AINDEX Methodology
Systematic Trading Intelligence powered by transparent, multi-layer AI
01 Context & Innovation
The project analyzes how the inclusion or exclusion of specific securities from an index is becoming a decisive factor in modern investment strategies, driven by the evolution of digital finance and artificial intelligence (AI).
Traditionally, investors replicated indices through ETFs. With Direct Indexing, however, they can build personalized indices by directly purchasing fractional shares of individual stocks β excluding or including securities based on ethical, strategic, or risk-based criteria.
This AI-enhanced approach enables the creation of actively managed thematic indices, which are immediately investable and adaptable to market trends.
02 AI System Architecture
The entire model is based on three classes of AI applications, coordinated by a Supervisor:
Unsupervised Machine Learning
Identifies patterns, key variables, and risk profiles through clustering and dimensionality reduction techniques.
Machine Learning (Tactical)
Generates daily buy ("Long") or inactivity ("Out") signals for each asset, aiming to maximize the risk/return ratio while minimizing trading frequency.
Swarm Intelligence & Deep Learning (Strategic)
Evaluates the optimal index composition and necessary daily adjustments based on the signals produced by the other modules.
A.I. is employed to estimate, for each financial instrument analyzed, the probability of upward or downward movement for each of the 20 market days following the most recent trading value.
03 Risk Management
Risk is classified into three categories:
Acceptable
Assets with manageable risk profiles
Unacceptable
Illiquid assets, short selling (excluded)
Unavoidable
Systemic market risk
The investable universe includes only instruments with acceptable risk profiles, explicitly excluding illiquid assets.
04 Cyclicality & Operational Signals
Through the analysis of over 50,000 instruments, the system identified 12 statistically significant zones within each market cycle, represented as a clock:
Peak
Trough
Sell Zone
Ascending
Core AI Trading Rules:
- Buy at 12h or 6h
- Sell at 3h or upon reaching the stop-loss
- Avoid short selling (classified as an unacceptable risk)
Zones between 6 and 3 (via 12) are statistically favorable.
05 Daily Indicators
Each day, the system calculates for every instrument:
It also generates an automated English commentary for each instrument, ensuring consistent, unbiased insights.
06 Trade Quality Rank
The Trade Quality Rank (TQR) is a statistical setup-reliability model for active BUY/LONG opportunities. It combines signal, market cycle, risk profile, momentum, statistical quality, and data confidence into one score from 0 to 100. The proprietary weighting model is not disclosed.
TQR is a point-in-time comparison tool: a higher tier identifies a more robust and higher-priority setup relative to other eligible opportunities at the same market snapshot. It is not a next-day return forecast, a guaranteed outcome, or a substitute for the system signal and stop discipline.
Strong relative setup
Good relative setup
Selective setup
Lower entry priority
How to read the market view
Color always represents the TQR tier. Selection decides which active setups enter the map; tile area represents either current provider-normalized Map Weight or equal area, never TQR quality. Percentile and distance from the market median describe relative position inside the selected synchronized cohort.
How evidence is measured
TQR is frozen at observation time T and subsequent returns begin at T+1. Outcomes are evaluated over the selected horizon and aggregated by synchronized market date, so a dense date cannot dominate the result. Tier separation becomes meaningful only across many mature observations and independent market dates.
07 Composition & Allocation Logic
Index Entry & Exit:
- Instruments enter the index with equal weighting (e.g., 10 stocks = 10% each)
- Exclusion rules triggered by:
- Lack of liquidity
- Excessive correlation
- Reaching maximum number of constituents
Cash Allocation (Golden Ratio):
The cash allocation is dynamically calculated based on the proportion of "Long" signals within the investable universe, following a ratio inspired by the Golden Ratio (1.6 β Ο).
08 Final Objective
To create dynamic, transparent, AI-managed thematic indices capable of:
- Adapting in real time to market changes
- Optimizing the risk/return profile
- Ensuring neutrality and the absence of bias
- Enabling personalized, automated portfolio management
Active Direct Indexing System
Powered by transparent, multi-layer AI
Combining quantitative analysis, risk control, and market cyclicality