{
  "id": 14188051,
  "title": "ASIA_X Multi-Horizon RL Research",
  "url": "https://urgent.news/2026/10/13/asia-x-multi-horizon-rl-research",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-13T11:38:32.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/veduisback/asiax-multi-horizon-rl-research-3l8p"
  },
  "original_language": "en",
  "account": "ASIA_X Multi-Horizon RL Research research project details:\n\nThe project analyzes ASIA_X (Asia Composite Index) quote data using multi-horizon neural networks to estimate the directional outcome of different future time horizons from historical price/quote sequences.\n\nIt trains a neural network to evaluate WAIT, BUY, or SELL actions across six prediction horizons: 5s, 15s, 30s, 45s, 60s, and 120s. The model considers relationships between input state and historical outcomes, rather than using manually written trading rules.\n\nThe current multi-head neural network architecture includes 128-unit layers, 64-unit layer, and three actions (WAIT, BUY, SELL) per horizon, resulting in 18 total outputs. The discount factor is set to 0.95 and learning rate to 3e-4.\n\nAlthough using the name DQN, the implementation is more accurately described as direct offline reward prediction, learning to predict expected historical reward for each action/horizon combination.\n\nRaw quote data is provided in a specific format, divided into training segments (0-20) and unseen evaluation segments (21-25) for separate testing. The state representation consists of a 22-dimensional vector constructed from recent quote history and market information. Rewards for each action are determined by comparing future prices to entry price, with a winning reward of +0.85 and losing reward of -1.0.\n\nTraining uses segments 0-20 and evaluates on separate segments 21-25 to avoid overly optimistic performance estimates. Repository includes Python scripts for training and evaluation, model checkpoint, and requirements.",
  "summary": "ASIA_X Multi-Horizon RL Research repo: https://github.com/Veduisback/olymptrade-rl AI-based research project for analyzing ASIA_X (Asia Composite Index) quote data using a multi-horizon neural network. The project explores whether a learned model can estimate the directional outcome of different future time horizons from historical price/quote sequences. Research / educational project. This…",
  "key_points": [
    "ASIAX index analyzed using multi-horizon neural networks",
    "Trains model to decide WAIT, BUY, SELL across six time horizons",
    "Uses raw quote data with specific format for training and testing"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}