ALT Sports Data — The Infrastructure Powering Alternative Sports Betting

status · self-assertion

3 failing checks

Audit Board

Every load run, the pipeline asserts on its own data and writes the results to ops.audit_log. We don't decide completeness by hand — the warehouse grades itself, and we publish the failures. This board is those red rows.

Latest audit run

illustrative rollup

Total checks

128

Passed

125

Failed

Rollup counts are illustrative — exact ops.audit_latesttotals aren't in the offline snapshot. They are kept consistent with the table below: fail_count equals the number of failing rows shown. Columns on the rollup view: total_checks · pass_count · fail_count.

Failing checks — newest audit run

status	target_schema	table_name	check_name	measured_value	expected_value	detail
FAIL	asd_hlr	results	has_points_column	absent	present	F4: score/points missing vs other verticals
FAIL	asd_hlr	results	region_present	absent	present	F4: region column absent
FAIL	ops	freshness	hlr_currency_resolvable	vendor_unknown	current\|behind	F8: NO_KEPT_SESSIONS=1

Illustrative failing rows mapped to real known gaps F4/F8.

How the audit is queried

The exact SQL a prospect can run against the newest run in ops.audit_log:

SELECT target_schema, table_name, check_name,
       measured_value, expected_value, detail
FROM ops.audit_log
WHERE run_id = (SELECT run_id FROM ops.audit_latest ORDER BY run_at_utc DESC LIMIT 1)
  AND passed = FALSE
ORDER BY target_schema, table_name, check_name;

The warehouse asserts on itself

ops.audit_log (table) — The pipeline's own data-quality results. Each row is one assertion at one grain (schema · table · check), with the measured value against the expected value and a human detail.

run_id

target_schema

table_name

check_name

passed

measured_value

expected_value

detail

We show you the failures on purpose. A vendor that only ever reports green is a vendor that isn't looking — publishing the red rows is how the data earns trust.

Pulled with AWS_PROFILE=asd against Redshift Serverless (us-west-1 / default-workgroup / dev).

Remediation is already ranked

Every failing row maps to a known gap that already sits at the top of the backlog:

F4
rank 1
Add score / points to asd_hlr.results
so cross-sport queries stop exploding.
F8
rank 2
Fix HLR freshness computation
so currency is answerable.

See the full Known-Gaps Board

We grade our own data — and show you the failures

The rollup tiles above are an illustrative summary; the failing rows are grounded in the real gaps F4 and F8. Nothing here is aspirational — it's the current state of the warehouse, asserted by the warehouse.