App builder & platform

You’re building a product on prediction-market data. Could be a market-discovery app, a wallet tracker, a copy-trading signal product, an analytics terminal, or a media property with live odds. This guide is about the architecture decisions — Predexon handles the indexing, normalization, and streaming so you don’t have to. You’ll be ready to:

Split reads correctly between REST and WebSocket
Serve market pages, wallet pages, and leaderboards from the right endpoints
Keep your quota bill near zero using free historical endpoints
Handle errors, caching, and latency expectations realistically

The core architecture

Almost every data product ends up with the same shape:

REST for state, WebSocket for change. Load a market page from REST once, then keep it live with the orderbook and trades channels. Don’t poll.
Cache reads you’ll repeat. Market metadata changes rarely — cache it hours. Prices and books change constantly — stream them instead of re-fetching.
Historical endpoints are free and unlimited on every plan (trades, orderbooks, wallet activity). Backfill as much as you want; it never touches your quota.

Serving the classic pages

Page in your product	Endpoints behind it
Market list / discovery	List markets sorted by volume, search, categories & tags
Market detail	Market price, candlesticks, orderbooks, top holders
Live tape	WebSocket trades — plus pending trades if you want your tape up to 5s ahead of everyone else’s
Wallet page	Wallet profile, positions, P&L
Leaderboard	Global leaderboard, per-market leaderboard
Smart-money feed	Smart activity, smart money per market

Code skeleton for a live market page:

import os, requests
HEADERS = {"x-api-key": os.environ["PREDEXON_API_KEY"]}
BASE = "https://api.predexon.com"

# Initial state — one REST call each
market = requests.get(
    f"{BASE}/v2/polymarket/markets", headers=HEADERS,
    params={"condition_ids": condition_id},
).json()
candles = requests.get(
    f"{BASE}/v2/polymarket/candlesticks/{condition_id}",
    headers=HEADERS, params={"interval": "1h"},
).json()

# Then keep it live over WebSocket (trades + orderbook channels)
# wss.predexon.com — see /websocket/overview

Differentiate with the hard datasets

The pages above are table stakes. The datasets that make a product stand out:

Pending trades (mempool) — show fills before they confirm. Your users see flow up to 5 seconds before apps built on Polymarket’s RTDS. WebSocket pending-trades →
Tick-level orderbook history — power “book at any moment” scrubbers, spread-history charts, and execution-quality analytics from the raw tick stream. Tick downloads →
Smart-money analytics — profitable-wallet flow per market, computed from full chain history. Nobody can fork your UI and get this from the venue API. Smart money →

Latency, errors, and operational realism

What to expect in production:

Surface	Typical latency	Notes
Data API (read)	50–200ms	Cacheable reads; use conditional fetching where possible.
WebSocket events	`<1s` end-to-end	Pending-trades lead confirmed trades — and beat RTDS by up to 5s.
Tick downloads	Seconds–minutes	Signed URLs; size depends on the slice you quote.

Error patterns you’ll see most:

429 — you outran your plan’s rate limit. Back off exponentially; consider caching or a higher tier.
401 / 403 — wrong or missing x-api-key.
503 rarely — surface as “temporarily unavailable” and let users retry; don’t hot-loop yourself.

See Best Practices for retry/backoff patterns, caching, and connection pooling.

Common builder recipes

Copy-trade signals

Stream a wallet’s fills in real time and surface them to your users.

Portfolio monitor

Positions + P&L + live updates for any wallet. Same patterns power most prediction-market UIs.

Smart-money entries

Alert when profitable wallets pile into a market.

Mempool signal feed

The 5-seconds-early tape, productized.

What you should read next

Data & Signals overview

Full map of every endpoint, organized by utility.

WebSocket overview

Channels, subscriptions, and plan limits for live data.

Best Practices

Retries, caching, rate limits, error handling.

Rate limits & free endpoints

What’s free and unlimited vs. what counts toward quota.

Welcome

Guides for your use case

Cookbook

Core concepts

Account & access

Integrations

The core architecture

Serving the classic pages

Differentiate with the hard datasets

Latency, errors, and operational realism

Common builder recipes

Copy-trade signals

Portfolio monitor

Smart-money entries

Mempool signal feed

What you should read next

Data & Signals overview

WebSocket overview

Best Practices

Rate limits & free endpoints

​The core architecture

​Serving the classic pages

​Differentiate with the hard datasets

​Latency, errors, and operational realism

​Common builder recipes

Copy-trade signals

Portfolio monitor

Smart-money entries

Mempool signal feed

​What you should read next

Data & Signals overview

WebSocket overview

Best Practices

Rate limits & free endpoints

The core architecture

Serving the classic pages

Differentiate with the hard datasets

Latency, errors, and operational realism

Common builder recipes

What you should read next