Files
signal-platform/scripts/fix_finnhub_market_cap.sql
T
dennisthiessen 5a531fd603
Deploy / lint (push) Successful in 9s
Deploy / test (push) Successful in 1m10s
Deploy / deploy (push) Successful in 38s
fix: convert Finnhub market cap from millions to absolute USD
Finnhub profile2 reports marketCapitalization in millions; storing it
as dollars made mega-caps like SPCX show as micro (e.g. 1.8M). Normalize
on ingest, add unit tests, and include a one-shot SQL backfill script.
2026-07-18 10:28:16 +02:00

114 lines
4.3 KiB
PL/PgSQL

-- Fix Finnhub market caps stored in millions instead of absolute USD.
--
-- Context:
-- Finnhub profile2.marketCapitalization is in millions of USD.
-- Pre-fix rows stored that value as-is (e.g. SPCX ~1.8e6 → "1.8M micro").
-- Correct absolute USD is value * 1_000_000 (e.g. ~1.8e12 → "1.8T mega").
--
-- Identification:
-- Chained fundamentals write source provenance into unavailable_fields_json:
-- "source_market_cap": "finnhub"
-- FMP / Alpha Vantage rows are already absolute USD — leave them alone.
--
-- Safety:
-- After a successful update we stamp market_cap_unit = "usd" so this script
-- is idempotent (safe to re-run).
--
-- Database: PostgreSQL (production).
-- Run against your app DB, e.g.:
-- psql "$DATABASE_URL" -f scripts/fix_finnhub_market_cap.sql
-- or step through the sections in a SQL client.
-- ---------------------------------------------------------------------------
-- 1) Preview: who would be fixed
-- ---------------------------------------------------------------------------
SELECT
t.symbol,
f.market_cap AS stored_millions_as_dollars,
f.market_cap * 1000000.0 AS corrected_usd,
CASE
WHEN f.market_cap * 1000000.0 >= 200e9 THEN 'mega'
WHEN f.market_cap * 1000000.0 >= 10e9 THEN 'large'
WHEN f.market_cap * 1000000.0 >= 2e9 THEN 'mid'
WHEN f.market_cap * 1000000.0 >= 300e6 THEN 'small'
ELSE 'micro'
END AS corrected_band,
f.fetched_at,
f.unavailable_fields_json
FROM fundamental_data f
JOIN tickers t ON t.id = f.ticker_id
WHERE f.market_cap IS NOT NULL
AND f.market_cap > 0
AND f.unavailable_fields_json::jsonb->>'source_market_cap' = 'finnhub'
AND COALESCE(f.unavailable_fields_json::jsonb->>'market_cap_unit', '') <> 'usd'
ORDER BY f.market_cap DESC;
-- Count
SELECT COUNT(*) AS rows_to_fix
FROM fundamental_data f
WHERE f.market_cap IS NOT NULL
AND f.market_cap > 0
AND f.unavailable_fields_json::jsonb->>'source_market_cap' = 'finnhub'
AND COALESCE(f.unavailable_fields_json::jsonb->>'market_cap_unit', '') <> 'usd';
-- ---------------------------------------------------------------------------
-- 2) Apply fix (run inside a transaction; COMMIT only if preview looks right)
-- ---------------------------------------------------------------------------
BEGIN;
UPDATE fundamental_data f
SET
market_cap = f.market_cap * 1000000.0,
unavailable_fields_json = (
f.unavailable_fields_json::jsonb || '{"market_cap_unit": "usd"}'::jsonb
)::text
WHERE f.market_cap IS NOT NULL
AND f.market_cap > 0
AND f.unavailable_fields_json::jsonb->>'source_market_cap' = 'finnhub'
AND COALESCE(f.unavailable_fields_json::jsonb->>'market_cap_unit', '') <> 'usd';
-- Sanity check after update (should include SPCX ~1e12 scale if present)
SELECT
t.symbol,
f.market_cap,
f.unavailable_fields_json
FROM fundamental_data f
JOIN tickers t ON t.id = f.ticker_id
WHERE f.unavailable_fields_json::jsonb->>'source_market_cap' = 'finnhub'
AND f.unavailable_fields_json::jsonb->>'market_cap_unit' = 'usd'
ORDER BY f.market_cap DESC NULLS LAST
LIMIT 30;
-- Remaining unfixed Finnhub rows (should be 0)
SELECT COUNT(*) AS remaining_unfixed
FROM fundamental_data f
WHERE f.market_cap IS NOT NULL
AND f.market_cap > 0
AND f.unavailable_fields_json::jsonb->>'source_market_cap' = 'finnhub'
AND COALESCE(f.unavailable_fields_json::jsonb->>'market_cap_unit', '') <> 'usd';
-- COMMIT; -- uncomment when satisfied
-- ROLLBACK; -- or roll back if something looks off
-- ---------------------------------------------------------------------------
-- Optional: rows with market_cap but NO source_market_cap provenance.
-- These cannot be safely auto-fixed (could be FMP absolute USD already).
-- Inspect manually if anything looks like Finnhub-scale millions.
-- ---------------------------------------------------------------------------
-- SELECT
-- t.symbol,
-- f.market_cap,
-- f.unavailable_fields_json,
-- f.fetched_at
-- FROM fundamental_data f
-- JOIN tickers t ON t.id = f.ticker_id
-- WHERE f.market_cap IS NOT NULL
-- AND f.market_cap > 0
-- AND (
-- f.unavailable_fields_json IS NULL
-- OR f.unavailable_fields_json = '{}'
-- OR f.unavailable_fields_json::jsonb->>'source_market_cap' IS NULL
-- )
-- ORDER BY f.market_cap ASC
-- LIMIT 50;