Every methodology change, published openly.
Quant models are only trustworthy when their changes are visible. This page records every material change to PRISM's scoring methodology, validation rules, and historical comparability — including the ones that made our numbers look worse.
Backtest realism upgrade (numbers revised down, on purpose)
A full audit of the research pipeline led to four methodology changes that make reported performance more honest and strictly harder to achieve:
1. Entry timing: simulated entries now use the first trading day after a scoring snapshot, never the snapshot day itself, removing a same-day information edge.
2. Transaction costs: backtests now support explicit per-rebalance trading costs instead of assuming free trading.
3. Currency normalization: non-USD listings are converted to USD with daily FX rates, so international returns are no longer flattered or penalized by currency moves.
4. Point-in-time fixes: a one-day lookahead in a legacy backtest path and a trend-component date bug were found and fixed.
Effect: previously shown research figures were optimistic. New figures are the baseline going forward. We consider this a feature.
Scoring epoch 2026-04-13: comparability reset
A major overhaul of scoring semantics, confidence handling, and component contracts (scoring version 3.0.0). Scores produced before and after this date are not directly comparable, so all public performance tracking starts fresh from this epoch. Historical pre-reset scores are retained internally but never mixed into published results.
Silent model changes are how backtests lie
A model that quietly changes its rules after seeing results can always look good. Publishing every change — with its effect on reported numbers — is the simplest defense we can offer against that failure mode, including our own.
How changes are tracked
Every scoring run is tagged with a model regime, a scoring version, and a configuration hash. A methodology change that breaks comparability triggers a new regime and a fresh performance track record — never a restated one.