AboutIndexResearchBeta
Sign in
Private Beta Now Open

Quantitative Risk Managementfor Pokémon Investors.

Sharpe-ranked cohorts, drawdown projection, and portfolio concentration analytics for serious Pokémon TCG investors. Methodology open. Picks private.

View the live Index
3,000+Cards modeled · ~1,700 with full cohort assignment
−40%Volatility in the lowest-risk bucket
2 papersPublic methodology
FreePrivate beta · invite required
What Hedonix does

Risk-adjusted analytics
for collectible markets.

Cohorts

Risk-adjusted cohort assignment

Every card sits in one of five Sharpe-ranked cohorts, rebalanced weekly. Multi-anchor validated.

See the Index
Portfolio

Portfolio risk analytics

Drawdown, concentration, and cohort composition for your holdings. Built for serious collectors.

Available in private beta
Research

Methodology open

Working paper, walk-forward backtests, and the underlying hedonic model. Published, not gated.

Read the research
The Product

Cohorts that carry
measurable risk profiles.

Hedonix sorts every card into a Sharpe-ranked cohort. The lowest-risk bucket has historically carried roughly 2× the Sharpe ratio and 40% less price volatility than the highest-risk bucket. Your watchlist gets analysed as a factor portfolio against those cohorts.

01

Sharpe-ranked cohorts, refreshed daily

Every card sits in one of five cohorts, ranked by risk-adjusted return potential. Validated across four independent walk-forward anchors. Your card is never just a price. It is a cohort assignment with a measurable risk profile. See the live track record on the Index page.

+40.5%
02

Portfolio risk projection

Your watchlist analysed as a factor portfolio. USD-weighted quintile composition, a 365-day forward-return cone, Herfindahl-Hirschman concentration score, and cohort-weighted drawdown projection under mild, moderate, and severe market shocks. The numbers a discretionary collector never sees, in the form an investor would want them.

+24.5%
03

Cohort migration tracking

Cards don't sit still. We snapshot every card's quintile daily so you can see who just entered the lower-risk cohort, who left the higher-risk cohort, who's been anchored for thirty days. Factor migration as a portfolio signal. The risk-state visibility a static snapshot can never give you.

Engine under the hood

Three calibrated models, methodology public.

Three calibrated models drive the cohort classification: an H6 hedonic OLS for PSA 10 (R² 0.914, LOSO 0.87), a sister OLS for PSA 9 (R² 0.920), and an XGBoost ensemble for raw (R² 0.97 in-sample, 0.83 LOSO). Each card is cross-checked against three data sources (PPT, JustTCG, PriceCharting) and flagged when they disagree. The cohort and its risk profile are what you actually trade on.

Read the working papers
What we don't do

Methodology,
not marketing.

Some things we deliberately won't ship.

No "Top 5 Cards to Buy"

We rank cohorts, not individual cards. Single-card signals don't survive multi-anchor validation. The Q5 cohort does.

No directional price calls

We don't predict where a card will go. We measure which cohort it sits in and the risk profile of that bucket.

No proprietary alpha hidden behind a paywall

The working paper is public. The model architecture is documented. What you pay for is the daily-refreshed app surface, not access to a secret.

No FOMO alerts

Cards move quintile slowly. The migration tracker exists. Push-notification spam does not.

How well does the model actually work

2.88 Sharpe across four anchors.

Risk-adjusted return validated across four independent walk-forward anchors. Q5 cohort: 2.88 Sharpe. Q1 cohort: 1.60. The lowest-risk bucket carried roughly 2× the Sharpe ratio and about 40% less price volatility than the highest-risk bucket, in every one of the four test windows.

Past performance does not predict future returns. Multi-anchor validated. Working papers at /research.

Read the working papers
Lowest-risk bucket
−40%
less price volatility than top bucket
Sharpe ratio
2.9 vs 1.6
risk-adjusted return, low vs high
Robustness
4 / 4
test windows where the effect held
How it works

Your portfolio,
scored daily.

Below is a fictional eight-card portfolio walked through the Hedonix pipeline. Cohort composition, drawdown projection, concentration, and cohort-weighted Sharpe sit on the same surface.

app.hedonix.tech / watchlist
Portfolio overview8 cards · $16,163
Umbreon ex 161 (SIR)

Umbreon ex 161 (SIR)

Prismatic Evolutions

$4,760Q5
Rayquaza VMAX Alt Art

Rayquaza VMAX Alt Art

Evolving Skies

$2,848Q2
Mew ex 232 (SIR)

Mew ex 232 (SIR)

Paldean Fates

$2,675Q5
Charizard ex 199 (SIR)

Charizard ex 199 (SIR)

SV 151

$1,685Q1
Umbreon VMAX Alt Art

Umbreon VMAX Alt Art

Evolving Skies

$1,650Q5
Charizard ex 234 (SIR)

Charizard ex 234 (SIR)

Paldean Fates

$935Q4
Lugia VSTAR Alt Art

Lugia VSTAR Alt Art

Lost Origin

$890Q3
Pikachu ex 238

Pikachu ex 238

Black Bolt

$720Q3
Beta status

What's shipped,
and what's next.

Hedonix is in private beta. Access is invite-only and free. The cohort engine, portfolio analytics, and drawdown stress-test are live. The roadmap below is the Pyramid plan from the strategy paper, phased and scoped. No “coming soon, maybe” vapor.

Available today
  • 3,000+ cards modeled across modern Pokémon (SV-era + selected SM/SWSH); ~1,700 receive full cohort + risk-model assignment (raw price ≥ $5 with full PSA pricing)
  • Daily Sharpe-ranked cohort assignment (Q5 lowest-risk → Q1 highest-risk), multi-anchor validated across four walk-forward anchors
  • Drawdown stress-test on your watchlist under mild, moderate, and severe market shocks
  • Portfolio analytics: USD-weighted quintile composition, forward-return cone, HHI concentration, sealed-market exposure
  • Cohort Pulse: live tracking of cards entering Q5 or exiting Q1, with 90-day per-card history
  • Performance benchmarked against the Hedonix Q5 long-only index (since-add basis)
  • Three calibrated quantitative models (PSA 10 OLS, PSA 9 OLS, raw XGB) running daily
  • Cross-source data triangulation (PPT + JustTCG + PriceCharting) with disagreement-driven risk flag
  • Watchlist with cohort-migration badges + plain-language driver breakdown
On the roadmap
  • Phase 1 · Tax-lot tracking (DE 1-year holding period, US 28% collectibles)
  • Phase 1 · Real-time email alerts when watched cards migrate between cohorts
  • Phase 1 · Saved screener views and shareable cohort filters
  • Phase 2 · Japanese Pokémon coverage (~5,000 cards)
  • Phase 2 · Cardmarket EU as a second exogenous price reference
  • Phase 2 · Cross-market correlation matrix between holdings
  • Phase 2 · Mean-variance rebalancing suggestions
  • Phase 3 · Public API access and webhook integrations
  • Phase 3 · Bulk CSV / PDF export and quarterly reporting
  • Phase 4 · Beyond Pokémon: sports cards, sealed products, enterprise access for auction houses and insurers

Beta access is gated by invite codes. Hedonix is a risk-management framework, not financial advice. Cohort assignment quantifies historical volatility and Sharpe across multi-anchor windows. Past performance does not predict future returns.

Philipp Baro
Built by

Philipp Baro

Bachelor's student, Goethe University Frankfurt

Methodology, not marketing. The model is published. The picks stay private.

More about Hedonix

Risk management
for serious collectors.

Hedonix is in private beta. Access is invite-only. Methodology is public. Picks are not.

Private beta · invite required · free during beta