Chiefs vs Bills
KC
LIVE
BUF
Intuitive, lively, and scalable prediction markets.
Context
Six months after launching Predictions, we had enough behavioral data and user feedback to understand where the experience was working and where it needed to evolve.
We learned that sports, crypto, and politics drove most engagement. Our existing navigation struggled to accommodate the growing breadth of markets, while a newer card-based approach was showing promise. At the same time, the core experience needed to become more competitive, and users were asking for more sophisticated ways to trade.
This created three priorities:
I worked with a small cross-functional team to define and deliver the next iteration of Predictions.
Scan markets across sports, crypto, politics, and new types.
Understand a market at a glance, then in detail.
Place a straightforward trade or a more advanced one.
Leave with a clear record of the position entered.
Prediction markets were fundamentally different from MetaMask’s traditional financial experiences. Users expected them to feel more immediate, lively, and approachable.
Rather than treating each improvement independently, I focused on establishing a consistent interaction model across discovery evaluation order confirmation.
Cards
Cards are the primary way users discover and evaluate markets, so I redesigned them around the information needed to make a decision quickly.
The system supports both standard prediction markets and versus formats for sports, while maintaining a consistent hierarchy around outcomes and odds. We also streamlined odds into ¢ values to more closely connect price with the value of a position.
Motion
I explored slot-reel-inspired motion for changing odds, reinforcing the dynamic nature of prediction markets.
Animating every market simultaneously quickly became visually overwhelming, so I constrained the behavior to cards entering the user’s focus. Motion became a contextual signal rather than decoration.
KC
LIVE
BUF
LIVE
M. Navone
F. Cobolli
Before January 2027
Before September 2026
Yes
No
Market details
Rather than creating a completely new detail-page pattern, I adapted the information architecture I had established for Token Details to Predictions.
This created greater consistency across MetaMask while allowing prediction-specific information and actions to live within a familiar structure.
Order experience
I extended our quick bottom-sheet pattern to support both straightforward trades and more advanced behaviors such as limit orders.
The goal was to introduce additional capability without making the default betting experience feel heavier.
Confirmation
Unlike many transactions elsewhere in MetaMask, research indicated that users wanted stronger confirmation after placing a prediction.
Instead of relying solely on a transient toast, we introduced a dedicated confirmation state that clearly communicated the position they had entered.
Supporting parlays
Sports represented an important part of Predictions usage, but the product lacked a behavior common to sports betting: combining multiple outcomes into a single position.
Rather than simply recreating a traditional sportsbook parlay flow, we explored how combinations could fit naturally into MetaMask’s existing interaction model.
Prebuilt combinations
Across two rounds of research, we found that users responded well to prebuilt combinations.
Instead of asking users to construct every combination from scratch, we provided starting points that could be reviewed and modified before placing a trade.
Building combinations
My initial explorations relied heavily on bottom sheets for constructing combinations. Through design critique and iteration, the cart metaphor emerged as a clearer model.
Users could progressively add outcomes while continuing to browse markets, with add-to-cart motion providing immediate feedback and reinforcing the more playful character of Predictions.
UXR version used a bottom sheet and colored buttons to convey selection. I ultimately decided that this was confusing as we're already using color to indicate active.
Revised mock after UXR leans on an outline treatment to indicate selection and works harmoniously with the new cart button.
Order experience
Although combinations introduced significantly more complexity underneath, placing one should not feel like learning an entirely new product.
I reused the core order experience established for individual markets, allowing the system to accommodate a new trading behavior while preserving familiarity.
Solution
The new Predictions experience addressed the core challenges we uncovered in V1. We created a scalable information architecture that supports every market category, outcome combination, and advanced order type, paired with a more inviting visual language that surfaces the information users need to make a prediction.
These updates establish a more expressive and scalable Predictions experience across core markets and combinations. The latest iteration is still rolling out, so we’re continuing to measure its impact on engagement and trading behavior.