HOW IT WORKS
Tailored Trading Approach
The groundbreaking ATG system features a global architecture built upon four vital components. By analyzing securities’ historical behavior spanning decades, ATG tailors unique strategies for each security to maximize trading returns.
As a self-sustaining system, ATG eliminates the need for human intervention in decision-making. The ATG master brain autonomously manages all customization, from parameter-setting to final investment execution. For instance, Energy futures and Equity index futures receive distinct trading approaches, as they exhibit differing volatility levels and behavioral patterns.
Hierarchical
Reinforcement
Centralized
Learning
Resilient Reward
System
Exploratory
Data Analysis
Behind The Scenes of ATG
Walkforward Optimization
Allows optimizations to be “walkedforward,” such that each new optimization is applied to a fresh out-of-sample actual trading session. This enables ATG to assess the out-of-sample performance of a strategy that is reoptimized on a regular basis using newer data.
Pyramiding/Scaling
Adds the option to gradually increase or decrease a position. The maximum number of entries and percentage change for each increase or decrease can be determined by ATG self learning algorithms.
Multiple Frequencies
on the same security
Allows ATG to combine multiple time frame data, indicators, predictions, and trading techniques on the same security.
Position Sizing
Determines how many shares/contracts/units to buy with each trade. Allows the master brain to determine the appropriate position sizing parameters and/or approach.
Multiple Template Optimization
& Backtesting
Enables the optimization and backtesting of multiple templates at the same time, utilizing the same data, cost, position sizing, position scaling, and optimization parameters.
How is ATG better than
A Human Portfolio Manager
Portfolio Manager
The market’s behaviour is the product of massive and complicated interactions between millions of market players, each of whom brings their own set of heuristics and biases to their decision-making processes. The resulting scenario causes human biases in the portfolio manager’s decision-making process, such as “ownership bias,” “risk tolerance asymmetry,” “anchoring,” and many more, which have a negative impact on the portfolio’s total returns.
A typical portfolio manager thinks of a formula to beat the markets and then automates it using traditional systematic or quantitative algorithm. These “static” algorithms can perform well for sometime but eventually fail as the market’s regime moves in one direction or another.
A human being has limited working hours per day and may overlook possible opportunities that arise dynamically in the derivatives market, resulting in ineffective and inefficient decisions.
Trading Software
Following the systematic process of deep reinforcement learning, ATG’s decisions are robust to human bias and identify imprints of human biases in overall market behaviour, resulting in positive portfolio returns.
ATG, like an enhanced human brain, understands shifting market behaviour, compares current volatility to scenarios from decades before, and modifies its “dynamic” algorithms as needed to optimise its risk return profile automatically.
Working tirelessly, 365 days a year, ATG constantly and automatically learns new data points, develops new features, and reacts to changing market regimes, resulting in the best decisions for the client’s portfolio.