multidude

"I divide my officers into four classes; the clever, the lazy, the industrious, and the stupid. Each officer possesses at least two of these qualities. Those who are clever and industrious are fitted for the highest staff appointments. Use can be made of those who are stupid and lazy. The man who is clever and lazy however is for the very highest command; he has the temperament and nerves to deal with all situations. But whoever is stupid and industrious is a menace and must be removed immediately!” ― Kurt von Hammerstein-Equord

At first i was thrilled at the incredible speed of my new coding partner. Chatting with the agent felt like talknig to a real person, which blinded me a bit and made me forget many rational rules of software development. A few refactorings and many learnings later i find resonance in HN where people critizize which at the beginning i was so enthusiastic about.

I have this incredibly industrious partner, blazingly fast but mindless and oblivious. Am i the guy Hammerstein told us to avoid, or is it my Agent(or worse, is it the combination of us both)?

Imagine an industrious and stupid one with an AI Agent.

tried to do financial sentiment analysis but kept hitting the same wall with FinBERT: "OPEC cuts production" scores negative. For crude oil it should be bullish — less supply means higher prices. Generic models don't know what phrases mean for a specific asset.

No existing dataset captured this systematically. Every financial sentiment project I found was either abandoned or producing generic polarity that was actively wrong for commodities and FX.

So I built swik — at first for myself, then I kept finding zero alternatives and thought this could interest other people. Many sentiment analysis projects out there are abandoned for this very reason: failure to produce meaningful, correct and reliable sentiment for a given asset-specific headline.

Then I thought: "build it and they will come." Now there it is — an open, community-maintained catalog of phrase inversions for 35+ assets (commodities, FX, indices, crypto). Each entry has the phrase, naive polarity, actual direction for that asset, and the economic reasoning behind it.

The inference API takes a headline + security symbol and returns direction, magnitude, relevance, and confidence — with the inversion catalog injected as context.

What's open:

• 267 inversion entries across 35 assets (CC BY 4.0, on GitHub) • Free API (100 req/day, no credit card) • Label queue to contribute corrections

Still very early — the catalog grows with contributions. Curious whether others have hit this problem and how they've handled it.

swik.io | catalog: github.com/polibert/sentimentwiki-catalog