A two-layer framework that separates ‘where we are in the cycle’ from ‘what to do about it’ - built for long-horizon decisions on assets with strong boom-bust structures.
This invite-only tool is a two-layer framework designed to simplify
long-horizon decision-making:
Valuation Engine - measures how extended price is relative to its own
regime. Outputs a single oscillator centered on zero, bounded by
configurable overbought / oversold thresholds.
Trend Model - aggregates several independent long-term subsystems
into a composite strength score, mapped to a long / cash / short
stance.
The separation of “where we are in the cycle” from “what to do about it”
makes this model particularly suited to assets with strong boom-bust
structures (like crypto). Each layer answers one question, and they’re
designed to be read together, not as standalone signals.
The valuation oscillator is a weighted blend of six standardized
components. Each component measures a different facet of “extension”
relative to the asset’s own historical regime, then gets normalized so
that all signals sit on the same scale before they’re combined.
Each component is converted to a z-score over a configurable lookback (default 1100 bars), so different scales become comparable.
Weighted blend
Components are combined with user-controllable weights so you can emphasize the lenses you trust most.
Optional smoothing
A moving average (SMA, EMA, RMA, HULL, WMA, DEMA, TEMA, LINREG, ALMA, T3) can be applied to filter whipsaws. Default is RMA over 45 bars.
Bounded scaling
A tanh-based compression keeps the oscillator readable across cycles - extreme outliers don’t break the chart.
Volatility-aware re-expansion
After compression, the signal is re-expanded by rolling standard deviation, so “overbought / oversold” stays meaningful when volatility regimes change.
Threshold logic
Thresholds can be fixed (default ±1.5) or dynamic (k·σ bands around the rolling mean or around zero).
Dynamic (k · σ) - thresholds become mean ± k · σ over a rolling
window (default 100 bars, k = 1.5). Optionally center at zero
to use ±k · σ regardless of mean drift.
A right-corner summary table prints each component’s current z-score plus
a 🟢 / 🔴 status emoji, the Final Score, and the overall valuation
emoji. Toggle on/off via Show Summary Table. The table works in light
or dark mode (controlled by the Light or Dark Mode input).
The trend layer combines seven independent long-term subsystems.
Each subsystem produces its own long / short / neutral vote on the
prevailing regime, and the votes are averaged into a single composite
score in the range [-1, +1].The subsystems are deliberately diverse - each looks at trend from a
different mathematical lens (price-anchor alignment, persistence checks,
macro regime confirmation, higher-timeframe overlays, etc.). You can
toggle any subsystem on or off independently.
Disabling a subsystem reweights the composite among the remaining
active subsystems - so you can trim the model down to only the lenses
you trust without throwing the score off scale.
A colored signal line is plotted on the price chart (overlay),
optional bar coloring is available, and a right-corner table reports
the current strength (%), rate of change, and position state.
The Long-Term Trend & Valuation Model is not designed to catch every
swing. It is built to keep you aligned with the dominant trend, help
manage risk around cycle extremes, and provide a consistent language
for allocation decisions across timeframes and assets.
