HOW TO READ SENTIVIO
Context you can inspect.
Understand what the metrics measure, and where their limits are.
Coverage, not market completeness
Counts describe items observed by Sentivio from Yahoo, Reddit, StockTwits and RSS publishers. They use publication time, not the time an item was processed. Delayed ingestion can change historical counts.
Company matching is automated. A ticker mention does not guarantee relevance, and duplicate stories can appear across sources.
Language balance, not expected return
The current classifier labels text as Bullish, Neutral or Bearish. Balance is 100 × (bullish count − bearish count) / classified items. Confidence is model confidence, not a probability of investment success.
Period comparisons require five classified items in each window. Source divergence compares the highest and lowest source balances when at least two sources have five items each.
Official evidence stays distinct
SEC EDGAR filings are linked to issuers using CIK identifiers. They are presented as official documents, without automated sentiment or a claim that every filing contains material news.
Initial SEC history covers a bounded recent period. A successful check with no filing is different from a failed check or missing history.
Portfolios are research watchlists
Sentivio tracks companies, not brokerage balances. Coverage shares measure information concentration. They do not measure financial allocation, diversification or investment risk.
Subscriber reports support a repeatable research process. Always inspect original documents and make decisions in the context of your own circumstances.