Methodology

Sources

• Pokémon: PokéAPI (pokeapi.co) — species, pokemon, encounters, evolution chains We use species metadata (generation, capture rate, flags, habitat), Pokédex flavor text (English), abilities, base stats, evolution chains, and wild encounter data.
• MBTI: the four preference pairs as described in standard Myers-Briggs literature — treated as preferences, not abilities or fixed traits.
• Enneagram: nine core motivations modeled on the Enneagram Institute's type descriptions — core fear, core desire, centers, wings, and stress/growth directions.

All source material is paraphrased; the shipped dataset stores behavioral signals, not copied prose.

The shared personality space

Users and Pokémon live in the same vector space. Each species gets an mbtiVector with four signed axes in [-1, 1]: EI (solitary↔social), SN (concrete↔abstract), TF (analytical↔relational), JP (structured↔spontaneous) — plus nine Enneagram affinities in [0, 1].

A behavioral lexicon of ~50 weighted rules maps canonical Pokédex phrasing (e.g. “lives in packs”, “senses emotions”, “created by genetic manipulation”) to signed axis and affinity contributions. Ability names, game types, habitats, and structural facts (legendary status, friendship evolution, extreme stats) add small secondary signals. Contributions accumulate with diminishing returns and shrink toward neutral when evidence is thin.

Personality Match

match = 75% × MBTI similarity + 25% × Enneagram affinity

• MBTI similarity: per-axis distance between your letter pole and the species' vector, weighted by our evidence confidence on that axis.
• Enneagram affinity: your core type weighted 80%, wing weighted 20% (or core alone at 100% with no wing), read against the species' nine affinities.
• Scores express profile similarity — never how strongly you possess a personality, and never compatibility with other people.

Derived Global Rarity Index

raw = 35%·encounterScarcity + 25%·availabilityScarcity + 15%·captureDifficulty + 15%·specialStatus + 10%·acquisitionComplexity

• Encounter scarcity: best wild encounter rate (0-55%+ maps to 1-0); species never found in the wild score ~0.92.
• Availability scarcity: fraction of game versions (in its generation or later) where it's obtainable.
• Capture difficulty: 1 − captureRate/255.
• Special status: Mythical 1.0, Legendary 0.9, otherwise 0.
• Acquisition complexity: evolution prerequisites — trade, items, friendship, special conditions, baby stage, no wild presence.

The raw score is ranked across all 1,025 species into a percentile, then tiered: Common <45th, Uncommon 45-70, Rare 70-88, Very Rare 88-97, Exceptional ≥97. Rarity is never part of the personality score.

Reading a result — mirrors, not verdicts

A raw similarity number invites a bad question — “only a 63% match?” — so the result page leads with relative standing, not the percentage:

• Closest mirror: the highest-scoring species is called your closest mirror deliberately — a mirror is a reflection, not an assignment. The raw score stays visible one layer down.
• Rank & percentile: every species is ranked 1–1,025 by the same score; “top 4%” means fewer species sit between you and it than above it.
• Triad & shelf: ranks 1–3 are presented together and ranks 2–9 as “other selves”, because adjacent scores are usually within noise — the constellation is the honest answer, not the single winner.

Lenses & categories

The same profile is re-read through weighted lenses — every lens is a deterministic re-weighting of the same model dimensions, never a separate fabricated answer:

• Closest Mirror — the base match above.
• Motivation — Enneagram affinity dominates (what drives you).
• Social — the EI and TF axes dominate (how you move among people).
• Rare Resonance — similarity multiplied by rarity percentile: species that are both like you and seldom met.
• Wild Card — affinity for dimensions orthogonal to your profile's center of gravity: the match that shares an unexpected axis.

Categories (Inner World, Social Self, Drive, Wild Card, Rare Resonance) draw from these lenses with a quality floor — if nothing clears the threshold for a lens, that facet simply doesn't appear rather than showing a weak pick.

Assessment depth

The explorer offers three depths of the same instrument: Quick (25 questions, ~3 min — 4 per axis, 1 per Enneagram type), Standard (42, ~7 min — 6 per axis, 2 per type), and Deep (68, ~11 min — the full bank).

Shorter reads are honestly noisier: a 4-item axis has a wide margin, so a 52/48 split should be read as “genuinely close”, not as a verdict. The scoring only ever uses the questions you answered — no padding with neutral responses — and the report shows the split so you can decide whether the axis was resolved at all.

Confidence

Profile confidence (0-100) reflects the quantity and consistency of usable canonical material: number of distinct English Pokédex descriptions, how many behavioral rules fired, and whether genus/ability signals corroborate. Dataset mean: 38/100; 610 species are flagged low-confidence because canonical behavioral description is thin. We surface this rather than inventing traits.

Limitations

• Pokédex entries are flavor text, not psychology. Some species have dozens of descriptions; some have three terse ones — confidence reflects this honestly.
• The lexicon can't capture nuance like irony or context; a “fierce” Pokémon that protects its young shows both signals, and the vector reflects the balance.
• MBTI preferences are treated as a stylized compass; Enneagram types as motivations. Neither is a clinical instrument.
• This is an entertainment and self-reflection experience — not an official assessment, not affiliated with The Myers-Briggs Company or Nintendo/The Pokémon Company, and not a diagnosis.