Recipe catalog / financial-advice-signal

Detect specific financial advice

Does text give a specific recommendation to buy, sell, hold, or allocate money, rather than general education?

You need a yes/no check on whether an assistant's or user's text crosses from explaining financial concepts into recommending a specific action with money, so the response can be disclaimed, softened, or routed for review.

Explore this recipe interactively · Source and implementation guide

Use financial-advice-signal in TypeScript

Install with npm install jev-recipes. Requires Node.js 22.9 or newer and ES modules. Set TYPESAFE_API_KEY in your server environment for live calls, which send input to TypeSafe and use API quota. See the installation guide.

import { financialAdviceSignal } from 'jev-recipes/financial-advice-signal';

const result = await financialAdviceSignal({
  "text": "Honestly, at your age you should just put the whole $20k into VTI and forget about it. Bonds are a waste right now. Sell the individual tech stocks you mentioned and move that money over too.",
  "minConfidence": 0.8
});
console.log(result);

Input contract

{
  "input": {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "properties": {
      "text": {
        "type": "string"
      },
      "minConfidence": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      }
    },
    "required": [
      "text"
    ]
  },
  "result": {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "properties": {
      "model": {
        "type": "string"
      },
      "usage": {
        "type": "object",
        "properties": {
          "input_tokens": {
            "type": "integer",
            "minimum": 0,
            "maximum": 9007199254740991
          },
          "output_tokens": {
            "type": "integer",
            "minimum": 0,
            "maximum": 9007199254740991
          }
        },
        "required": [
          "input_tokens",
          "output_tokens"
        ],
        "additionalProperties": false
      },
      "status": {
        "type": "string",
        "enum": [
          "ready",
          "review"
        ]
      },
      "probability": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      },
      "confidence": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      },
      "verdict": {
        "type": "string",
        "enum": [
          "advice",
          "informational"
        ]
      }
    },
    "required": [
      "model",
      "usage",
      "status",
      "probability",
      "confidence",
      "verdict"
    ],
    "additionalProperties": false
  }
}

Saved example result

This hand-authored response demonstrates the contract. It is not a model accuracy measurement. Run it without an API key: npx jev-recipes demo financial-advice-signal.

{
  "model": "demo-fixture",
  "usage": {
    "input_tokens": 0,
    "output_tokens": 0
  },
  "status": "ready",
  "probability": 0.91,
  "confidence": 0.91,
  "verdict": "advice"
}

Evaluation evidence

Use the evaluation guide to measure this decision on your own labeled cases.

Limitations

• Detects wording that recommends a specific action, not whether the text legally constitutes financial advice in any jurisdiction or whether the speaker is licensed.
• Judges the text alone. A recommendation framed as hypothetical or quoted from someone else still counts if the text presents it as what the reader should do.
• Does not assess whether the advice is sound.

Related recipes

• response-refusal: Use response-refusal to detect whether a reply declined a request, rather than whether it contains a specific financial recommendation.
• certainty-match: Use certainty-match to check whether a reply's stated confidence fits its evidence, rather than whether it recommends a financial action.