Validation Layer Architecture

The ValidationLayer (monitoring/validation_layer.py) provides real-time data sanity checking on provider responses before metrics are exposed via the API.

Range Checks (FIELD_RANGES)

Every incoming numerical metric is validated against absolute domain boundaries:

Field Name	Minimum Allowed	Maximum Allowed	Unit / Description
pe_ratio	-500.0	2000.0	Price to Earnings ratio
dividend_yield	0.0	0.40	Dividend yield (0% to 40%)
beta	-5.0	10.0	Market Beta
roe	-5.0	5.0	Return on Equity (-500% to +500%)
roa	-5.0	5.0	Return on Assets
market_cap	1,000,000	10,000,000,000,000	Market Capitalization in local currency

Values falling outside these bounds are discarded and replaced with None.

Inter-Provider Consensus Check

When 2 or more providers supply a numerical value for the same field:

1. The median value across all agreeing providers is computed: M = median(V_1, V_2, ..., V_k).
2. For each provider value V_i, the absolute percentage deviation is calculated: deviation = |V_i - M| / M
3. If deviation > VALIDATION_OUTLIER_THRESHOLD (default 0.50 or 50%), V_i is flagged as an outlier, logged in provider_health_log, and set to None.