Guides / Cookbooks

Cookbooks · Text · Beginner · about 5 min

Is a headline relevant to a symbol?

For a headline and a company, ask whether it is about the company, which way it cuts, and how much it could matter, before it reaches a model or a person.

1. When to use it

   News feeds are mostly noise for any one name. Score each headline before it reaches a desk channel or a downstream model. The headline below is a sample written for this recipe, not a real news item.

2. The request

   {
     "model": "ora-fin1",
     "state": "NVIDIA (NVDA) shares slip after a large cloud customer says it will design more of its own AI accelerators instead of buying them.",
     "questions": {
       "about_nvda": {
         "type": "yes_no",
         "instructions": "Is this headline about NVIDIA or its products?"
       },
       "tone": {
         "type": "choice",
         "instructions": "For NVIDIA shareholders, this news is",
         "criteria": {
           "positive": "Good: more demand, wins, upgrades",
           "negative": "Bad: lost demand, a customer leaving, downgrades",
           "neutral": "Neither, or unclear"
         }
       },
       "materiality": {
         "type": "score",
         "instructions": "How much could this matter to NVIDIA's revenue?",
         "criteria": [
           "Not at all",
           "A little",
           "Noticeably",
           "A lot"
         ]
       }
     }
   }

   Open this in the Playground

3. The real response

   {
     "model": "ora-fin1",
     "answers": {
       "about_nvda": {
         "type": "yes_no",
         "yes_no": 0.92
       },
       "tone": {
         "type": "choice",
         "choice": "negative",
         "confidence": 0.693,
         "probabilities": {
           "positive": 0.143,
           "negative": 0.795,
           "neutral": 0.062
         }
       },
       "materiality": {
         "type": "score",
         "score": 1.046,
         "confidence": 0.954,
         "legend": [
           "Not at all",
           "A little",
           "Noticeably",
           "A lot"
         ],
         "probabilities": [
           0,
           0.958,
           0.038,
           0.004
         ]
       }
     },
     "usage": {
       "input_tokens": 440,
       "output_tokens": 19
     },
     "version": "ora-text-0.2.0"
   }

   About the company: 0.92. Tone: negative (confidence 0.693). Materiality: 1.046 on 0–3, mostly “A little”.

4. The code

   cURLPythonJavaScriptPHPGoJavaC#RubyCopy

   Requires curl 7.76+. Save as run.sh, then: bash run.sh

   curl -sS https://oramodel.com/v1/systemone \
     -H "Authorization: Bearer $JEV_API_KEY" \
     -H "Content-Type: application/json" \
     -d '{
     "model": "ora-fin1",
     "state": "NVIDIA (NVDA) shares slip after a large cloud customer says it will design more of its own AI accelerators instead of buying them.",
     "questions": {
       "about_nvda": {
         "type": "yes_no",
         "instructions": "Is this headline about NVIDIA or its products?"
       },
       "tone": {
         "type": "choice",
         "instructions": "For NVIDIA shareholders, this news is",
         "criteria": {
           "positive": "Good: more demand, wins, upgrades",
           "negative": "Bad: lost demand, a customer leaving, downgrades",
           "neutral": "Neither, or unclear"
         }
       },
       "materiality": {
         "type": "score",
         "instructions": "How much could this matter to NVIDIA'\''s revenue?",
         "criteria": [
           "Not at all",
           "A little",
           "Noticeably",
           "A lot"
         ]
       }
     }
   }'

5. Act on it

   a = data["answers"]
   if a["about_nvda"]["yes_no"] >= 0.7 and a["materiality"]["score"] >= 1.5:
       post_to_desk(headline, tone=a["tone"]["choice"])
   elif a["about_nvda"]["yes_no"] >= 0.7:
       archive(headline, tag="nvda")

   Text answers describe the headline, not the price. To ask about the stock itself, send the symbol (Decide on a stock with just its symbol).

6. Variations

   • Name the company in the question, not only in the text: answers are sharper.
   • Ask several companies' questions about one headline in one request.

Information, not investment advice. Every number on this page is a real response for the session shown; answers change with each session.

What's next

• Back to the market recipes