SparklingAI
What Is SparklingAI? An AI Trading Agent Research Stack
A high-level explanation of the SparklingAI stack: market data, alpha research, execution policy, hard risk, evaluation, and the future AI agent/API layer.
Read research noteResearch notes
AI trading agent research, starting with XAUUSD
Evidence-driven notes on walk-forward testing, backtesting, execution layers, risk controls, and the development path toward SparklingAI as a broader market intelligence system.
XAUUSD historical data
How SparklingAI Builds a Research Data Pipeline for XAUUSD
A public-safe look at the XAUUSD research data pipeline behind SparklingAI, from local gold archives to mixed-feed diagnostics and model-ready datasets.
Read research notexLSTM time series forecasting
Why SparklingAI Is Exploring xLSTM for Alpha Research Instead of LLMs
A research note on why xLSTM-style sequence models may fit alpha research better than LLMs, while LLMs remain useful for agent reasoning and live-trade assistance.
Read research notemulti horizon forecasting
How SparklingAI Evaluates Multi-Horizon XAUUSD Alpha Signals
SparklingAI studies XAUUSD alpha across multiple horizons instead of relying on one short-term prediction. This note explains selected public diagnostics without exposing the private model recipe.
Read research notebacktest no trades
Why a Zero-Trade Backtest Can Still Be Useful in AI Trading Research
A zero-trade backtest is not automatically useless. In AI trading research, it can show that signal diagnostics and trade permission are being kept separate.
Read research notetake profit stop loss backtesting
What Target-Hit Diagnostics Reveal About AI Trading Signals
Target-hit diagnostics help explain whether a selected AI trading signal reaches its intended path before the stop path. That is more useful than looking at win rate alone.
Read research noteAI trading signals
How SparklingAI Validates XAUUSD Alpha Signals Before Execution
SparklingAI treats alpha as evidence, not as an automatic trade command. This note explains the validation checks between an XAUUSD AI signal and possible execution.
Read research noteAI trading signal API
What an AI Trading Signal API Should Return Beyond Buy or Sell
A useful AI trading signal API should return structured evidence, not just buy or sell. SparklingAI is designing signals around context, confidence, uncertainty, and decision status.
Read research notewalk forward testing
Walk-Forward Testing for AI Trading Strategies
A practical explanation of walk-forward testing, fold planning, and why out-of-sample validation matters for AI trading research.
Read research noteXAUUSD backtesting
XAUUSD Backtesting Case Study: 4-Fold Walk-Forward Rerun
A public XAUUSD walk-forward research case study showing the latest 4-fold rerun without exposing proprietary alpha logic.
Read research noteAI trading agent
What a Live AI Trading Agent Must Monitor Before Placing Orders
A public-safe engineering note on the monitoring layer a live AI trading agent needs before any signal can become an order.
Read research note