How to Build a Prediction Market Screener With an API

A screener narrows a few thousand live markets down to the handful worth your attention. It is the boring half of market tooling and the half people actually use daily, because the alternative is scrolling a venue's front page and calling it research.

The mechanics are simple. The judgement is in choosing filters that surface tradeable markets rather than merely interesting ones.

Screener or Scanner

Build the screener first. A scanner without a shortlist is scanning noise.

Pulling the Catalog

import Predictefy from '@predictefy/sdk';

const client = new Predictefy({ apiKey: process.env.PREDICTEFY_API_KEY });

// One call, every supported venue.
const markets = await client.router.fetchMarkets({ limit: 500 });

The catalog changes slowly. Fetch it on a schedule, cache it, and run your filters against the cache. Re-pulling thousands of markets on every user keystroke wastes credits and adds latency for no benefit.

Filters That Earn Their Place

Four filters do most of the work. Price is the least useful of them, despite being the one everybody starts with.

function screen(markets, rules) {
  const now = Date.now();

  return markets.filter((m) => {
    const price = m.lastPrice;
    if (price == null) return false;

    // Ignore markets that have effectively already resolved.
    if (price < rules.minPrice || price > rules.maxPrice) return false;

    const days = (new Date(m.closeTime) - now) / 86_400_000;
    if (days < rules.minDays || days > rules.maxDays) return false;

    if (rules.venues.length && !rules.venues.includes(m.venue)) return false;

    return true;
  });
}

The Filter That Needs a Second Call

Liquidity is the filter that decides whether a market is tradeable, and it is not in the catalog. You have to read the book, which is why the cheap filters run first.

const withLiquidity = [];

for (const m of shortlist) {
  const book = await client.exchange(m.venue).fetchOrderBook({
    marketId: m.marketId
  });

  const bid = book.bids[0]?.price;
  const ask = book.asks[0]?.price;
  if (bid == null || ask == null) continue;

  const spread = ask - bid;
  if (spread <= rules.maxSpread) {
    withLiquidity.push({ ...m, bid, ask, spread });
  }
}

Spread is a better liquidity proxy than volume for this purpose. Volume tells you what happened; spread tells you what it would cost to get in and out right now.

Going Past the Top of Book

A tight spread on one contract says nothing about a hundred. If your screener feeds real trades, ask what your intended size would actually cost.

const quote = await client.exchange(m.venue).getExecutionPrice({
  marketId: m.marketId,
  side: 'buy',
  contracts: rules.size
});

// Slippage against the top of book, in cents.
const slip = quote.averagePrice - m.ask;

Sorting and Presenting

Sort by whatever you are optimizing for, but sort by something. A screener that returns an unordered list has moved the problem rather than solved it. Slippage ascending is a good default: the markets you can actually get in and out of, first.

Show the fields that drove the decision, not every field you have. Venue, price, spread, days to close and slippage is enough to act on. Anything more becomes a spreadsheet nobody reads.

Why the Router Endpoint Matters

Written per venue, a cross-venue screener means one client per venue, one pagination scheme per venue, one field naming convention per venue, and a normalization layer you maintain forever. Most people give up and screen one venue, which defeats the point: the best price for a given view is often on a venue you were not looking at.

Predictefy's router searches 15+ venues in one call and returns one shape, so the filtering code above never learns which venue a market came from. The API key is free to start with a monthly credit allowance, which is enough to run a real screener on a schedule.

Frequently Asked Questions

Conclusion

A screener is a filter chain with one expensive link. Order it correctly, cheap filters against a cached catalog first and book reads only for the survivors, and it runs comfortably on a free key.

The filters worth having are the ones about tradeability rather than interest: how wide is the spread, how long until it resolves, and what would your actual size cost. A market that looks compelling and cannot be entered at size is not a result, it is a distraction.

One housekeeping note: this is information, not financial advice. Endpoints and credit costs change, so confirm anything that matters against the current documentation.