Data and Automation

Building a Market Dashboard That Tells the Truth

Design notes for a page that shows prediction market data without overstating it: which fields to show, what to filter, and what to label loudly.

A panel divided into four uneven regions

A dashboard is an argument about what matters. Choosing which numbers to show is most of the design, and the temptation is always to show the exciting ones.

The fields every row needs

The price, as a mid from live quotes. Not the last trade.

The spread. A price without its spread overstates precision. "62, market 60/64" is honest; "62%" alone is not.

Size. Depth at the top of book, on both sides. Determines whether the row is actionable at all. Depth and slippage.

Age. How old the observation is. On any derived figure, the age of its oldest input. Detecting stale data.

Time to settlement. Determines what any edge is worth annualised. Capital lockup.

Without size and age, a number is decoration.

What to filter before rendering

Filtering in the pipeline rather than leaving it to the reader is the main thing that separates a useful dashboard from a noisy one.

Exclude one-sided books. Exclude quotes older than a threshold. Exclude markets below a minimum depth. For cross-venue rows, exclude pairs whose observations are too far apart in time, and pairs that cannot be structurally matched. Matching markets across venues.

Then say somewhere what was excluded and why. A table that silently drops rows is hard to trust.

Default ordering

Not by the biggest number. The largest cross-venue gaps are dominated by staleness, thin books and mismatched questions, so sorting by gap size is a machine for surfacing artefacts. How to read a gap table.

Sort by something that already incorporates the filters — a quality-weighted measure, or a return over the actual holding period — and let the reader sort by raw size if they want it.

A small tag attached to a larger block
Every number needs an age and a size next to it.

Language

This is not a stylistic question, it is an accuracy question.

Say theoretical cross-market edge, because that is what a price difference is: what two prices imply, before costs. Never say profit, never say guaranteed, never imply a return. The distance between the headline and an outcome is made of spread, fees, slippage, capital cost and the chance that two questions resolve differently. What a gap is worth.

Say what a figure is derived from. If a grade is automated, say it is automated and that the contract rules are the source of truth. Equivalence grades.

Show the method

A dashboard whose numbers cannot be explained is a dashboard nobody can check. Publishing how a figure is computed — including its filters and its limits — costs nothing and is the difference between a tool and an oracle. The methodology.

Performance notes

Precompute. A page that runs a full comparison per request will be slow and will hammer your sources. Compute on a schedule, store the result, serve the stored version with its timestamp visible.

That also makes freshness a first-class property of the page rather than an afterthought, which is the right way round.

Next

Paper trading a signal before it costs anything.

Educational content only. Nothing here is financial advice, a solicitation, or an offer to trade. Prediction markets can move against you and a position can lose its entire value. Where this guide describes a difference between venues it describes a theoretical cross-market edge in observed prices, which is not a profit: fees, spreads, capital lockup, transfer time and differing settlement rules all come out of it first.