What is a sports card worth before you bid?
Valhalla buys and sells blue-chip sports cards and memorabilia. We guided the brand, built the website, and built a pricing model that values a card, forecasts its price a week out, and gives a maximum bid. Medon calls it a pre-screener, and that is exactly what it is.
Card sales the model learned from
Buy or no-buy calls correct
In testing on the current version
First call to a working website
- Industry
- Sports collectibles
- Scope
- Brand guidance, website, pricing model
- Engagement
- Ongoing since February 2026
The challenge
Judgment does not scale one card at a time.
Valhalla's edge is people. Two members of its team bring 45 years of combined experience in the top end of the card and memorabilia market. That experience is the product, and it is also the bottleneck: every card they consider needs a number before anyone can decide whether it deserves a closer look.
They wanted two things from us. A public face that reads as serious and stays inside strict marketing rules. And a tool that puts a fast, consistent price on any card, so their judgment goes to the hard calls instead of the routine ones.
We came in with just a concept. OneWave built the brand, the website, and a custom AI model, and helped us look like a serious investment firm from day one.
The brand and the site
Guidance on the mark, a site built around it.
We had a working website eight days after the first call, under the firm's original name. When the team moved to the Valhalla name, we argued for a flat, simple mark and generated AI mockups to show the direction. Medon picked one, and his designer drew the final crest from it.
We rebuilt the site around that crest: navy and silver, editorial serif type, and a short brand film on the home page. The copy then went through four rounds of compliance review with the firm and its advisers. What launched says less than the first draft did, on purpose.
The pricing model
From a hundred trades to 300,000 sales.
The first version learned from the team's own track record: about a hundred past trades. That is not enough to train on, and we said so in the review. We used synthetic data to prove the mechanics, then rebuilt the model on real market sales. It now trains on more than 300,000 rows of sales data, with another 300,000 scraped and queued.
An analyst enters a card (player, year, set, grade) and gets a current price, a forecast one week out with a 90% range, a buy or no-buy call, and the maximum price that still clears the firm's buying criteria. In testing it gets about 70% of buy and no-buy calls right. It is a pre-screener, not a sourcing engine, and the firm describes it exactly that way. The model and the code belong to Valhalla.
How it came together
A working site in 8 daysFebruary
The first working version of the website, eight days after the first call and before the name changed.
New name, new markMay
Our guidance and AI mockups set the direction. Medon chose one, his designer drew the crest, and we rebuilt the site and added a brand film.
Review and launchJune
Four rounds of compliance review on the copy before valhallasam.com went live.
Pricing on real sales dataSeptember
The model retrained on 300K+ sales rows: current price, one-week forecast, buy or no-buy call, and a maximum bid.
The result
Judgment first, model second.
The website is live. The model prices any card the team enters and improves as more sales data goes in. The decision stays with the people who have spent their careers making it.
Disclosure: OneWave co-founder Gabe Keding sits on Valhalla's advisory board.
After
- valhallasam.com, through four rounds of compliance review
- Brand guidance and AI mockups behind the new crest
- Navy and silver site system with a brand film
- Card pricing: current value, 7-day forecast, 90% range, max bid
- A model whitepaper for the team’s own use
Expertise that needs a second opinion?
We build models that put a fast, consistent number in front of your team, so their judgment goes where it counts.
Start your project