farabiCollector market intelligence · AI product · 2026
Pokémon Demand Lab
An AI-powered Pokémon card market-intelligence platform connecting live price evidence, collector demand, grading economics and private portfolio tracking.
Live product · Market analysis · Grading tools · Portfolio intelligence
Every card commands the market
Pokémon Demand Lab is an independent collector-market intelligence product created to help people understand more than a card’s latest price. It brings together current Pokémon card prices, collector interest, market direction, grading costs, potential upside and risk in one focused experience.
The product turns scattered public evidence into a clearer decision journey: discover cards, understand the market, establish a sensible buying price, test grading assumptions and follow the result over time. Its guiding idea is simple: every card commands the market, but the strongest decisions come from reading the full signal.
Today’s Market: evidence made visible
Today’s Market combines a daily Pokémon Card Index with live price checks, historical movement, market breadth and explainable confidence signals. Collectors can explore price direction, shortlist momentum, buying zones, grading economics and public attention without reducing the market to a single headline number.
The interface presents evidence with its scope and limitations visible. Methodology pages explain how indices, estimates and modelled scenarios are constructed, while refresh controls let users request current available prices. This creates an evidence-led market view designed for confident exploration rather than opaque predictions.
The methodology behind the market-intelligence model
Demand Lab grounds its AI-assisted analysis in a reproducible daily data pipeline. The Pokémon Card Index began at 100 on 8 February 2024 across 2,317 historical card versions, with every card equally weighted. Each date creates a repeatable random screen of 10,000 catalogue cards spanning different sets, eras, characters and rarities. From that broad screen, the system collects at least 200 valid public price comparisons against each card’s 30-day average, excludes unavailable checks and retains the last complete reading when the threshold cannot be met.
The daily score averages those price movements after limiting unusual one-off jumps, then applies the result to the latest index value. A separate graded-market model triangulates grading-company population reports with public sold-price bands grouped by age, language and grade. Its middle case uses 45 million graded cards at an average value of $133, deducts 5% for likely crossover, reholder and label-change duplication, and places the $5.7bn middle estimate within a transparent $3.8bn–$7.5bn range. PSA, GemRate, PriceCharting and Pokémon production figures provide the public evidence and reasonableness checks behind the model.
AI-assisted decisions, explained
The AI Advisor ranks card opportunities using current market evidence, buy limits, price momentum, grading potential and risk. Card Insight opens the reasoning behind an individual recommendation, while the Scenario Planner lets collectors test their own purchase price, grading probability, fees and return assumptions.
These tools work together as a transparent decision-support system. Recommendations are connected to observable evidence and clearly labelled confidence, helping users compare opportunities while retaining control of the final collecting or buying decision.
From discovery to a private portfolio
Demand Lab connects public discovery with personal action. Popular-card exploration and search lead into a Wishlist, Portfolio and grading workflow where users can organise owned cards, acquisition costs, grading progress and changing value in one place.
Owner-scoped persistence keeps saved collector activity private. By joining research, planning and record-keeping inside one product, Demand Lab replaces the fragmented mix of price sites, notes and spreadsheets with a coherent collector journey.
Product strategy, design and engineering
Farabi conceived and designed Pokémon Demand Lab as a complete product: its proposition, collector journeys, evidence model, information architecture, visual identity, editorial system and roadmap. He shaped the AI-assisted tools around explainability, useful market context and practical collector decisions, with implementation support from Codex.
The live platform uses Python, React, TypeScript, Vinext/Vite and Node.js, supported by Cloudflare D1 and R2, Drizzle ORM, owner-scoped data and an editable Content Studio. The architecture also exposes bounded read-only tools for compatible AI agents without placing private collector data on the public surface.
Built for a wider collector world
Pokémon is the live market focus, supported by the strongest available data and the deepest product workflows. The wider architecture has been designed to support comic, entertainment and sports cards, including Marvel, Star Wars, DC Comics and football when sufficiently reliable evidence becomes available.
That staged approach gives Demand Lab room to grow while protecting the clarity of the current experience. It is a living London-built product that connects collector culture, market evidence, grading intelligence and ownership into one ambitious platform.
Model blueprint / Explainable ranking rules
How the Pokémon card
shortlist is calculated.
Each card is compared using the same measures: price direction, ease of resale, likely grade, grading profit, buying price and current collector interest. Verified grading population is shown when available and left blank when it cannot be confirmed.
Collect
Prices, sales, graded values, release age and population
Compare
Put each measure on a consistent scale
Score
Estimate price-growth probability over 90 days, 12 months and 24 months
Check
Require the exact card, current evidence, at least 70% confidence and usable sales activity
Core equation
Σ wᵢ × xᵢ30% 90-day + 20% 12-month + 10% 24-month + 15% grading economics + 10% entry + 10% liquidity + 5% relevance.Grading economics
Σ P(grade) × resale value − costsThe average resale value across possible grades, minus grading, postage, insurance, packaging, tax, upcharges and selling fees.expected return after costs, placed on a 0–100 scalePSA 10 upside can improve the score, while PSA 9 downside remains visible.Safety checks
Owned card? → excludeAbove buy limit? → waitMissing PSA 9/10 or exact identity? → INSUFFICIENT EVIDENCEWeak evidence? → Unknown risk, never Low90-day weakness with 12–24 month strength? → label long-term recoveryVintage card? → maximum one in a 16-card shortlistPopular Cards method / Separate daily ranking
How today’s popular cards are selected.
Popular Cards is separate from AI Advisor. It checks 720 catalogue cards, removes cards priced at £10 or below, reviews public collector interest and market evidence for the strongest candidates, then selects 16. The list mixes lower- and higher-priced cards, with no more than three cards for one character.