How Radiant works

Community Fair Value (CFV) turns the crowd’s read on a card into a single number you can trust. It is built in three steps.

1. Collectors cast a forecast. Instead of an up/down opinion, you put a dollar figure on what a card actually sells for in a given condition. A forecast, not a wish.

2. Votes are weighted and trimmed. Forecasts are reputation-weighted and combined with a trimmed median — the top and bottom outliers are dropped before the middle is taken — so a burst of hype or a handful of trolls can’t hijack the result.

3. The result is anchored to reality. The community number is blended with and hard-clamped to real market data — what we call the band.

CFV stays suppressed until at least 5 distinct voters have weighed in on a specific card and condition. Until then we show the market anchor alone, so a single loud voice never sets the price.

Real market data sets a floor and a ceiling around every card — roughly ±15–25% of the market anchor, and tighter when the comps are rich and recent.

The community can move CFV anywhere within that glowing band, but it can never detach from real market data. That guardrail is the whole point: the crowd adds nuance the market hasn’t priced in yet, while the band keeps it from drifting into fantasy.

It is also a deliberate posture. CFV is a non-binding community estimate, never a “recommended price” — Radiant is price intelligence, not a marketplace.

Every estimate carries a confidence badge so you know how much weight to give it. Confidence reflects how rich the comps are and how many distinct voters have weighed in.

High— deep comps and a healthy pool of forecasts. Medium and low step down as the data thins. Comp-onlymeans there isn’t enough community input yet, so we show the market anchor alone rather than pretend a crowd exists.

Alongside the dollar value, a separate bullish / bearishgauge captures hype and momentum — where the community thinks a card is heading.

Crucially, sentiment is hype only and is never blended into the dollar value. We keep the two apart on purpose, so popularity and excitement can’t quietly distort the price.

Your vote weight is earned. As your past forecasts land near real sales, your influence grows — this is a skill and calibration game, not a popularity contest. Consistently right callers get louder; fresh accounts carry near-zero weight, which is our front line against Sybil attacks and brigading.

To be straight with you: accuracy-based scoring is still rolling out. Today the leaderboard ranks an activity-based XP score; the move to fully accuracy-weighted reputation is in progress.

Each grade is its own market. A PSA 10 is not a PSA 9, and neither is a raw card. We never average across grades to invent a blended number.

We track raw prices by condition (NM, LP, and so on) and graded prices across PSA, CGC, and TAG, alongside population reports. Together those power the “Should I grade this?” tool, which compares your raw card against the realistic graded outcomes.

Each printing is its own SKU, too. Holo, reverse holo, and normal are priced separately — for a card with no holo printing, the reverse holo is usually the sought-after one, and lumping them together would hide that.

Market data — catalog, images, raw and graded prices, and population reports — comes from Scrydex, our licensed market-data oracle.

These are market estimates, not realized sale prices. A realized-sold feed is on the way, and it will firm up graded values in particular.

Everything here is for informational purposes only and is not financial advice. Radiant is not affiliated with, endorsed by, or sponsored by Nintendo or The Pokémon Company— all card names and images are the property of their respective owners.