Research standards
How Synoptiv analyzes stocks
Synoptiv is produced by an automated research engine. It gathers public financial information, computes a deterministic baseline, and uses artificial intelligence to synthesize the evidence into a balanced, versioned report. No report is individually reviewed by a human analyst before publication.
Published · Author: Synoptiv Research Engine
1. Public information we gather
Each run combines up to 27 collection and computation steps across 10 categories of public data. We crawl the public web for company news and market sentiment; collect OHLCV price history and corporate actions; read publicly available SEC filings, forms, and XBRL facts through EDGAR; and gather company fundamentals, financial statements, earnings history, analyst consensus, institutional ownership, insider Form 4 activity, disclosed congressional activity, lobbying records, federal contracts, peer data, fund holdings, market-attention measures, and macroeconomic series.
Coverage varies by security and date. Missing inputs stay missing; the engine does not invent a value to fill a gap. Data can still be delayed, incomplete, duplicated, or wrong.
2. Deterministic calculations
Before any AI-written analysis, code computes technical indicators and a rule-based baseline from the collected data. These calculations include trend and momentum measures such as moving averages, RSI and MACD; volatility measures such as ATR; price levels such as support and resistance; valuation and growth ratios; earnings-surprise history; and a normalized composite sentiment score on a −100 to +100 scale.
Composite sentiment fuses available news tone, technical posture, analyst recommendations, insider and congressional activity, short-interest context, and attention signals. It is a descriptive summary of the inputs, not a recommendation to buy or sell.
3. AI-written synthesis
A large language model receives the structured evidence and deterministic calculations, then writes the company overview, news takeaways, bull and bear cases, scenario assessment, risks, invalidation conditions, and final thesis. The model is instructed to distinguish facts from interpretation, acknowledge conflicting evidence, and avoid unsupported precision.
AI output can misunderstand data or state something incorrectly. Readers should verify material claims against primary public records before making a decision.
4. Signals and “no trade”
When the evidence supports a defined setup, a report may include a directional signal with an entry, stop, targets, and conditions that would invalidate the thesis. When evidence is weak, contradictory, stale, or does not offer a defensible risk/reward setup, the engine returns “no trade.” That is an intentional outcome, not a failed analysis.
Signals are research outputs, not personalized recommendations. They do not account for a reader’s objectives, finances, time horizon, taxes, liquidity needs, or risk tolerance.
5. Versions, freshness, and track record
A report is a timestamped snapshot. Re-running a symbol creates a new immutable version rather than silently replacing the old one, so readers can see what changed and which information was available at the time. Public stock pages show the latest analysis date and link the version history; full versioned reports require access.
Any displayed signal outcome or track-record statistic is hypothetical: it is calculated by comparing a published signal with later market prices. It is not the result of an executed trade and does not include every real-world cost or constraint.
6. What this methodology does not promise
The process improves consistency and makes the evidence inspectable; it does not make forecasts certain. Markets change after publication, historical relationships can break, and both source data and generated text can contain errors. Synoptiv is informational research, not investment advice, and past or hypothetical performance does not guarantee future results.
Read the full disclosures for limitations, risk, data-quality, and AI-generation details.