Prediction markets still contain significant inefficiencies. Top traders or “whales” and highly specialized players in specific categories — weather, particular sports leagues, niche political and cultural events — demonstrate consistent positive mathematical expectancy across large samples of trades.

Our system, built on principles proven effective in real-world automated analysis cases (including analysis of over 86 million trades and identification of elite wallets among 14,000+ active addresses), can:

• Automatically identify wallets and strategies with sustainable edge (minimum 100 trades, >70% win rate, positive expectancy, high consistency)
• Detect player specialization by category and timing effects
• Track exit discipline (successful players exit before settlement in 91% of cases)
• Automatically replicate the behavior of elite strategies with adaptation to current market conditions

For every potential trade, a multi-layered filtering and decision-making system operates:

Automatic selection of markets based on liquidity and order book depth, spread size, time to resolution, presence of confirming activity from pre-selected elite wallets, and correlation with other positions in the portfolio.

The AI system evaluates each potential trade across dozens of parameters: historical success of similar patterns, current market regime, alignment with selected strategy specialization, expected probability movement, and risk/reward profile.

Instead of a single algorithm, we deploy several independent AI agents, each voting for or against entry:

• Whale-Copying Agent — Tracks entries and behavior of selected leaders and assesses conviction level based on position size, timing, and historical success.
• Market Edge / Arbitrage Agent — Searches for arbitrage opportunities, discrepancies between market probability and external data or correlated markets, and other forms of statistical advantage.
• Risk & Portfolio Agent — Evaluates current portfolio exposure, correlation with existing positions, compliance with overall risk limits, and current market regime.

Entry into a position occurs only when a sufficient level of consensus is reached (for example, at least 2 out of 3 agents agree). This significantly reduces false signals and emotional or impulsive entries.

A key element of our statistical advantage is not only entry but timely exit. Unlike most retail investors, successful players on prediction markets exit with profit before settlement in 91% of cases.

Our system automatically decides to exit when one or more of the following conditions are met:

• Target level is reached (e.g., 85% of the calculated probability/price movement according to the internal AI model)
• Abnormal volume or activity appears, signaling the exit of “smart money” (whales)
• Internal model conviction falls below the threshold
• Time limits or portfolio triggers are reached (maximum time in position, changes in correlations, etc.)

This approach allows capturing the majority of potential profit while significantly reducing the risk of an unfavorable outcome in the final hours or days before resolution.

We deliberately sacrifice some potential upside to preserve capital, reduce result volatility, and ensure long-term strategy sustainability. Risk management is not an add-on — it is the heart and foundation of everything we do.

• Position Sizing — The size of each position is calculated dynamically based on the current AI model confidence score, overall portfolio risk budget, and established limits. We use conservative modifications of fractional Kelly and other proven position-sizing methodologies.
• Diversification — No more than 20–50 high-conviction setups are active simultaneously, selected from 500+ pre-filtered markets. Strict limits on maximum exposure to a single category, single “whale,” and correlated groups of positions.
• Correlation & Concentration Checks — The AI continuously monitors relationships between active positions and prevents excessive risk concentration in one direction or category.
• Drawdown Controls & Circuit Breakers — Automatic reduction of new position sizes or full suspension of new trade openings when certain portfolio drawdown levels are reached (daily, weekly, maximum).
• Stress Testing & Scenario Analysis — Regular testing of strategies on historical stress periods and modeled scenarios (sharp changes in probabilities, sudden liquidity drops, correlated movements across multiple markets).
• Human Oversight — Despite the high level of automation, a team of professional risk managers and quants provides constant monitoring of system performance and can adjust parameters in exceptional market situations.

Prime Dex uses a transparent and aligned model where platform and investor interests are fully aligned:

• Performance Fee — Standard 20% of profits above the established hurdle rate (minimum return level below which no performance fee is charged).
• Management Fee — Small annual fee for asset management (typically 1–2% AUM).
• Net Profit Distribution — All profits after deducting fees and operating expenses are distributed to investors proportionally to their share of assets under management (AUM).
• Dividends — Payouts are made regularly (monthly or quarterly) as dividends to the specified wallet or account.
• Auto-Compounding — Option for automatic dividend reinvestment to maximize the power of compounding and accelerate capital growth.

Thanks to the combination of powerful statistical edge, extremely strict filtering (less than 0.5% of trades pass selection), and institutional-grade risk management, Prime Dex aims to deliver stable positive returns with controlled volatility and limited drawdowns — something traditionally available only to professional market participants and hedge funds.