Ok, not sure how YouTube algorithm hasn’t pushed this podcast to me, but the guest list is amazing…I’m tempted to just start at the beginning and listen in ascending order. https://youtube.com/@allinbitcoinpod @CK is the best kept secret podcast

Great pod with WillyWoo on @CK … Very prescient and I wish I had seen it sooner. I finally got around to plotting bitcoin price in a clever way for perhaps seeing into the future and I’m making it available for anyone to verify/use/modify:

GitHub

GitHub - bg002h/BitcoinPowerLawModelExtrapolation: Junyper notebook with log log monthly price data plot, curve fit, and extrapolation

Junyper notebook with log log monthly price data plot, curve fit, and extrapolation - GitHub - bg002h/BitcoinPowerLawModelExtrapolation: Junyper n...

I think the chart (see attached image in bright or dark mode) suggests that we are not in a bear market yet, but of course there are always two ways to look at the same data…the competing hypotheses are we are in a bull market still with a bear trap vs. bitcoin decreasing volatility has trimmed the extremes of price movement. While my job in medicine is to tell the future, I think only time will tell which Bitcoin hypothesis is correct!

I reran my Bitcoin power law analysis with monthly, weekly, and daily data…5686 data points used for daily graph:

DIY Bitcoin Power Law Price Model

GitHub

GitHub - bg002h/BitcoinPowerLawModelExtrapolation: Junyper notebook with log log monthly price data plot, curve fit, and extrapolation

Junyper notebook with log log monthly price data plot, curve fit, and extrapolation - GitHub - bg002h/BitcoinPowerLawModelExtrapolation: Junyper n...

Made with Jupyter notebook code (free software for iPhone: https://apps.apple.com/us/app/carnets-jupyter-with-scipy/id1559497253, paid software for iPhone: https://apps.apple.com/us/app/juno-python-and-jupyter/id1462586500). To calculate price extrapolation in a future year like 2027 by hand, do this: (2027-2009)^5.655 x 0.0133=$166,885.13, but beware the very wide confidence intervals!

Final Bitcoin power law extrapolation chart. I wish I had done this 10 years ago! Code will soon be up on github:

GitHub

GitHub - bg002h/BitcoinPowerLawModelExtrapolation: Junyper notebook with log log monthly price data plot, curve fit, and extrapolation

Junyper notebook with log log monthly price data plot, curve fit, and extrapolation - GitHub - bg002h/BitcoinPowerLawModelExtrapolation: Junyper n...

If anyone knows Matthew Meženskis @1basemoney, show him this please!

Update to what is now perhaps the most important plot in bitcoin history, now with 1 and 2 std deviation bands (68 and 95% confidence intervals, and colors that AI says is pleasing and “premium”):

This just might be the most important chart in bitcoin history: And glancing at it reminds me of how much the 2014-2016 bear market sucked.

Bitcoin price extrapolation using log log plot and Jupyter notebooks:

GitHub

GitHub - bg002h/BitcoinPowerLawModelExtrapolation: Junyper notebook with log log monthly price data plot, curve fit, and extrapolation

Junyper notebook with log log monthly price data plot, curve fit, and extrapolation - GitHub - bg002h/BitcoinPowerLawModelExtrapolation: Junyper n...

Free app for iOS to run the Jupyter notebook code:

App Store

Carnets - Jupyter (with scipy) App - App Store

Download Carnets - Jupyter (with scipy) by Nicolas Holzschuch on the App Store. See screenshots, ratings and reviews, user tips, and more apps like...

Just download the csv file (mine is stored in a folder named bitcoin inside the notebook folder) and copy the jupyter-code into your notebook…

Bitcoin price extrapolation using Juno to run Jupyter notebook on iOS, vibecoded via grok…main point: $1M Bitcoin by mid 2033 would be reasonable. And $10M Bitcoin in 20 years would be reasonable.

Want to do your own Bitcoin price power law fitting at home? On iOS? Get data here (see image 3)

Bitcoinity.org

Average all the prices using this formula (iOS numbers) =AVERAGEIF(B2:K2,">0") (see image 4) Then copy the average column and paste as values and strip down the data to just the date and the average price, export as CSV file (see image 2). Then buy Juno for iOS ($30) and paste this script in a new Jupyter notebook (you’ll see output like image 1 after you run it): # Bitcoin Monthly Prices — Linear + Log-Log + Power-Law Fit & 20-Year Projection # Y-axis: major grid at EVERY power of 10 + minor grid at 2×–9× # FIXED: floating-point exponentiation to avoid ValueError on negative powers # Juno notebook compatible — February 2026 import pandas as pd import numpy as np import matplotlib.pyplot as plt from scipy.stats import linregress from matplotlib.ticker import FixedLocator, ScalarFormatter # ── CHANGE THIS TO YOUR ACTUAL FILE PATH ───────────────────────── csv_path = './bitcoin/BitcoinMonthlyPricesD.csv' # ── Load and clean data ─────────────────────────────────────────── try: df = pd.read_csv(csv_path) print("CSV loaded successfully!") print("\nFirst 5 rows:\n", df.head()) print("\nColumns:", list(df.columns)) print("Shape:", df.shape) except FileNotFoundError: print(f"File not found: {csv_path}") print("Tip: Use !pwd or check Files → Juno folder") except Exception as e: print("Error loading CSV:", str(e)) date_column = 'Time' value_column = 'Avg' df[date_column] = pd.to_datetime(df[date_column], format='%m/%d/%y', errors='coerce') df = df.dropna(subset=[date_column]) df = df.sort_values(date_column) x_dates = df[date_column] y_data = df[value_column].astype(float) # Summary print("\nDate range:", x_dates.min().strftime('%Y-%m'), "to", x_dates.max().strftime('%Y-%m')) print("Valid points:", len(y_data)) print(f"Price range: ${y_data.min():,.2f} – ${y_data.max():,.2f}") # ── Plot 1: Linear scale ────────────────────────────────────────── plt.figure(figsize=(11, 6)) plt.scatter(x_dates, y_data, color='blue', alpha=0.7, s=50, edgecolor='none') plt.xlabel('Date') plt.ylabel('Average Price (USD)') plt.title('Bitcoin Monthly Average Price – Linear Scale') plt.grid(True, alpha=0.3) plt.gca().xaxis.set_major_locator(plt.matplotlib.dates.AutoDateLocator(maxticks=12)) plt.gca().xaxis.set_major_formatter(plt.matplotlib.dates.DateFormatter('%b %Y')) plt.xticks(rotation=45, ha='right') plt.tight_layout() plt.show() # ── Prepare data for log-log plot ───────────────────────────────── valid = y_data > 0 y_valid = y_data[valid] x_dates_valid = x_dates[valid] # Years since Bitcoin genesis (2009-01-03) genesis = pd.to_datetime('2009-01-03') years_since = (x_dates_valid - genesis).dt.days / 365.25 years_valid = years_since.astype(float) # Filter to start from ~1 year onward MIN_YEARS=1.0 reasonable = years_valid >= MIN_YEARS years_log = years_valid[reasonable] y_log = y_valid[reasonable] # ── Log-Log Plot + Power-Law Fit + 20-Year Projection ───────────── if len(y_log) < 10: print("Not enough positive data points for fitting (after year 1 filter).") else: log_x = np.log10(years_log) log_y = np.log10(y_log) slope, intercept, r_value, _, _ = linregress(log_x, log_y) B = slope A = 10 ** intercept print("\nPower-law fit: y ≈ A × (years since 2009)^B") print(f" Exponent (B) = {B:.4f}") print(f" Prefactor (A) = {A:.4e}") print(f" R² = {r_value**2:.4f}") print(f" Points used = {len(log_x)} (from year {years_log.min():.1f} onward)")#print(f" Points used = {len(log_x)} (from year ~1 onward)") # ── Create figure ───────────────────────────────────────────── plt.figure(figsize=(12, 7.5)) # Historical data plt.scatter(years_log, y_log, color='darkred', alpha=0.85, s=60, edgecolor='none', label='Historical Data') # Projection years_plot = np.linspace(1.0, 38.0, 800) y_plot = A * years_plot ** B plt.plot(years_plot, y_plot, color='royalblue', lw=2.4, label=f'Fit + projection: y ≈ {A:.2e} × t^{B:.3f} (R² = {r_value**2:.3f})') plt.xscale('log') plt.yscale('log') plt.xlabel('Years since Bitcoin genesis (Jan 2009)') plt.ylabel('Average Price (USD)') plt.title('Bitcoin Monthly Average Price – Log-Log + Power-Law Fit & ~20-Year Projection') # ── EXPLICIT Y-AXIS TICKS & GRID ──────────────────────────────────────────── # Define exponents covering typical BTC price range exponents = np.arange(-2, 10) # 0.01 to 100,000,000 # Major ticks = every power of 10 major_positions = 10.0 ** exponents # use 10.0 to force float exponentiation plt.gca().yaxis.set_major_locator(FixedLocator(major_positions)) # Formatter: plain numbers, no scientific notation, no offset fmt = ScalarFormatter(useOffset=False) fmt.set_scientific(False) plt.gca().yaxis.set_major_formatter(fmt) # Minor ticks = 2× to 9× each decade minor_positions = [] for exp in exponents: for sub in range(2, 10): minor_positions.append(sub * (10.0 ** exp)) # FIXED: 10.0 ** exp plt.gca().yaxis.set_minor_locator(FixedLocator(minor_positions)) # ── Grid lines ────────────────────────────────────────────────────────────── # Major grid: solid at every power of 10 plt.grid(which='major', axis='y', color='gray', linestyle='-', linewidth=1.0, alpha=0.65) # Minor grid: dashed, faint at 2×–9× plt.grid(which='minor', axis='y', color='lightgray', linestyle='--', linewidth=0.5, alpha=0.45) # Keep x-grid subtle plt.grid(which='major', axis='x', color='gray', linestyle='-', linewidth=0.7, alpha=0.5) # Reasonable y-limits to show many decades plt.ylim(0.01, 1e8) # increase to 1e9 or 1e10 if projection goes higher # ── X-axis: starts at 1, log ticks ──────────────────────────── x_tick_values = [1, 2, 3, 5, 7, 10, 15, 20, 25, 30, 35] plt.xticks(x_tick_values, [str(int(y)) for y in x_tick_values]) # Today marker today_years = (pd.to_datetime('today') - genesis).days / 365.25 plt.axvline(today_years, color='gray', ls=':', lw=1.3, alpha=0.65, label=f'Today (~{today_years:.1f} years since 2009)') plt.legend(loc='upper left', fontsize=10.5) plt.tight_layout(pad=1.5) # ── Dual labels on bottom x-axis ────────────────────────────────────────────── # Get current tick locations and "years since" labels current_xticks = x_tick_values years_since_labels = [str(int(y)) for y in current_xticks] # Corresponding calendar years calendar_labels = [f"{2009 + int(y)}" for y in current_xticks] # Create combined labels (two lines per tick) combined_labels = [f"{ys}\n{cal}" for ys, cal in zip(years_since_labels, calendar_labels)] # Apply the two-line labels plt.xticks(current_xticks, combined_labels) # Optional: make the "years since" smaller or italic to distinguish for label in plt.gca().get_xticklabels(): label.set_fontsize(10) # slightly smaller overall # or more advanced: make top line smaller via manual Text objects (harder) plt.show()

I’ve overpaid for bitcoin…like really MASSIVELY overpaid…everyone did at the beginning. I plotted monthly price data since the beginning, did a log-log transform, fit a straight line to the data and noticed when I first bought Bitcoin, I overpaid. This is ironic, given that at the time I genuinely felt like I wasted my money, of which I had very little at the time (first week of residency). And now I know I literally about 10^3 time too much, if you follow the trend line bitcoin price would later establish in reverse. If you delete the data from über early idiots like me, you see the slope of the line substantially increases…and if you delete even more datapoints, everyone who bought the first year and a half overpaid by a considerable amount. It wasn’t until the 22nd month of trading that people started paying a “fair” price. Note that I get a much smaller exponent than most power law folks…huge difference between 4.4 and 5.8, so be careful extrapolating power laws too far into the future.