Image of market data displayed by Quantarix AI’s AI market analysis platform
AI DATA INTELLIGENCE / FINTECH

Analyze market data in real time and back up your buying and selling decisions with numbers

Quantarix AI uses predictive models and back-tested algorithms to analyze minute-by-minute market fluctuations. Generated signals are recorded in public logs and disclosed for anyone to verify.

access the system Check performance log
Verification log recording method: All signals are stored separately with time stamps, expected entry prices, and judgment results, and are retroactively reflected in the public log after they occur.
PROBLEM CONTEXT

The amount of information and the speed of change exceed the processing capacity of humans.

In today's markets, multiple data streams are updated simultaneously and asynchronously, including market depth, trading volume, news feeds, and social media sentiment. It is structurally difficult for individual traders to manually integrate these and react to millisecond changes.

As a result, important signals can get lost in the noise. Quantarix AI is designed with this processing load in mind, and rather than replacing human judgment, it continuously supplies analysis results that serve as basis for judgment.

Image of multiple market data streams analyzed by Quantarix AI
CORE TECHNOLOGY

Configuring predictive models and real-time analysis engines

We avoid marketing-like abstract expressions and explain the processing target and mechanism of each component as technical specifications.

Prediction model (time series analysis)

Using a time series model trained on past price and volume data, we estimate the probability distribution of short-term price fluctuations. The model is regularly retrained to follow changes in market regimes.

Real-time analysis engine

It receives tick data distributed from exchanges via streaming and operates in a configuration that minimizes the delay from reception to completion of processing. When the processing load increases, the analysis targets are sorted by priority.

Backtest environment

New models and parameters are validated on historical data before being published. The verification results are saved in the same structure as the log format and can be compared later.

Risk management logic

Signals are accompanied by guidelines for position sizing and drawdown control. This is a recommendation and the final ordering decision is left to the user.

Data processing flow

Data acquisition → Preprocessing/normalization → model inference → signal generation → Log recording/distribution
TRANSPARENCY LOG

performance log

All generated signals will be reflected in the public log after the results are determined. Below is a sample showing the display format of the log screen.

* The table below is sample data showing the log display configuration. The actual numbers will be updated from time to time on the public log screen.

signal id Date and time of occurrence Estimated entry price Judgment criteria Validation status
SG-00231 09:30:12 2,480.00 Closing price standard/next business day Verified
SG-00232 10:14:48 1,192.50 Closing price standard/next business day Verified
SG-00233 13:02:05 874.00 Closing price standard/next business day Waiting for judgment

Verification method: For each signal, the timestamp at the time of generation and the expected entry price are immediately recorded, and the results are added after the judgment conditions are met. The recorded occurrence time and price will not be changed later, and only the judgment result will be added, so there will be no subsequent rewriting of the numbers.

USE CASES

Utilization design from day trading to medium- to long-term operations

Signal distribution for high frequency day trading

Targeting short-term price fluctuations on a minute-by-minute and hour-by-hour basis, we continuously process everything from receiving tick data to generating signals. In situations where speedy execution decisions are required, analysis results can be received via a trading app or API.

Design guidelines It is configured to keep the processing delay from tick reception to signal generation to a certain level, and the processing time can be checked for each signal on the public log.

Mid- to long-term portfolio optimization support

We combine daily and weekly market data with macro indicators to provide analytical reports that help you review your asset allocation. It is designed to emphasize changes in correlation structure and risk indicators rather than short-term price movements.

How to use it In the monthly report, you can check the risk indicators of the entire portfolio and the judgment results of past signals side by side.
FAQ

Technical questions regarding API integration, latency, and model learning

Can the API be linked with existing trading systems?

We are delivering signal data via REST API and WebSocket. It uses an API key for authentication and is designed to be integrated into existing automatic ordering systems and bots. We do not provide the ordering function itself, but concentrate on distributing analysis results.

What is the latency (processing delay)?

The processing from tick data reception to signal generation is designed with the goal of low latency, but actual values ​​will vary depending on market load conditions and communication environment. The time each signal occurs is recorded in the public log, so you can check the actual delay between delivery and reflection after the fact.

How is the predictive model trained and updated?

The model is trained using past price and volume data, and has relearning cycles in response to changes in the market environment. The performance difference before and after relearning will be compared in a backtest environment, and then it will be decided whether to reflect it in the production environment.

Which market/stock data are you analyzing?

Focusing on stocks listed on major domestic exchanges, we target stocks whose trading volume and trading frequency meet certain standards. The scope may be revised depending on the availability of data sources.

Can anyone verify the results in the logs?

Public logs can be viewed after registering an account. The recorded occurrence time, estimated price, and judgment result are fixed and have a structure that cannot be changed later, so they can be aggregated and verified over any period of time.

Check verifiable logs with your own eyes first

The Quantarix AI system is designed to be evaluated by records, not claims. After creating an account, you will be able to view public logs and analysis reports.

access the system See feature details