Defining the “Advanced Trader”

Business Vertical: DraftKings Predict

Role: Sole Researcher

Tl;dr:

  • Corrected who the product was being built for

  • Set the design criteria with the Product Director and a VP

  • Cut the first release down to what traders actually needed, so development started in Q4 2026, earlier than planned

Background & Context:

  • DraftKings along with a few other nascent prediction markets wanted to reach serious traders: user who trade on desktop and use platforms like Interactive Brokers.

    • These are customers that the mobile version of DK Predict does not currently serve. And they are very valuable. They bring liquidity and market making.

  • Two assumptions stood in the way:

    • (1) The team believed it needed a big feature set before launch.

    • (2) An existing estimate said ~6% of Predictions users were advanced traders. It was based on rough criteria, and more research was about to build on it.

My Approach:

  • Advanced traders are rare and hard to find, so I had to adjust my methods as I went.

  • Screener survey. Too few usable people came back.

  • Behavioral data. I queried Snowflake myself, using an AI-assisted setup in VS Code, and pulled users whose trading behavior looked advanced.

  • Long interviews. How they trade, what tools they use elsewhere, and what it would take for them to switch.

  • Prototype sessions. I watched them use the desktop build to see what they used and what they ignored.

  • The data found the candidates. The interviews checked whether the data had found the right people.

Findings:

  • Behavior alone overcounts. Only ~40% of the traders my criteria flagged were truly advanced. The ~6% estimate rested on similarly rough criteria, so it likely overstated the segment too.

  • Looking advanced isn't the same as being advanced. The build had features that looked professional but lacked the basics that make them usable. Traders noticed right away.

  • The order logic didn't match how traders think. One trader pointed out that once shorting was allowed, the Sell button duplicated buying the other side. Selling Yes at 60¢ is the same bet as buying No at 40¢. Two ways to place one trade is confusing for traders and messy in the system behind it.

  • Traders were clear about what had to come first: the market-making and liquidity tools that let them trade seriously. Much of the rest could wait.

Impact:

  • I took the findings to the product team, and we set the design criteria on a call with the Product Director and a VP.

  • The first release was scoped to the essentials.

  • The trade slip was planned around one way to place each trade.

  • Development started in Q4 2026, ahead of plan.

  • Desktop is usually secondary at DraftKings. Here it was central to the strategy, and the research helped set its direction.

What I'd do next:

  • Use the sharper criteria to size the real segment. Then measure the first release against similar traders outside the beta, looking at trading volume, retention and desktop adoption.