What data sources does ManiBot use?

ManiBot uses: FinViz Elite (stock screening and market overview), Finnhub API (real-time quotes and fundamentals), ORTEX (short interest and institutional data), Yahoo Finance (historical OHLCV fallback), Alpha Vantage (additional historical data fallback), and TradingView (chart widgets).

Is my data secure on ManiBot?

Yes. User API keys are encrypted at rest using AES-256-GCM encryption. ManiBot uses Google OAuth 2.0 and TOTP-based two-factor authentication. No real financial account or brokerage data is collected — ManiBot is a paper trading platform only.

AI Safety & Methodology | Mani$Bot — How Our AI Works, Data Sources & Limitations

Q: How does ManiBot's AI select stocks?

ManiBot uses a multi-stage AI pipeline. Stage 1: FinViz Elite screens 3,830+ NASDAQ stocks using technical filters (price floor $5+, volume, price action). Stage 2: AI models (GPT-5, Gemini 2.5 Flash, or Perplexity Sonar Pro) score the resulting candidates against named trading strategies. Stage 3: Risk filters apply ATR-based position sizing, confidence thresholds (≥80%), and cross-source deduplication before any paper trade executes.

Q: Can AI analysis be wrong?

Yes. AI models can hallucinate facts, misinterpret data, present stale information, and be overconfident. A high confidence score from an AI model does not guarantee accuracy. ManiBot uses multiple AI providers and cross-checks signals, but no AI system is infallible. Always verify analysis with primary sources before making any decisions.

Q: What are the limitations of backtested strategies?

Backtested strategy results are simulated using historical data. They do not account for: real-world slippage, bid-ask spreads, order execution delays, liquidity constraints, or psychological factors like fear and greed. Past backtested performance does not predict future results. Results in paper trading may also differ significantly from live trading.

gavel What ManiBot Is NOT

We Are Not Financial Advisors

ManiBot is NOT a registered investment adviser, broker-dealer, or financial adviser. We are not registered with the SEC, FINRA, or any state or federal regulatory body. Nothing on this platform constitutes personalized investment advice.

Not personalized investment advice — We provide educational content and AI-generated analysis, not recommendations tailored to your financial situation, goals, or risk tolerance.
Not a guarantee of profits — Past performance of any strategy — simulated or real — does not guarantee future results. All trading involves substantial risk of loss.
Not a substitute for professional advice — Always consult a licensed financial adviser before making real investment decisions.
Not real trading — Paper trading uses virtual money. Simulated results can differ significantly from live trading due to slippage, market impact, and emotional factors.
Not a brokerage — ManiBot does not execute real trades, hold real funds, or connect to any brokerage account.

psychology How ManiBot's AI Works

ManiBot uses a multi-stage AI pipeline that combines traditional stock screening with multiple large language models (LLMs). Here is exactly how it works:

1

Technical Screen

FinViz Elite filters 3,830+ NASDAQ stocks by price ($5+), volume, and technical criteria

2

AI Strategy Scoring

LLMs score candidates against named strategies. Confidence threshold ≥80% required

3

Risk Filters

ATR-based position sizing, cross-source deduplication, market regime check (VIX + SPY)

4

Paper Trade

Surviving picks execute as simulated paper trades with AI-predicted targets and stops

AI Routing Logic

ManiBot uses Intelligent Perplexity Routing — queries that require live web data (news, earnings, recent filings) are routed to Perplexity Sonar Pro, which has real-time web access. All other queries go to the configured primary model (default: GPT-5 or Gemini 2.5 Flash). Users can override routing by selecting a specific provider or using their own BYO API key.

The 4 Core AI Providers

OpenAI GPT-5 — General-purpose analysis, strategy matching, narrative summaries. Best for balanced, comprehensive stock breakdowns.
Google Gemini 2.5 Flash — Fast multi-modal analysis. Excellent for high-volume screening and technical pattern recognition.
Perplexity Sonar Pro — Live web access. Best for real-time news, recent earnings, FDA approvals, and breaking events. Automatically routed when live data is needed.
DeepSeek Chat — Cost-efficient alternative for users who need many analyses and want to minimize API costs when using BYO keys.

BYO API Keys

Users can bring their own API keys for any supported provider. Keys are encrypted at rest using AES-256-GCM. This gives you full control over which model is used, your own rate limits, and your own cost. ManiBot never charges for AI usage beyond the subscription fee.

How Strategies Are Applied

ManiBot's library includes 446 named trading strategies across 41 types (momentum, breakout, mean reversion, value, reversal, volatility, and more). The AI is given a strategy definition — including entry criteria, exit criteria, and risk parameters — and asked to score a given stock's match quality. The strategy library is maintained in the database and updated by the platform team. Strategies are never generated on-the-fly by AI; they are pre-defined and reviewed by humans before activation.

storage Data Sources

ManiBot relies on several external data providers for market data, screening, and analysis. Here is a full list of what we use and what each provides:

Provider	Used For	Data Type	Update Frequency
FinViz Elite	Stock screening, market overview, sector performance	Real-time	Intraday (market hours)
Finnhub API	Real-time quotes, company fundamentals, earnings calendar	Real-timeFundamentals	Real-time / daily
ORTEX	Short interest data, days-to-cover, institutional flows	Short Data	Daily (end of day)
Yahoo Finance	Historical OHLCV data (fallback for backtesting)	Historical	Daily (end of day)
Alpha Vantage	Additional historical data (secondary fallback)	Historical	Daily
TradingView	Advanced chart widgets (displayed in-app)	Charts	Real-time
Perplexity Sonar Pro	Live web search for news, filings, recent events (via AI routing)	Live Web	On demand

Data Quality Caveat

Market data from third-party providers can contain errors, delays, or gaps — especially during pre-market, after-hours, or periods of high market volatility. FinViz Elite has been observed to occasionally ignore price filters under rate-limiting conditions. ManiBot applies a JavaScript-level price floor filter as a secondary safeguard, but no data pipeline is immune to errors. Always verify prices with your own brokerage or exchange data before acting on any information.

report_problem AI Model Limitations

Understanding the limits of AI is as important as understanding its capabilities. Here is an honest breakdown of known risks:

Limitation	Description	Risk Level
Hallucination	AI models can state incorrect facts confidently — including wrong prices, earnings dates, or company details. Always verify numbers with primary sources.	High
Training data cutoff	GPT-5, Gemini, and DeepSeek have knowledge cutoffs. They may not know about recent management changes, product launches, or regulatory actions. Perplexity mitigates this with live web access.	High
Overconfidence	A high confidence score (e.g. 92%) from an AI model does not correlate to prediction accuracy. Confidence scores reflect internal model certainty, not real-world outcome probability.	High
Pattern extrapolation failure	AI trained on historical patterns can fail during unprecedented market conditions (black swan events, pandemics, rate shocks). Past market regimes do not repeat identically.	Medium
No real-time news awareness	Non-Perplexity models do not watch live news. A breaking FDA rejection or earnings miss can invalidate AI analysis instantly. Use Perplexity routing for time-sensitive stocks.	Medium
Prompt sensitivity	Slightly different phrasing of the same question can produce different AI responses. ManiBot uses standardized, tested prompts in the database-driven prompt management system to minimize this.	Low–Medium
Market regime blindness	ManiBot fetches live VIX and SPY data to detect market regime (bullish/bearish/volatile), but AI models are still largely trained on historical regimes and may underweight extreme conditions.	Low

Best Practice: AI as One Signal Among Many

Use ManiBot's AI analysis as one data point among several. Cross-reference with SEC filings, official company IR pages, reputable financial news outlets, and your own technical analysis before making any real investment decisions.

history Backtesting & Simulated Performance Caveats

ManiBot displays backtested strategy performance and paper trading P&L results. These are simulations only. Be aware of the following:

Survivorship bias — Backtests typically only include stocks that survived to the test date. Companies that went bankrupt or were delisted during the test period may be excluded, making results look better than they actually were.
Look-ahead bias — Strategies may inadvertently use data that would not have been available at the time of the trade decision. ManiBot tests strategies to minimize this, but it cannot be completely eliminated.
Slippage not modeled — Real orders fill at different prices than the quoted price. In illiquid stocks, slippage can be substantial. Paper trades execute at the last known price without modeling spread or market impact.
No commissions or fees — Simulated paper trades do not account for brokerage commissions, exchange fees, or margin interest that would apply in real trading.
Emotional factors excluded — Paper trading does not simulate the psychological pressure of real money on the line. Discipline in paper trading does not guarantee equivalent discipline with real capital.
Market impact not modeled — Paper trades do not move the market. Large real trades in small-cap or low-volume stocks can move the price against you, increasing your effective cost.

Past Simulated Performance Does Not Predict Future Results

Any strategy performance metrics shown in ManiBot are based on historical simulations. They are not a projection or guarantee of future returns. Markets change, conditions change, and any strategy can fail.

warning Trading Risk Disclosure

Trading Involves Substantial Risk of Loss

You can lose some or all of your investment in real trading. Only risk money you can afford to lose completely. ManiBot's paper trading system uses virtual funds — no real money is involved.

Market volatility — Stock prices can move rapidly and unpredictably. Losses can exceed initial expectations.
Leverage amplifies losses — If you use margin or options in real trading, losses can exceed your initial investment.
Small-cap and penny stock risk — ManiBot enforces a $5 price floor on all scanner picks to reduce penny stock exposure, but stocks above $5 can still be highly volatile and illiquid.
Short interest risk — Stocks with high short interest can experience sudden short squeezes. ORTEX data helps identify this but does not predict timing.
Paper vs. real performance — Results achieved in paper trading may differ substantially from real trading results.

account_balance Regulatory Posture

ManiBot is operated as an educational technology platform. We hold no regulatory registrations at this time and operate within the following framework:

Not a registered investment adviser (RIA) under the Investment Advisers Act of 1940
Not a broker-dealer registered with FINRA or the SEC
Not a commodity trading adviser (CTA) registered with the CFTC or NFA
Not a money services business (MSB) — ManiBot does not hold, transfer, or invest real funds
General educational content — All content is designed to be educational and informational, not advice targeted to any individual's financial situation

Your Responsibility

By using ManiBot, you acknowledge that you understand these limitations, that you are responsible for your own investment decisions, and that you will consult appropriate licensed professionals before placing real trades. If you are unsure whether trading is right for you, please speak to a qualified financial adviser.

health_and_safety Risk Management Guidelines

Before transitioning from paper trading to real trading, consider these principles that experienced traders use:

Position Sizing

Never risk more than 1–2% of your total portfolio on a single trade
Diversify across multiple stocks and sectors to reduce concentration risk
Use stop-losses on every trade to cap potential losses before entering
Scale position size with conviction — highest conviction, largest (within limits) position

Emotional Discipline

Create a written trading plan and follow it without improvisation
Never chase losses by increasing position sizes (revenge trading)
Take breaks after significant losses to avoid emotional decision-making
Keep a trading journal and review it weekly to identify patterns

Education Before Real Money

Paper trade for at least 3–6 months and achieve consistent positive results before using real money
Understand every strategy you use thoroughly — if you can't explain it, don't trade it
Start with a very small real-money account if transitioning — treat early losses as tuition
Track real vs. paper results honestly to identify where your psychology diverges from your plan

privacy_tip Your Data & Security

API keys are encrypted — If you bring your own AI provider keys, they are stored with AES-256-GCM encryption. ManiBot staff cannot read your API keys.
No real financial data collected — ManiBot does not connect to brokerage accounts, banks, or any real financial institution. No real account numbers, balances, or transaction data are ever collected.
Authentication — ManiBot uses Google OAuth 2.0 and TOTP-based Two-Factor Authentication (2FA) to protect accounts. Passwords are not stored in plaintext.
Your trading activity is private — Paper trades, analysis history, and preferences are not sold to third parties.
Data deletion — You can request deletion of your account and associated data at any time through the Support Center.

update Safety & Quality Update Log

ManiBot continuously improves its AI safety guardrails and data quality. Here is a log of recent changes:

May 2026 Expanded safety methodology page. Added AI routing transparency, data source table, model limitations table, and regulatory posture section.
May 2026 Added JavaScript-level $5 price floor post-filter in FinViz screener as hard backstop against penny stocks slipping through when FinViz is rate-limited.
Mar 2026 Daily picks auto-trade confidence threshold raised to 80% (from 75%). Strategy pipeline minimum price raised to $5. Game Theory scanner R:R fixed to 2:1 (target 12%, risk 6%).
Mar 2026 Real RSI(14) using Wilder's smoothing from historical price data replaced synthetic estimates. Live VIX and SPY market regime detection added with 10-minute cache.
Mar 2026 ATR-based risk position sizing added to trade execution to reduce losses on volatile stocks. Cross-source ticker deduplication added to prevent over-concentration.
Mar 2026 Replaced hardcoded penny-stock fallback strategy set with diversified momentum fallback [Trend_Following_Improved, Mean_Reversion_Improved, Volume_Surge].
Jan 2026 AI Prompt Management System deployed. All scanner and analysis prompts are now database-driven, version-tracked, and tested before activation. Eliminates ad-hoc prompt drift.
2025 Initial launch of paper trading platform, safety disclaimers, and AI analysis with multi-provider fallback logic.

quiz Frequently Asked Questions

Is ManiBot financial advice?

No. ManiBot is not a registered investment adviser, broker-dealer, or financial adviser. All AI analysis, scanner signals, and strategy matches are for educational and informational purposes only. Always consult a licensed financial adviser before making real investment decisions.

How does ManiBot's AI select stocks?

ManiBot uses a multi-stage pipeline: FinViz Elite screens 3,830+ NASDAQ stocks using technical filters (price ≥$5, volume criteria), then LLMs score candidates against named strategies with a minimum confidence threshold of 80%, and finally ATR-based risk filters and cross-source deduplication are applied before any paper trade executes.

Can AI analysis be wrong?

Yes — and often. AI models can hallucinate facts, misinterpret data, present stale information, and be overconfident. A high confidence score does not guarantee accuracy. Always verify any AI-generated analysis with primary sources (SEC filings, company IR pages, reputable financial news) before acting on it.

What are the limitations of backtested strategies?

Backtested results are simulated using historical data. They do not account for real-world slippage, bid-ask spreads, order execution delays, liquidity constraints, commissions, or psychological factors. Past backtested performance does not predict future results.

Is my data secure?

Yes. User API keys are encrypted at rest using AES-256-GCM. ManiBot uses Google OAuth 2.0 and TOTP 2FA for authentication. No real financial account or brokerage data is collected — ManiBot is a paper trading platform only. See our Privacy Policy for full details.

What regulators govern ManiBot?

ManiBot is an educational technology platform. It is not registered with the SEC, FINRA, CFTC, or NFA. It does not hold or transfer real funds, execute real trades, or provide personalized investment advice. If you need regulated financial advice, please consult a licensed professional in your jurisdiction.

contact_support Questions or Concerns?

If you have questions about our platform, safety measures, or need help, please visit our Support Center or Help Center.

For legal terms and conditions, see our Terms of Service and Privacy Policy.

ManiBot Safety & AI Methodology

Education First, Always