Have you tried handing it over some test money on a trading platform to see the results? What prediction algorithm is it using?

>>Have you tried handing it over some test money on a trading platform to see the results?

Not everybody has test money to play with. I'm in that group due to the crap economy left behind by previous politicians.

>> What prediction algorithm is it using?

As money is an issue, all I have at the moment is Google's AI Studio which is definitely not opensource, in my view. Though it is more forgiving than other vibe tools out there with their usage quotas. The market predictor has no capabilities to connect to any proprietary APIs as that requires money. That being said, it should be possible to modify the code so that it uses proprietary APIs should anyone want to do it. For now, I'm just adding stuff to it via VibeCoding that I think could be helpful. The main goal is to turn it into a research App that finds potential assets that will double to triple in the near feature or stocks that are prime for shorting, etc. I am also thinking of adding investment simulation capabilities but not quite sure how to go about at the moment. Turning some of its capabilities as a learning platform could be a good way to monetize the app. Not really a serious app at this moment, though, as I lack lots of tools to accomplish this.

Based on the code provided in services/geminiService.ts, here is the breakdown of the algorithms and sources used by the market prediction AI application:

Prediction Algorithm: The application does not use traditional quantitative statistical models (like ARIMA, LSTM, or Linear Regression) running on raw numerical data. Instead, it uses a Generative AI / Large Language Model (LLM) approach:

Model: The app utilizes Google's Gemini 2.5 Flash (gemini-2.5-flash).

Methodology: Context Gathering: The AI first performs a real-time Google Search (tools: [{ googleSearch: {} }]) to gather the latest text-based data, news, sentiment, and technical analysis summaries available on the web.

Semantic Analysis: The AI acts as a "Senior Financial Analyst" to interpret this unstructured data. It synthesizes a sentiment score (0-100) and predicts future price targets based on the qualitative data found (news catalysts, earnings, etc.).

Chart Generation: --The historical and forecast charts are generated mathematically within the code using Deterministic Linear Interpolation. --It connects the current price to the AI's predicted future price targets (1 week, 1 month, etc.).

--It adds algorithmic "noise" (randomness seeded by the asset name and date) to simulate market volatility visually, ensuring the chart looks realistic but stable for the specific day.

Data Sources: --The application relies entirely on Google Search Grounding. It does not connect to specific hardcoded financial APIs (like Bloomberg or Yahoo Finance API directly). Instead, it instructs the AI to search the public web for specific types of information.

Based on the prompts defined in the code, here are the sources and data points the app targets:

Real-Time Aggregators & Search Engine Results: --Live Google Search results for real-time price estimates. --Global financial news outlets (indexed by Google). --Market sentiment analysis from web summaries.

Regulatory & Official Documents: --SEC EDGAR Database: Specifically targeted in prompts for Stocks, Quantum, and Cannabis strategies to find filings.

Company Press Releases: --Used to identify fresh catalysts like contracts or product launches.

Technical Data: --Support and Resistance levels (retrieved from technical analysis articles found via search). --Relative Volume (RVOL) and Intraday Volatility data (for Day Trading strategy). --Chart patterns (Bull Flags, Pennants, Opening Range Breakouts).

Sector-Specific Sources: --Crypto Launchpads: Seedify, DAO Maker, Polkastarter (specifically for the "AI Presale" strategy).

Venture Capital Reports: --Data on VC investments and insider buying.

Industry News: Specific searches for Quantum Computing breakthroughs, Cannabis legalization news, and AI technology updates.

Market Dynamics: --Short interest data and borrow fees (for the "Short" strategy). --Analyst upgrades/downgrades and price targets. Macro-economic trend reports.

As I mentioned earlier, it actually found lots of legit assets with breakout potential but also finds crypto asset for that I suspect are scams but have a presence on the web.

Hope that provides a bit of context to your query.