A score is only useful if it’s honest. We backtest the model against history to check that when it says “X% likely,” the real outcome actually happens about X% of the time.
Across 9,335 historical predictions on 12 stocks/ETFs (2011–2025), the average gap between predicted and actual was 9.9 percentage points. Lower is better; under ~5 points is well-calibrated.
What this means: the model runs slightly conservative — over this period real outcomes tended to land a bit above its estimates, so read the odds as a cautious floor, not an optimistic promise.
Each row groups predictions of a similar confidence. The two bars should line up.
Each holding’s odds of reaching its target come from a standard lognormal price model, using its volatility (from ~4 years of history) and the time to your end date.
Expected growth is a market baseline (8%/yr), tilted by analyst consensus when available — not a stock’s own past return, which the backtest showed is misleading.
The whole-plan probability and range come from a Monte Carlo simulation of your holdings together (with their correlations) plus your cash.
The formula is transparent, and the same code runs both the live score and this backtest — no divergence.
These are model estimates, not a guarantee, and never investment advice.