Methodology
Why rankings differ between users
Same data, different ratios. Every number on AllincaRatio is calculated — never generated randomly or invented by AI.
The formula
Personalized Match Score = Σ (criterion score × criterion weight). Weights are normalized so the effective weights always total 100%. Example: 88×0.30 + 82×0.25 + 90×0.20 + 85×0.25 = 86.0.
Normalization
Each criterion is mapped to 0–100 against fixed category benchmarks: e.g. range 200→600 mi, 0–60 mph 9→3 s, annual ownership cost $5,500→$1,800. Higher is always better after normalization.
Price & Value
Not a simple inverse of price. 70% is a value ratio — a feature index (reliability, prestige, design, technology, safety) per $1,000 — and 30% is affordability within the current pool. A cheap, poorly equipped car does not win automatically.
Avoiding double counting
Ownership Costs uses running costs and depreciation only; purchase price lives in Price & Value. Efficiency measures consumption, while Range & Charging measures distance and charging speed.
Missing data
Missing values are never treated as zero or average. The criterion is dropped for that car, remaining weights are re-balanced, and the data-coverage indicator shows how much of your requested weight was backed by data. Below 80% is flagged as low confidence.
Two scores
Match score reflects your priorities. Quality score uses fixed weights (reliability 30%, safety 25%, functionality 15%, technology 15%, efficiency 15%) so it stays the same for everyone.
Search
Natural-language requests are parsed by transparent rules into year, budget, powertrain, body type, exclusions and priorities. Follow-ups like “Remove Tesla” or “Make prestige twice as important” modify the current ranking.
Data
This version uses a curated sample dataset of 16 vehicle configurations. Specs follow published manufacturer and EPA figures for one representative trim. Prices, ratings and ownership costs are illustrative estimates — not live listings. A live data-provider layer is the next step.