Context-aware financial sentiment analysis.
Built on an open inversion catalog.

Standard NLP models get financial headlines wrong. “OPEC cuts” is not bearish for oil — it is bullish. swik applies Aspect-Based Sentiment Analysis (ABSA) with an asset-specific inversion catalog to return signals that reflect how markets actually react. Open data. Free API.

35 assets
ยท
345 inversions
ยท
371 human labels
ยท

Try it now

Why swik?

The Inversion Catalog

267+ known phrase inversions across 35+ assets. Context-dependent polarity corrections for commodities, FX, indices, and crypto. CC BY 4.0. Browse inversions →

Asset-Specific Sentiment APIs

Each security has its own price driver catalog. Direction, magnitude, and relevance calibrated per asset. Built for quantitative research and applications in trading.

Fine-tuning on Real Labels

Community labels feed LoRA fine-tuning per asset. Every contribution improves cross-asset spillover analysis and domain-specific model accuracy. Contribute →

Community-driven Accuracy

Every label sharpens the model. Top contributors earn API credits and appear on the public leaderboard.

See full leaderboard →