I Tested Whether Terence McKenna's "Timewave Zero" Predicts Bitcoin's Price Cycles — Here's What I FoundI Tested Whether Terence McKenna's "Timewave Zero" Predicts Bitcoin's Price Cycles — Here's What I Found

A curious experiment at the intersection of psychedelic philosophy and on-chain math.

The Setup: Two Weird Theories Walk Into a BarThe Setup: Two Weird Theories Walk Into a Bar

A few months ago I found myself down a rabbit hole with two ideas that, on the surface, have no business being in the same sentence:

Bitcoin's Power Law Theory (PLT) — the observation, championed by analysts like Harold Christopher Burger and PlanB's predecessors, that Bitcoin's price has followed a remarkably consistent power law since its first traded days. Plot log(price) against log(days since genesis block) and you get a nearly straight line with R²≈0.94. The booms and busts are just oscillations around this trend. The trend itself is eerily stable.

Timewave Zero (TWZ) — Terence McKenna's psychedelic-era mathematical theory that time itself has structure. He proposed that history moves in fractal waves of "novelty" — periods of increasing complexity and connectedness — and that these waves were converging toward a singularity point he called the "zero date." He originally pegged this to December 21, 2012 (the Mayan calendar end date). His collaborator John Meyer later recalculated it to July 8, 2018. The waveform is derived from the I Ching's 64 hexagrams, recursively self-similar across timescales.

Both theories make a similar structural claim: time has a shape, and that shape governs big events. So I thought — what if I actually tested whether McKenna's waveform correlates with Bitcoin's deviations from its power law trend?

Why This Is Actually a Reasonable Question (Sort Of)Why This Is Actually a Reasonable Question (Sort Of)

I want to be upfront: I didn't expect to find anything. This was a curiosity experiment, not a belief. But the question is at least well-formed:

• Bitcoin's PLT residuals (the boom/bust cycles above and below the trend line) are real, measurable, and not fully explained by the power law itself
• TWZ produces a deterministic mathematical waveform — it's not vague, it generates specific numbers for specific dates
• If TWZ had any relationship to Bitcoin's timing, it would show up as a statistical correlation between the waveform and the residuals

The honest version of this question is: does TWZ's specific phase structure correlate with Bitcoin's price cycles better than random noise?

How I Did It (The Methodology, Plain English)How I Did It (The Methodology, Plain English)

I had Claude write ~1,500 lines of Python to do this properly. Here's the pipeline:

Step 1 — Get the data. Pulled Bitcoin daily prices from Blockchain.com (going back to 2010) merged with Yahoo Finance. 3,790 trading days from August 2010 to December 2020 as the training set.

Step 2 — Fit the power law. Fit a log-log regression of price vs. days-since-genesis. Got slope = 5.68 (the theory predicts ~6.0), R² = 0.937. Extracted the residuals — the daily log-difference between actual price and what the power law predicts.

Step 3 — Generate the TWZ waveforms. Implemented McKenna's actual algorithm (from the published source code) for three candidate "zero dates":

• TWZ-A: McKenna's original 2012-12-21
• TWZ-B: Meyer's recalculated 2018-07-08
• TWZ-C: An arbitrary future date 2025-01-01

Step 4 — Test the correlation. For each TWZ waveform, computed the Pearson correlation with the PLT residuals at every lag from -180 to +180 days (361 lags × 3 signals = 1,083 comparisons).

Step 5 — The critical part: the null test. This is where most amateur analyses go wrong. A correlation being "statistically significant" (p < 0.05) means almost nothing with 3,600 data points — even r = 0.03 gives p ≈ 0.05. The real question is: does TWZ beat random signals with the same spectral shape?

I generated 10,000 "surrogate" signals — random waveforms that have the exact same frequency content as TWZ but with randomized phases. Then I asked: does the real TWZ waveform correlate with Bitcoin's residuals better than these random look-alikes? The 95th percentile of this null distribution was r = 0.473. That's the bar TWZ needs to clear.

The ResultsThe Results

Here's the honest summary:

None of the pre-registered TWZ variants beat the null threshold of r = 0.473.

The best TWZ signal (Meyer's 2018 date, TWZ-B) achieved r = 0.317 at a lag of +133 days. That's real — it's not zero — but it's only 67% of what you'd need to beat a random signal with the same spectral shape. It doesn't clear the bar.

Meanwhile, a simple 4-year halving sinusoid — just a sine wave with a 4-year period anchored to Bitcoin's first halving in November 2012 — achieved r = 0.662. That's 40% above the null threshold and more than twice as strong as the best TWZ signal.

What This Actually MeansWhat This Actually Means

The boring but important answer: Bitcoin's boom/bust cycles are explained by the halving cycle, not by Timewave Zero.

This makes intuitive sense. Every ~4 years, the block reward halves. This creates a supply shock. Miners sell less, scarcity increases, price rises. The market overshoots, corrects, and the cycle repeats. This is endogenous to Bitcoin's design — Satoshi literally programmed it in.

TWZ, by contrast, is an external signal derived from the I Ching. There's no mechanism by which it would influence Bitcoin's price. And the data confirms: it doesn't.

A few nuances worth noting:

1. The tiny p-values are misleading. TWZ-B's correlation of r = 0.317 has a Pearson p-value of ~10⁻⁸⁴. That sounds insane — but it just means "this correlation is distinguishable from exactly zero." With 3,600 data points, even a trivially weak correlation looks "significant" by that measure. The surrogate null test is the right test, and TWZ fails it.
2. The optimized TWZ date is interesting but circular. When I ran a grid search to find the best possible TWZ zero date (not pre-registered, just exploratory), the algorithm found December 16, 2022 — right near the FTX collapse bottom — with r = 0.520. That does beat the null. But this is pure data mining: I searched 522 candidate dates and picked the best one. That's not a discovery, it's overfitting.
3. TWZ-B's +133 day lag is curious. The best correlation for Meyer's 2018 date occurs when the TWZ waveform leads Bitcoin's residuals by 133 days. I don't know what to make of this. It could be noise. It could be a coincidence of the waveform's shape. It's not strong enough to act on.
4. The training set starts in 2010, not 2009. The first 592 days of Bitcoin's history had zero-price entries (pre-exchange era). This means McKenna's 2012 zero date (TWZ-A) only has about 2 years of pre-singularity data in the training set — not ideal.

The Philosophical FootnoteThe Philosophical Footnote

McKenna was a brilliant, strange thinker. Timewave Zero was his attempt to mathematize a mystical intuition — that history accelerates toward moments of maximum novelty. Whether or not the math works, the idea resonates with anyone who's watched the pace of change in the last decade.

Bitcoin itself is arguably one of the most "novel" things to happen in human history — a spontaneous, decentralized monetary network that emerged from the internet and has grown by 10 orders of magnitude in 15 years. If McKenna's theory had any empirical traction anywhere, Bitcoin seems like a reasonable place to look.

It doesn't have traction here. But the question was worth asking, and now we have a clean answer.

The CodeThe Code

Everything is open and reproducible:

• TWZ generator: Implements McKenna's published algorithm, verified against his original outputs
• PLT fitter: Log-log OLS regression with residual extraction
• Correlation engine: Pearson cross-correlation with Benjamini-Hochberg FDR correction
• Null testing: 10,000 spectrally-matched surrogates (Theiler et al. 1992 method)
• Pre-registration: Hypotheses were locked before touching the data

The pre-registration document was written and committed before any real data was analyzed — this prevents the common failure mode of "I found something interesting, let me call it a hypothesis."

FiguresFigures

(Four charts are attached — descriptions below)

Figure 1 — Bitcoin's Power Law: 15 Years of Eerily Consistent Growth (share_fig1_power_law.png)

This is the foundation of the whole experiment. The top panel shows Bitcoin's daily price on a log₁₀ scale (y-axis) plotted against calendar year (x-axis), with the orange power law fit running through it (slope = 5.68, R² = 0.937). The shaded band is ±0.5 log units — roughly a 3× price range in either direction. The grey dashed vertical lines mark the three halvings in the training set (Nov 2012, Jul 2016, May 2020).

The bottom panel is what we actually tested: the residuals — the daily difference between actual log price and the power law prediction. Positive (green) = above trend (bull market). Negative (red) = below trend (bear market). You can see the four distinct boom/bust cycles clearly, each one triggered by a halving. This residual series is what we tried to predict with Timewave Zero.

Figure 2 — Does Timewave Zero Predict Bitcoin's Boom/Bust Cycles? (share_fig2_results.png)

The main result chart. Each bar shows the peak absolute Pearson correlation (|r|) achieved by that signal against Bitcoin's residuals, across all tested lags. The red dashed horizontal line at r = 0.473 is the random noise threshold — the 95th percentile of 10,000 spectrally-matched surrogate signals. Any bar below this line is indistinguishable from random noise.

• TWZ-A (McKenna's 2012 date): r = 0.180 — well below threshold
• TWZ-B (Meyer's 2018 date): r = 0.317 — the best TWZ variant, still only 67% of what's needed
• TWZ-C (future 2025 date): r = 0.143 — near-zero
• Solar cycle (~11 years): r = 0.142 — also near-zero
• TWZ-D (hatched bar, exploratory only): r = 0.520 — this does beat the threshold, but the zero date was chosen after seeing the data via grid search, so it doesn't count as a valid result
• 4-Year Halving Cycle (green): r = 0.662 — clearly above the threshold, the dominant signal

The annotation on TWZ-B says "FAILS: Doesn't beat random noise." The annotation on the halving bar says "PASSES: Beats random noise by 40%."

Figure 3 — Correlation at Every Time Lag (−180 to +180 days) (share_fig3_lag_profiles.png)

This shows how the correlations were computed. Rather than just testing at zero lag (simultaneous), we tested whether TWZ leads or lags Bitcoin's cycles by up to 180 days in either direction. Positive lag = the signal predicts Bitcoin's future. Negative lag = Bitcoin's past predicts the signal.

The left panel shows all three TWZ variants. The orange line (TWZ-B) is the strongest, peaking at r = 0.317 at lag +133 days — meaning TWZ-B correlates best when it leads Bitcoin by about 4.5 months. But even at its peak, it never crosses the red dashed threshold lines.

The right panel shows the 4-year halving cycle for comparison. The green line rises steadily and peaks at r = 0.662 at lag +149 days, clearly above the threshold. The contrast between the two panels tells the whole story visually.

Figure 4 — How the Null Test Works (share_fig4_null_explainer.png)

This figure explains the methodology for anyone unfamiliar with surrogate testing. The left panel shows the actual TWZ-B waveform (orange, thick) alongside 6 example random surrogate signals (grey, thin). The surrogates have the exact same frequency content as TWZ-B — same power spectrum, same overall shape — but with randomized phase. They look similar to TWZ-B but are definitionally random.

The right panel shows the lag-profile correlations for those same signals against Bitcoin's residuals. The grey lines are the surrogates; the orange line is real TWZ-B. The red dashed lines are the 95th percentile threshold from all 10,000 surrogates. You can see that TWZ-B (orange) stays well within the cloud of random signals — it's not doing anything the random look-alikes can't do. This is the core of why H1 fails.

TL;DRTL;DR

• I tested whether Terence McKenna's Timewave Zero waveform correlates with Bitcoin's price cycles
• Used 10 years of training data (2010–2020), 10,000 random surrogate signals as a null test
• Result: TWZ doesn't beat random noise. Best TWZ signal: r = 0.317. Null threshold: r = 0.473.
• The 4-year halving cycle does beat random noise, at r = 0.662 — more than twice as strong as TWZ
• Bitcoin's timing is explained by its own supply schedule, not by McKenna's I Ching mathematics
• Null results are results. The experiment was worth doing.