How Our Ratings Work
Two Signals, One Rating
Every card rating combines two independent models. The Fair Value Model estimates what a card should be worth based on its fundamentals — character popularity, rarity tier, artwork quality, set prestige, and artist premium. The Temporal Model predicts where the price is heading over the next 12 months based on price trajectory, volatility, set age, and market conditions.
When both models agree — fair value says undervalued AND temporal says appreciation ahead — that's a Strong Buy. When both say overvalued and declining, that's a Strong Sell.
Backtested Results
We backtested our combined signal across four market conditions on $20+ cards with 12-month horizons:
| Period | Market | Strong Buy | Strong Sell | Spread |
|---|---|---|---|---|
| Bear (Jan 2022) | +4.9% | +11.2% | -20.0% | +31.2pp |
| Recovery (Jan 2023) | +6.8% | +31.2% | -17.7% | +48.9pp |
| Bull (Jan 2024) | +55.7% | +83.3% | +35.0% | +48.3pp |
| Bull (Jul 2024) | +77.3% | +116.5% | +28.6% | +87.9pp |
Past performance does not guarantee future results. Spreads measure the difference in average 12-month returns between top-third (Strong Buy) and bottom-third (Strong Sell) rated cards.
Update Cadence
Ratings are published as monthly editions. We don't update ratings daily because short-term price noise would cause ratings to flip-flop without adding signal. The underlying data (PriceCharting monthly prices, PSA population) changes on a monthly cadence, so our ratings match that rhythm.
Between editions, we display live market data (latest prices, recent sales) clearly labeled as raw data — separate from the rated recommendation.
What We Don't Predict
Our model identifies cards where fundamentals don't match the current price, and cards whose trajectory suggests appreciation. It does not predict market crashes, reprint announcements, or cultural shifts. These are exogenous events no model can foresee. We recommend using our ratings as one input into your decisions, not as the sole basis.