Continuous Price-Weighted DCA — Interactive Model

I(p) = min{ b · p−k , m · b },  p = P / Pref,  I = 0 for p > θ
A tunable middle ground between dollar-cost averaging (k = 0) and value averaging (large k). Enter your own figures — nothing is stored or transmitted; all computation runs in your browser. Price paths are illustrative stress-test scenarios, not forecasts.

Your capital

Everything you're willing to deploy. No new money added later.
Optional. Excluded from cost-basis math; shown in total only.
25%
Withheld from the model entirely — for the case where the framework itself is wrong.
Main: 7,500 Reserve: 2,500

Model parameters

2.0
k = 0 → plain DCA (price ignored). k = 1 → fixed units per period. k ≥ 2 → convex; accelerates into declines.
50,000
Your "neutral" price. Below it, intensity > 1× ; above it, < 1×. The heaviest subjective prior in the model.
400
The sum deployed when price sits exactly at the anchor.
2.5×
Circuit breaker. Engages at p* = m−1/k
62,000
Above this price, skip the period; funds accrue to a pool, released on deep declines.
Choppy decline to ~35% below anchor, then bounce.
Avg cost
Units acquired
Deployed / main
Total held

Deployment intensity I(p)/b vs normalized price

your k k = 0 (DCA) k = 1 cap

Simulated path × per-period deployment

price deployed cumulative avg cost
#PriceActionDeployedUnits Cum. unitsAvg costMain leftPool
What this is. An interactive companion to a methodological working paper on continuous price-weighted DCA for cyclical assets. The point is not to produce a recommendation — it is to make visible what your parameters actually encode. Drag k and watch the average cost move: that difference is the price of a belief you cannot verify.

What it is not. Not investment advice. Not a forecast. The three price paths are stress-test scenarios chosen to expose the model's character under different regimes, not predictions of any asset's future. Backtests on hand-picked paths prove nothing. Crypto assets are extremely volatile; total loss of principal is possible.

Privacy. Everything runs client-side. No figures you enter are stored, logged, or sent anywhere.