Model-based estimates of compound annual growth over the next 12 years. The Range column shows the 5th-95th percentile band from historical forecast errors. The Years column is the number of years of data in the model: for Bitcoin it is the trailing 8-year span of prices used in the power-law fit; for other methods it is the calendar span of weeks where a 12-year forward return could be computed.
This is not investment advice: do not invest money on the strength of these figures. The estimates may rely on incorrect inputs or flawed models, and past performance is not a guarantee of future performance. Nobody can know what is going to happen. Do your own research. You are responsible for your own choices. Don't blame me if anything goes wrong.
| Asset | Method | R² | Years | 12y forecast | Range (5th-95th pct historical) | |
|---|---|---|---|---|---|---|
| Loading… | ||||||
Each point is one forecast from the table above: trailing 12-month annualised volatility on the horizontal axis and forecast 12-year CAGR on the vertical axis. Hover a point to see its label. Light grey lines show the best-fit gradient with each forecast held out in turn; the accent line is the full fit. Drag the slider below to replay history using each week’s trailing volatility and model input; forecasts still use today’s fitted regression (see note).
Loading…
Week by week from the linear fit above. The shaded band is the leave-one-out gradient range (same as the scatter plot). The vertical axis is Δ CAGR / Δ vol (e.g. −0.2 means +1pp Δvol associates with about −0.2pp ΔCAGR). This is a measure of the forecast reward per marginal unit of risk.
Implied forecast CAGR when trailing volatility is zero (the line’s Y-intercept). The shaded band is the leave-one-out intercept range. This is a meaure of the implied risk-free rate. The grey line is the effective federal funds rate (FRED FEDFUNDS). It doesn't seem to track very well, I don't know if you would expect it to.