// learn · Data & tools

Prompt Lab

Pick your parameters. A structured, copy-ready research prompt assembles live. Run it on Perplexity, Claude, GPT or Gemini. Honest by default: every pick must come with its risk and what would invalidate it.

Your prompt 

        

          
Paste it into an AI research tool

          
    
            
  • Perplexity live web + citations
    • 
            
  • Claude deep reasoning, long briefs
    • 
            
  • GPT strong general reasoning
    • 
            
  • Gemini Google data, long context
    • 
            
  • Grok real-time market chatter (X)
    • 
            
  • DeepSeek capable and low-cost
    • 
            
  • Kimi long context, strong on finance
    • 
            
  • Venice AI private, uncensored
    • 
          

          
Not sure which model is best for finance? Check independent benchmarks: Artificial Analysis (finance & accounting) and FinanceBenchmark.ai. Rankings shift often, so re-check before you trust a model on numbers.

          
We only suggest tools. Pick the one you trust, and always verify the output against the live signals on Peaky Radar.

        

      

    


    

      
How to build an AI research prompt for stocks or crypto

      
A good research prompt does four things: it fixes the universe, fixes the horizon, names the evidence, and demands an exit. Skip any of them and the model answers with the same famous tickers it gives everyone else.

      
    
        
  1. Set the universe. Market, region and market cap. "Find good stocks" returns mega-caps; "European-listed, $1B–$50B" returns a screen.
    2. 
        
  2. Set horizon and upside. Days to twelve months. The horizon decides whether catalysts or fundamentals dominate the answer.
    3. 
        
  3. Name the evidence. Which research inputs and which indicators must be cross-referenced, so the reply cites signals rather than narrative.
    4. 
        
  4. Demand deliverables and invalidation. Entry, stop, targets, and what would prove the thesis wrong. A pick with no invalidation level is an opinion, not research.
    5. 
      


      
Prompt Lab FAQ

      
What is an AI research prompt for investing?

      
A structured instruction that tells the model exactly which universe to scan, over which horizon, using which evidence, and in which format to answer. That structure is what separates a citable research brief from a generic list of popular tickers.

      
Which AI model is best for stock and crypto research?

      
Use a model with live web access for anything time-sensitive: Perplexity and Grok read the web in real time, Claude and GPT reason better over long briefs, Gemini handles long context. Running the same prompt on two models is a cheap consistency check.

      
Can I generate the prompt in a language other than English?

      
Yes. The whole prompt, not just the interface, is generated in English, Italian, Spanish, French, German or Chinese. Switching language rewrites the prompt text, so you can paste it and get an answer in the same language.

      
Which regions can I screen?

      
U.S.-listed, U.S. and Canada (TSX), Europe (Euronext, LSE, Xetra, Borsa Italiana, SIX), Asia-Pacific (Tokyo, Hong Kong, Shanghai/Shenzhen, Seoul, Taiwan, Sydney) or global. The region goes into the prompt as named exchanges, which stops the model from silently defaulting to U.S. mega-caps.

      
Is the output financial advice?

      
No. This is an educational tool. AI output can be wrong or out of date, so verify every figure against a primary source before acting. Markets carry risk of loss.

    


    
More ready-made prompts

    

      

        
Long-term investment plan

        
Act as a fee-only financial planner. Given my age, monthly savings capacity, target retirement age, and risk tolerance (ask me for these if not provided), build a long-term investment plan: a target asset allocation (equities/bonds/cash/alternatives) with a rationale, index funds or ETF categories to use for each slice, a rebalancing schedule, and how the allocation should shift as I get closer to retirement. State every assumption explicitly and flag anything that depends on my tax jurisdiction. Not financial advice, for my own research.

        
      

      

        
Crypto discovery scan

        
Using real-time data, find 10 crypto projects (market cap under $500M) with rising on-chain activity, growing developer commits, and real product usage, not just social hype. For each, give: what the project actually does, why current traction looks organic rather than paid, the main risk (team, unlock schedule, competition), and what would invalidate the thesis. Rank by risk-adjusted conviction, not by hype. Not financial advice, for my own research.

        
      

    


    
Educational tool, not financial advice. AI output can be wrong or out of date, verify every figure before acting. Markets carry risk of loss.