Skip to main content

Activity Detector, Activity Detector TD

Structural changes in the price discovery of assets.

Model appearance.

Model appearance.

Model description​

The main purpose of the Activity Detector model is to identify structural changes in the price discovery of financial assets by analyzing volatility and trading volume. The model captures sharp spikes in these metrics, evaluates their significance, and classifies them as anomalous or not. When it detects anomalies, the system alerts the user right away, helping them react quickly to changing market conditions.

Methods for analyzing volatility and trading volume​

Activity Detector tracks changes in an asset's volatility and liquidity dynamically. It analyzes market data time series in real time. To do this, it:

  • Uses approximate statistical estimates of averaged and instantaneous volatility and volume values.
  • Applies smoothing filters and adaptive algorithms that minimize noise in the data and pick out significant changes.
  • Analyzes periods of accelerated change in volatility and volume, comparing them with the historical context of the asset's market behavior.
  • Applies thresholds and dynamic adaptation to exclude false signals.

Detecting anomalous activity​

Anomalous activity is determined from several factors at once:

  1. How far current volatility and volume deviate from their typical values over similar periods.
  2. The speed of change in market parameters: sharp jumps in price, volume, and spread are analyzed.
  3. Context parameters: general market conditions and macroeconomic factors are taken into account, which excludes natural periods of higher activity (for example, when important economic news is released).

Use of machine learning​

Activity Detector uses machine learning algorithms to improve the accuracy of the analysis, but the specific models are not disclosed. The main principles of how the system works with AI include:

  • Training the model on historical data to recognize typical and atypical patterns of market behavior.
  • Automatic updating of statistical thresholds and logical dependencies based on the dynamic behavior of the market.
  • Independent adaptation to changing market conditions without external intervention, which gives relevant and accurate signals even in volatile markets.

High accuracy and the factors behind it​

The model reaches high accuracy through:

  • A combination of adaptive statistical algorithms and machine learning methods that minimize the number of false alarms.
  • Continuous self-learning of the model on updated data, which makes the detection of anomalous events more reliable.
  • Detailed analysis of market conditions and exclusion of external factors that lead to false signals.
  • Optimized computations that process large volumes of data quickly and keep signals up to date.

User interaction and alert mechanisms​

Activity Detector is fully integrated into the SpreadFighter platform and gives the trader convenient tools for monitoring market activity. The main ways to interact include:

  • Instant alerts and signals shown in the platform interface when anomalous market activity is detected.
  • An indication of the model's confidence level in the detected changes.
  • The ability to view the history of recorded anomalies, so users can analyze past signals and how patterns changed.

Technical aspects of the model​

Activity Detector works in real time and processes data with minimal delay. The main technical characteristics of the model:

  • Processing of large volumes of market data at high speed.
  • Minimal false alarms thanks to adaptive filtering algorithms.
  • Automatic adaptation without manual parameter tuning.
  • Robust computational methods that keep the model stable even in conditions of high market volatility.

Activity Detector is a powerful analytical tool for traders. It lets you react quickly to critical market changes and reduce the risks of sudden swings in volatility and liquidity.

Model tests​

tip

Coming soon…

Location​

To add the "Activity Detector" tool to the workspace, first click the "Screeners" button in the platform header. Then click the button of the same name in the "Microstructure" section.

Model location.

Model location.

Functionality​

Widget​

The Activity Detector model has 2 widget variants: Activity Detector and Activity Detector Time Dominance.

  • Activity Detector finds the assets with the largest structural changes.
  • Activity Detector TD calculates the time in minutes during which the asset went through significant structural changes, and ranks the assets with the longest period of changes.

Data arrives in a table with the following columns:

  • Value - the Activity Detector model value.
  • Cum - the number of minutes during which the asset was stationary.

Widget.

Widget.

Indicator​

The model indicator is an oscillator. Blue shows anomalous activity, gray shows its absence. The indicator has background highlighting. The most intense highlight colors correspond to the largest structural changes in the asset.

Indicator.

Indicator.

Use cases​

tip

Coming soon…