How it works
The feed shows less so you can trust it more.
Most stock trackers show you everything. Every tick, every mention, every 0.2% move. BestStocks does the opposite — it filters aggressively and only surfaces events that clear a real bar. Here's what that bar looks like.
Daily data collection
Every day we pull fresh prices, volume, and earnings data from Yahoo Finance for all 101 tracked companies. SEC filings come directly from EDGAR — no third-party summary, just the raw document metadata.
News runs through Marketaux, which scores how relevant each article is to each company by entity. We only accept articles where the tracked company ranks in the top two entities and scores above 0.5 relevance. That alone cuts out the majority of noise articles.
The significance gate
A stock has to clear at least one of these to appear in the feed:
- Price moved 2.5% or more from the previous close
- Volume was at least double the five-day average
- Price hit a new 30-day high or low
Anything below all three thresholds gets dropped. On a quiet day that means most of the 101-company list never shows up at all. That's intentional.
Evidence audit
Once a move clears the gate, we don't just report the percentage. We check nine categories of evidence within a ±3-day window around the event date:
- Price and volume patterns
- Recent news articles (scored for relevance)
- Earnings surprises or guidance changes
- SEC filings (8-K, 10-Q, 10-K, proxy statements)
- Analyst upgrades, downgrades, or price target changes
- Corporate actions (splits, buybacks, leadership changes)
- Peer-company moves in the same controlled sector group
- Broader market or macro factors
- Technical signals (new highs, breakouts)
We use Claude Haiku to investigate, not to invent. If a real catalyst shows up in the evidence, it gets cited. If nothing concrete turns up, the explanation says so — sector correlation, volume anomaly, or unknown.
Confidence levels
Every feed card gets a confidence level before it publishes:
- High — timely news or earnings data found that explains the move
- Medium — sector correlation, volume spike, or new high/low, but no single clear driver
- Low — move cleared the significance gate but no catalyst was found
Low-confidence cards still appear in the feed — sometimes a move is real even without a visible reason. But the label is there so you can decide how much weight to give it.
Peer-group context
We don't compare every stock to every other stock. We use controlled peer groups: semiconductors, mega-cap internet, cloud software, payments, pharma/biotech, energy, financials, and others. When AMD moves, we check NVDA, INTC, and QCOM — not AMZN.
If the whole peer group moved together, the explanation flags sector-wide pressure or tailwind. If only one company moved while peers sat flat, that gets flagged too. Context changes how you should read the event.