●——● TECHNOLOGY

Built in-house.
Operated continuously.

The infrastructure stack is proprietary from signal loop to order routing. No vendor dependencies at the execution layer. Every component is owned, tested, and understood.

●——● INFRASTRUCTURE STACK

Five systems. One machine.

Each system is independently designed and proprietary. Together they form a continuously operating trading infrastructure.

SIGNAL ENGINE
60-second OI and price signal loops via Upstox API. Real-time open interest data aggregated across 5-min, 15-min, and 1-hour timeframes. Claude AI inference layer for confluence scoring — each signal scored before routing to the execution agent.
EXECUTION AGENT
Autonomous trading agent with OAuth2 order routing. T1/T2/T3 exit ladder at +30% / +60% / +100% enforced programmatically. Daily loss circuit breaker halts all activity on trigger. No manual overrides permitted at risk boundaries.
OI INTELLIGENCE
Real-time open interest aggregation across multiple timeframes with WhatsApp alert delivery for critical threshold breaches. OI-based institutional flow detection feeds the primary signal generation layer.
KAPTRAX ENGINE
The Kaptrax Analog Engine performs cosine similarity matching and FastDTW dynamic time warping against 65,000+ validated historical Nifty windows. Pattern analog identification surfaces structural similarities across years of market data — validated statistically before live signal pool entry.
RISK CONTROLS
File-based kill switch enables instant system shutdown without API calls. Forced market-close exits at 15:10 IST — system-enforced, not discretionary. Daily PnL limits, behavioral pattern monitoring, and systematic journaling of all failure modes.
●——● INFRASTRUCTURE METRICS

The numbers behind the machine.

Research depth, pattern coverage, and operational precision — measured.

XAUUSD rows researched
5-min, 2004–2025
Pattern windows in
Kaptrax database
60s
Signal loop
interval
NSE F&O bhavcopy
rows indexed
●——● KAPTRAX

Pattern intelligence at scale.

Kaptrax is Salt Route's proprietary analog engine. It identifies structural market patterns by matching current price behavior against a database of 65,000+ validated historical windows from the Nifty 50.

The matching uses two techniques in combination: cosine similarity for shape comparison (fast, computationally light) and FastDTW (Fast Dynamic Time Warping) for temporal alignment that handles compression and expansion of patterns across different market velocity regimes.

No pattern enters the live signal pool without statistical validation against the full 1.86M-row NSE F&O bhavcopy dataset. Edge is measured before it is traded.

●——● AI INFERENCE

Claude AI as confluence scorer.

The signal engine routes real-time OI data, price action, and pattern analog results through a Claude AI inference layer. The model evaluates multi-factor confluence — not as a decision maker, but as a structured scoring layer that must clear before execution routing begins.

AI augmentation also powers the research environment: systematic journaling of trade hypotheses, behavioral pattern tagging, and performance attribution — each trade feeds the next research cycle.

Infrastructure as moat. The edge is not the signal alone — it is the integrated system that surfaces, filters, executes, monitors, and learns.

●——● EXPLORE FURTHER

Understand the philosophy that governs every system decision.

View Philosophy →