants changed the holder base).

Mayer Multiple

Created by Trace Mayer ~2017. Bitcoin price / 200-day moving average. <1.0 = buy zone, 1.0-2.4 = fair value, >2.4 = overheated. Descriptive (where market is) rather than predictive.

Realized Cap

Sum of every UTXO valued at the price when it last moved. Reflects holder cost basis. Less volatile than market cap. As of early 2026: ~$700-800B for Bitcoin vs ~$1.2T market cap. Introduced by Coinmetrics 2018.

Realized Price

Realized cap divided by total supply. Average cost basis of all holders. ~$35,000 for Bitcoin in early 2026. Long-term-holder realized price: ~$26K; short-term-holder realized price: ~$77K.

Wyckoff schematic

Early-20th-century framework by Richard Wyckoff identifying Phase A-E patterns of accumulation/distribution. Frequently overlaid on Bitcoin charts. Useful as descriptive heuristic; over-applied as predictive tool with poor track record.

Logarithmic regression bands

Bitcoin price plotted on log scale with regression curve and standard-deviation bands. Tops and bottoms historically near upper/lower bands. Descriptive rather than predictive.

Survivorship bias

Tendency to remember winning trades and forget losing ones. In TA case studies, examples where the technique worked are featured; counter-examples omitted. Across the population of TA practitioners, average return after fees is ~0 or negative.

Reflexivity

Self-referential dynamic where actions based on a prediction cause the prediction to materialize. When a TA indicator becomes popular, its triggers become self-fulfilling — and then self-defeating as everyone front-runs.

This time is different

Phrase warned against in finance, but sometimes genuinely true. Each Bitcoin cycle has had different participant composition; structural changes (e.g., spot ETFs in 2024) genuinely change market dynamics, reducing reliability of historical-fit patterns.

Long-term holder (LTH) accumulation

Rising LTH-cohort supply during a bear market. Has been a leading indicator of cycle bottoms. The signal is noisy and lagged but has been more reliable than price-based TA across multiple cycles.