System Operation

    Kronos operates through a structured execution pipeline with strict risk controls

    Execution Pipeline

    1

    Market Data Scanning

    Real-time market data is collected via WebSocket from connected exchanges (Binance Futures, Bybit, OKX, KuCoin). Technical indicators are calculated across multiple timeframes to assess current market conditions.

    Data sources: Price data, volume, order book depth, and 20+ derived technical indicators (RSI, MACD, Bollinger Bands, Ichimoku, ADX, ATR, Volume Profile, etc.) across 1m, 5m, 15m, 1h, and 4h timeframes.

    2

    Rule-Based Trade Evaluation

    Each potential trade is evaluated against predefined rules and configured parameters. No trade is executed without meeting all configured criteria.

    Technical Criteria

    Signal strength, confluence, trend alignment

    Risk Validation

    Exposure limits, position sizing, SL/TP verification

    3

    Risk Parameter Validation

    Before any execution, the system validates that all configured risk parameters are satisfied. No trade bypasses these controls.

    Stop-Loss

    Mandatory on every trade

    Position Size

    Within exposure limits

    Isolated Margin

    Risk contained per position

    4

    Order Execution via Exchange API

    Trades are executed directly via the connected exchange API (Binance, Bybit, OKX or KuCoin) with mandatory stop-loss and take-profit orders placed simultaneously through the Algo Order endpoint.

    Automated Execution

    Orders sent directly to Binance with SL/TP attached

    Risk Controls

    Isolated margin, predefined exposure limits enforced

    5

    Trade Logging & Performance Reporting

    Every execution is logged with complete details. Performance metrics, exposure data, and execution history are available through reports and logs.

    Complete audit trail including entry/exit prices, P&L, execution timestamps, and risk metrics. For informational purposes only.

    Adaptive Intelligence Layer

    On top of the execution pipeline, three optional layers continuously adapt the bot to market and user performance.

    Adaptive Mode

    Auto-tunes minimum confidence, position size and leverage based on your real win rate, drawdown and consecutive losses. Overrides static settings when ON.

    Strategy Selector

    Classifies each pair as Trend / Range / Volatile / Breakout and assigns one of 10 master presets (3x → 50x). High-leverage presets are reserved for ultra-liquid pairs.

    Auto Profit Protection

    Dynamic trailing stop: SL is moved only in the direction of profit. TP stays open while the AI confirms the move still has strength, with a 15-min Anti-Revenge cooldown after a loss.

    System Limitations

    Understanding what the system cannot do is essential for realistic expectations

    The System Cannot:

    • ✗Predict future market movements with certainty
    • ✗Guarantee profits or prevent all losses
    • ✗Eliminate the inherent risks of trading
    • ✗Withdraw or transfer user funds
    • ✗Override configured risk parameters

    Potential Issues:

    • Execution delays during high volatility
    • Slippage in fast-moving markets
    • Exchange outages affecting operations
    • Market conditions differing from historical patterns
    • Past performance does not predict future results

    Foundation Models and System Architecture

    Kronos combines classical quantitative analysis with modern machine learning. Foundation models — large models pre-trained on broad market and language data — are used to read context: news tone, market regime, and the relationship between assets. Those signals never trade on their own. They are one input among several, and every decision still has to clear deterministic risk rules before an order is sent to an exchange.

    Real market data

    Prices, order books, funding rates and volume are read directly from connected exchange APIs in real time. No simulated or delayed feeds are used for live decisions.

    Foundation models and ML

    Language and time-series foundation models classify market regime (trend, range, high volatility) and summarize context. A learning loop compares past predictions with realized outcomes and adjusts confidence over time.

    Confluence scoring

    Technical indicators, model output and historical performance on that pair are merged into a single confidence score. A trade is only considered when several independent indicators agree.

    Deterministic execution

    Execution is rule-based, not model-based: position size, leverage, mandatory stop loss and take profit are set by your configuration and validated before the order reaches the exchange.

    How a decision travels through the system

    Every order follows the same path. Each layer can reject the trade, and a rejection at any point stops the process.

    • Ingestion — live market data is collected from each connected exchange.
    • Analysis — indicators are computed and foundation models classify the current market regime.
    • Scoring — signals are merged into one confidence value with a minimum threshold.
    • Risk gates — daily loss limits, cooldown, exposure and leverage rules are checked.
    • Execution and protection — the order is sent with a mandatory stop loss that only moves in the profit-locking direction.

    What the models do not do

    Foundation models do not predict prices and do not guarantee results. They estimate context and probability. Markets change, models degrade, and past performance does not indicate future results. Capital preservation rules always override any model output, and you can stop the system at any time.

    Risk Disclosure

    This platform is for informational and experimental purposes only. Automated trading involves risk, including loss of capital. No investment advice is provided. Past performance does not guarantee future results. Consider seeking independent financial advice before trading.

    Explore How Kronos Works

    Explore our security measures and risk disclosure