Here is the prompt snippet to switch modes in Grok3:

-----

Add MODE TOGGLE: Verbose (VBS) and Laconic (LAC) | [Toggle response mode; VBS for 350–500 words with context, LAC for 50–100 words, brief, no context
- VBS: “VBS ON” in prompt; full exposition, Markdown structure.
- LAC: “LAC ON” in prompt; concise, butler-style, minimal Markdown.
- Default: VBS; auto-revert next prompt unless re-specified.
- Log mode switches in JRN

suko 

In the case of Grok3 it's not a million tokens but in reality more like 128K tokens and that works out, for me, to 50000 words with 25 images per chat.

The images you share with Grok are charts? Reason Im asking is I tried this with GPT and felt that sometimes image comprehension is not perfect, after discussing with it, "we" settled it would be best to exchange trading related data in some sort of csv export. I coded it, but haven't used it yet, but can share experience in the future.

I am not using it for technical analysis but rather analytical results my platform spews out in image format that I have no CSV access to. Don't need CSV now, can just screenshot the table of numbers and let the AI sort it out and then analyze.

Grok itself says GPT4 has an advantage in that area as Grok cannot analyze images directly, but even this is wrong because it's famous for its medical image diagnostic use such as MRIs, and I have been using it for identification in the garden.

I was hoping to experiment LLM for analysis / prediction (recommendation rather ). After spending around $25 on tokens with opeai using gpt-4 analyzing stocks from SP500, Nasdaq 100 and Dow, results look ok. I am using Technical, Fundamental and market sentiment to predict. I will host it on AWS and provide a link to check the results. Here is the prompt I am using if any one is curious.

# Construct prompt with all available data
prompt = f"""Analyze the following stock data for {symbol} and provide a comprehensive price prediction:

Current Market Data:
- Last Price: ${current_price:.2f}
- Bid: ${market_data.get('bid', 'N/A')}
- Ask: ${market_data.get('ask', 'N/A')}
- VWAP: ${daily_stats.get('vwap', 'N/A')}
- Daily Range: ${daily_stats.get('low', 'N/A')} - ${daily_stats.get('high', 'N/A')}
- Price Change: {price_change:.2f}%

Key Metrics:
- Market Cap: {market_cap}
- EPS: {eps}
- Dividend Yield: {dividend_yield}%
- P/E Ratio: {pe_ratio}
- P/B Ratio: {pb_ratio}
- Debt to Equity: {debt_to_equity}
- Current Ratio: {current_ratio}
- Quick Ratio: {quick_ratio}
- Profit Margin: {profit_margin}%
- Revenue Growth: {revenue_growth}%
- Earnings Growth: {earnings_growth}%
- Free Cash Flow: {free_cash_flow}
- Operating Margin: {operating_margin}%
- Return on Equity: {return_on_equity}%
- Return on Assets: {return_on_assets}%
- Beta: {beta}
- Shares Outstanding: {shares_outstanding}
- Average Volume: {avg_volume}
- Sector: {sector}
- Industry: {industry}
- Volume Change (vs 30-day avg): {f"{volume_change:.2f}%" if volume_change is not None else "N/A"}

Technical Indicators:
- RSI: {technical_indicators.get('rsi', 'N/A')}
- MACD: {technical_indicators.get('macd', 'N/A')}
- Signal Line: {technical_indicators.get('signal_line', 'N/A')}
- 20-day SMA: {technical_indicators.get('sma_20', 'N/A')}
- 50-day SMA: {technical_indicators.get('sma_50', 'N/A')}
- Bollinger Bands:
* Upper: {technical_indicators.get('bollinger_bands', {}).get('upper', 'N/A')}
* Middle: {technical_indicators.get('bollinger_bands', {}).get('middle', 'N/A')}
* Lower: {technical_indicators.get('bollinger_bands', {}).get('lower', 'N/A')}
- Volume Ratio: {technical_indicators.get('volume', {}).get('ratio', 'N/A')}

Fundamental Data:
{json.dumps(fundamental_data, indent=2)}

News Sentiment:
- Sentiment: {news_sentiment.get('sentiment', 'neutral')}
- Score: {news_sentiment.get('score', 0.0)}

Based on this data, provide a comprehensive analysis and prediction. Your response must be a valid JSON object with the following structure:
{{
"direction": "up/down/neutral",
"confidence": <number between 0 and 1>,
"reason": "<detailed explanation>",
"analysis": {{
"technical": "<technical analysis summary>",
"fundamental": "<fundamental analysis summary>",
"market_sentiment": "<market sentiment analysis>",
"risk_factors": ["<list of key risk factors>"],
"opportunities": ["<list of key opportunities>"]
}},
"metrics": {{
"price": {current_price},
"volume_change": {volume_change},
"market_cap": "{market_cap}",
"eps": {eps},
"dividend_yield": {dividend_yield},
"pe_ratio": {pe_ratio},
"pb_ratio": {pb_ratio},
"debt_to_equity": {debt_to_equity},
"current_ratio": {current_ratio},
"quick_ratio": {quick_ratio},
"profit_margin": {profit_margin},
"revenue_growth": {revenue_growth},
"earnings_growth": {earnings_growth},
"free_cash_flow": {free_cash_flow},
"operating_margin": {operating_margin},
"return_on_equity": {return_on_equity},
"return_on_assets": {return_on_assets},
"beta": {beta},
"shares_outstanding": {shares_outstanding},
"avg_volume": {avg_volume},
"sector": "{sector}",
"industry": "{industry}",
"vwap": {daily_stats.get('vwap', 'N/A')},
"daily_range": {{
"high": {daily_stats.get('high', 'N/A')},
"low": {daily_stats.get('low', 'N/A')}
}}
}}
}}"""

messages = [
{"role": "system", "content": "You are a professional stock market analyst. Provide clear, data-driven predictions with confidence scores and comprehensive analysis. Always respond with a valid JSON object."},
{"role": "user", "content": prompt}
]

Fundamental and Market Data : polygon.ai and yahoo finance ( fill some gaps ).

Here is a temporary link to hosted results to try. This is just to verify the results and showcase the above prompt.
Top UP stocks and Top Down stocks list 50 stocks that have the highest probability / confidence to go up/down n the near future. This is not backtested, so no value as of now. Just for demo. Stock Market Analyzer

Great thread! I've been trying to get a indicator for 2nd Entries (Mack's PATS) coded for Tradestation. The Ninja Trader indi is perfect so the logic is there. I finally secured the Ninja script and have been working with Gemini (flash & pro) to convert to Easy Language. Not doing too well, syntax issues Gemini can't figure out. So I'm curious if anyone has a suggestion for AI models that may better understand EL or coding in general, thanks

Tripken 

Great thread! I've been trying to get a indicator for 2nd Entries (Mack's PATS) coded for Tradestation. The Ninja Trader indi is perfect so the logic is there. I finally secured the Ninja script and have been working with Gemini (flash & pro) to convert to Easy Language. Not doing too well, syntax issues Gemini can't figure out. So I'm curious if anyone has a suggestion for AI models that may better understand EL or coding in general, thanks

Claude.AI should be able to do it in a few minutes. Free for quick usage.

shortski 

Claude.AI should be able to do it in a few minutes. Free for quick usage.

Thanks. Haven't tried Claude, none of the other mainstream models could get it. Of course only using free, may need to pay for a month!!

Tripken 

Thanks. Haven't tried Claude, none of the other mainstream models could get it. Of course only using free, may need to pay for a month!!

if you just copy/paste the code to Claude, if it is working code, the bot should convert it quickly. Maybe some errors but if it is working code, it should be sorted quickly.

I would also suggest trying Claude. I'm using the $20/month "pro" version. Here's my experience.

I asked Claude to write from scratch an intraday EMA indicator in TradingView that reset on the first bar of each day, to remove the effect of opening gaps. That involved some careful timebase handling and EMA scaling. I'm experienced with TV code, and Claude's code was perfect.

I haven't yet tested Claude writing or converting to EasyLanguage but I plan to use Claude to convert some of my TV indicators to EL.

If you give Claude a try, post your results here. I'm curious to know how well Claude handled EL.