Investing.com had a prediction markets concept sitting in Figma, designed by another designer. The stakeholders weren't happy with it. The design was heavily dark-themed, felt outdated, and didn't match the trust and clarity expected from a platform used by millions of traditional investors.

We were brought in to take over and rebuild the visual direction from scratch. The scope was specifically the trending page, delivered as a proof of concept. If stakeholders approved it after development, the rest of the prediction markets section would follow.

We rebuilt the visual system with a light mode priority and an investor-first design language. Every screen used real market data, including charts pulled from the sibling Market Predictions project we had already shipped for the same client.

The most complex piece was the poll/voting cards. Prediction markets are dense with data: charts, percentages, voting mechanics, outcome probabilities. Users don't want to read paragraphs about each market event. We went through seven to eight card iterations to find the right balance between contextual imagery, compact data, and scannability.

After multiple rounds of design revisions, the client requested a mobile version near the end of the delivery window. We delivered that within the same 48-hour timeline, no scope extension.

Speed without cutting corners. 48 hours, full web and mobile delivery. This was possible because of the collaborative rhythm we had already built with this repeat client across previous projects.

Visual calibration for the right audience. The first iteration came out too gamified. We dialed it back to feel credible for traditional investors while keeping the experience engaging. Knowing when a design is "too fun" for the context is a judgment call that saves weeks of back-and-forth.

Data density, handled. Seven to eight card layout iterations to make prediction market polls instantly scannable. Each card pairs a contextual image with compact voting data so users can engage without reading a wall of text.