Store Pinnacle odds in Postgres

Answer: keep one append-only table of price ticks keyed by market id, outcome and timestamp, and write every market upsert from the PinnFeed WebSocket into it. Snapshots give you the starting state; delta frames give each change. Add /v1/closing rows later for CLV. Works with any Postgres (self-hosted, Supabase, Neon).

Schema#

create table if not exists odds_ticks (
  ts          timestamptz not null default now(),
  event_id    bigint      not null,
  market_id   text        not null,
  market      text        not null,   -- moneyline | spread | total | team_total
  period      int,
  line        numeric,
  outcome     text        not null,
  decimal     numeric     not null,
  fair        numeric,
  max_stake   numeric,
  primary key (market_id, outcome, ts)
);
create index if not exists odds_ticks_event on odds_ticks (event_id, ts);

Node worker (Builder)#

// npm i ws pg   ·   PINNFEED_KEY=ps_live_... DATABASE_URL=postgres://... node worker.js
import WebSocket from 'ws';
import pg from 'pg';

const db = new pg.Pool({ connectionString: process.env.DATABASE_URL });
const sql = `insert into odds_ticks (event_id, market_id, market, period, line, outcome, decimal, fair, max_stake)
             values ($1,$2,$3,$4,$5,$6,$7,$8,$9) on conflict do nothing`;

async function save(m) {
  for (const o of m.outcomes || []) {
    await db.query(sql, [m.eventId, m.id, m.type, m.period, m.line ?? null, o.name, o.decimal, o.fair ?? null, m.maxStake ?? null]);
  }
}

function connect() {
  const ws = new WebSocket('wss://pinnfeed.io/v1/stream', { headers: { 'x-api-key': process.env.PINNFEED_KEY } });
  ws.on('open', () => ws.send(JSON.stringify({ type: 'subscribe', sports: ['soccer'], markets: ['moneyline', 'total'], main: true })));
  ws.on('message', async (raw) => {
    const msg = JSON.parse(raw);
    if (msg.type === 'ping') return ws.send(JSON.stringify({ type: 'pong' }));
    if (msg.type === 'snapshot') for (const m of msg.markets || []) await save(m);
    if (msg.type === 'delta') for (const ch of msg.changes) if (ch.kind === 'market' && ch.op === 'upsert') await save(ch.market);
  });
  ws.on('close', (code) => { console.log('closed', code); setTimeout(connect, 2000); });
}
connect();

If inserts fall behind the stream, the server can close with 4008 slow_consumer; batch inserts (or queue them) and subscribe to fewer markets. Reconnect logic in full: Node reconnect.

Query examples#

-- last price per outcome
select distinct on (market_id, outcome) market_id, outcome, decimal, ts
from odds_ticks order by market_id, outcome, ts desc;

Use the table for backtesting against closing lines and line movement charts.

Plans#

Streaming needs Builder (€99/mo) or Pro; closing-line history is included on Builder. On Free or Hobby you can poll /v1/odds into the same table. Paid plans: 3-day trial; a card is required and it is charged after day 3 unless you cancel.

Educational content for adults (18+). Data for research, not betting advice.

Questions people ask#

Should I store the snapshot?#

Yes, as the first row per outcome. Otherwise the table only holds prices that changed after you connected.

How do I avoid duplicates after a reconnect?#

The snapshot after a reconnect repeats current prices; on conflict do nothing with a timestamp key keeps them as new ticks. If you want only changes, compare with the last stored price before inserting.

TimescaleDB?#

Works the same; make odds_ticks a hypertable on ts.