We tradethe future.
Pantheon runs volume across a wide set of sportsbooks, plus listed options. We build our own price for an event, take whichever side the books have wrong, and carry the whole lot as one risk book.
Most desks hunt for one big edge inside one market.
We take a small one and run it across every book.
Sportsbooks publish thousands of prices an hour and they do not agree with each other. One is slow to move, one is protecting a lopsided position, one has not seen the injury news yet. That disagreement is the trade. We price the event ourselves and take whichever side is mispriced, at whichever book is offering it.
None of that required permission. It required deciding the problem was solvable and then staying on it long enough to be right. Every part of this desk was built by people who were told they were too early, too small or too young to build it, and did it anyway. That is the whole method: put your mind to something and refuse to put it down.
The Pantheon withstands the test of time.
§ 02 — About
Four people who kept building the same thing.
Four people running trading strategies across sportsbooks and listed options. We came out of algorithmic esports competitions and kept the habits: write it, watch it lose, work out why, rewrite it.
Who runs it
Pantheon is run by students. We are spread across a few universities and time zones, which is most of the reason the desk covers as many books as it does.
- Georgia Institute of Technology
- Massachusetts Institute of Technology
- University of Oxford
§ 03 — MARKETS
Two markets, same math.
Sportsbooks
Thousands of prices an hour, and no two books agree.
A sportsbook has to quote both sides of a market whether or not it wants the risk, and it has to do it fast enough to get things wrong. We build our own view of the event, compare it to what is posted, and work the difference at as many books as will take the volume.
- Win-probability models built per league
- True prices synthesised across books
- Re-priced live as the game state changes
- Correlated positions carried as one book
Options
The same job, priced in variance.
Listed equity and index options run on machinery we already had. The model gives a fair value, we quote around it, and the difference between implied and what actually gets realised is the part we get paid for. Defined risk, not open exposure.
- Implied against forecast realised variance
- Skew and term structure
- Pricing into known catalysts
- Defined-risk structures, hard limits
§ 04 — EDGE
How we got here.
The rules fell out of the work: price it before looking at the screen, run a thin edge at size across a lot of books, hedge the leg you do not want to hold, and treat everything open as one exposure.
Origin
One person, one bot
It began as a solo entry in algorithmic esports competitions: write an agent, submit it, watch it get beaten, work out why, resubmit. Most of what we do now is still that loop.
The team
Grew to four
Three more joined over the following seasons. Everyone owns strategy code, and everyone is expected to argue with someone else's.
Now
From competition to trading
The competitions had a ceiling; markets do not. We moved the same approach onto sportsbooks and listed options, where being right about a distribution actually pays.
Design targets for the desk currently in build, not a record of past trading.
§ 05 — CAREERS
We are hiring people who argue with the number.
The desk is being built right now, so the first people in shape it instead of inheriting it. You will own a strategy end to end within a month and you will be asked to defend it to three people who want it to be wrong.
- Quantitative researchers
- Sports modellers
- Options traders
- Low-latency engineers
- Data engineers
careers@pantheonofducks.com