Logistic regression

Combines every indicator into a single score and turns it into a probability that price will be higher within N days.

Why it matters

It's the baseline every other model has to beat: if a more complex model doesn't clearly outperform it, that extra complexity isn't paying for itself. It's also the easiest to audit — every indicator has a fixed, visible weight, so it's possible to see exactly how much each one contributes to the final prediction, something the tree-based models below can't offer as simply.

Build it yourself

Drag the indicators' combined score and watch the logistic function turn it into a probability.

Probability % = 100 ÷ (1 + e^(−score))
Probability of a rise

Illustrative example numbers for practice — not real data.

How it works, step by step

  1. Each indicator (MVRV, SOPR, NVT Signal...) is multiplied by its own weight, learned from historical data during training: a positive weight means "when this indicator rises, the probability of a rise increases"; a negative one, the opposite.
  2. All of those products are added into a single combined score, which can be any number — positive, negative, large, or small.
  3. That score passes through the logistic ("sigmoid") function, which smoothly squashes it into a range between 0 and 1 — no combination of indicators can ever produce a probability of exactly 0% or 100%.
Honest limitation. It can only capture linear relationships between indicators and the outcome — if two indicators only matter when they appear together (an interaction between them), this model can't see that on its own.

Predictions

These are the real predictions stored day by day, each one produced by walk-forward validation: every day's was produced by a model trained only on prior cycles. They're published for transparency, not because there's an edge behind them — for price direction, the project's research found none that is defensible.

Probability of a rise within 7 days

Probability of a rise within 30 days

Probability of a rise within 90 days

Data last updated on 2026-09-09.

← All models