Automatic Alarm Classification using supervised learning techniques
A Random Forest is trained on a high imbalanced dataset of automatically raised alarms to recognize false positives.
Portfolio Margin Minimization and Combinatorial Optimization
Aiming at minimizing the margin requirements for an option portfolio by employing a combinatorial optimization based approach.
Anomaly Detection in Realtime Series
A real-time anomaly detection tool based on the Hierarchical Temporal Memory (HTM) network.
Reinforcement learning for optimized trade execution in financial markets
An analysis based on a reinforcement learning (RL) technique to address the problem of optimized trade execution in financial markets.
Recommendation Systems as anomaly detection tool for market abuse
Applying Recommendation Systems techniques in a reverse way to create an anomaly detection tool: our first experiment in the Market Abuse field.

Market Abuse. Results for the reduction of false positives alarms

RF MA imbalance FP

In this last post we finally present if the Random Forest trained on the past activity of the compliance officer is able to classify an alarm as false positive or not.

Market Abuse. Classification with high imbalanced dataset

RF MA imbalance

In this second post of the series, we introduce: the dataset used for the classification problem, the ML approach that better manages the high-imbalanced dataset and finally the statistical metrics used to measure the goodness of the results.

Market Abuse. Is it possible to reduce false positive alarms?

RF MA

In this series, we explore the possibility to use classical supervised learning techniques to classify alarms generated by a system that implements the Market Abuse Regulation MAR. If successful, this machine learning application could help Compliance Officers to reduce the effort spent in the analysis and classification of the alarms generated.

Margin Minimization model

MarginOpt CombOpt

In this second and last episode of the series we will propose a combinatorial optimization model to heuristically solve the margin minimization issue applied to a portfolio of options.

Portfolio Margin Minimization and Combinatorial Optimization

MarginOpt CombOpt

In this blog post series, we will present an analysis based on a combinatorial optimization approach to address the problem of margin minimization of an option portfolio. In this first post we will describe in detail the general context in which the problem is inserted and the various techniques which in principle can be used to solve this kind of issues.

NFK model and LIST Smart Order

OptExec RL

In this last episode of the series, we will present an extended analysis to test the performance of the NFK model against the LIST Smart Order strategy.

NFK model and Surrogate Smart Order

OptExec RL

In this blog post, we will present a preliminary analysis to test the performance of the NFK model against a naive strategy that we have called Surrogate Smart Order.

The NFK model

OptExec RL

In this blog post, we will present in detail the characteristics of our reinforcement learning model to approach the optimal execution problem.

Daily volume profiles and market VWAPs

OptExec RL

Before starting to describe our reinforcement learning model for optimal execution, let us introduce a few elements which might be helpful to better understand some passages of the following posts.

RecSys for MAD: backtesting results

RecSys AD

As a final review of our experiment, we performed a backtesting analysis — sort of, since we are in a non-supervised learning setup. Here we present the main outcomes with an attempt to statistically inspect the resulting top anomalies.

Universal Anomaly Score

RecSys AD

In this post we show some examples of anomaly rank results, highlighting the need for transformation of such values that makes the score interpretable in a universal way, overcoming the specific scale and shape of each RecSys outcome.

Evaluation of a RecSys as Anomaly Detector

RecSys AD

Using a RecSys as an anomaly detector undermines the possibility of a supervised-learning approach, so traditional mean average precision metrics are not enough. But even before that, we had to face a fitting convergence evaluation problem.

Anomaly detection with HTM: an interesting case study

ADTS AD

In this last blog post of the ADTS series we will show an interesting example of anomaly detected on the MTA market data series.

Anomaly detection with HTM: a real world example

ADTS AD

In this blog post of the ADTS series we will show an application of an HTM network on a real-time series.

Anomaly detection with HTM: Anomaly Score and Anomaly Likelihood

ADTS AD

In this blog post we will show how to use an HTM network to solve an anomaly detection problem.

Evaluation of a RecSys

RecSys

Next steps are tuning the remaining hyperparameters and training the model: but how to establish the soundness and the accuracy of the calibration outcome? The ROC curve, the AOC and the precision/recall at k are the standard metrics aimed at this purpose.

Hierarchical Temporal Memory (HTM)

ADTS AD

In this blog post we will present a short overview of the Hierarchical Temporal Memory theory.

RecSys for MAD: an empirical study

RecSys

In this second episode of the series we'll introduce the ‘real-world’ dataset we've been dealing with. In particular we will discuss the foremost step of the hyperparameter selection phase, namely the mapping we've adopted in order to be able to feed this dataset into a RecSys facility.

Anomaly Detection in Time Series

ADTS AD

In this blog post series, we will present a real-time anomaly detection tool based on the Hierarchical Temporal Memory (HTM) network. In this first post we will briefly describe the problem of anomaly detection in time series.

Optimal Order Execution and Reinforcement Learning

OptExec RL

In this blog post series, we will present an analysis based on a reinforcement learning (RL) technique to address the problem of optimized trade execution in financial markets. We will then compare the performance of our model with the outcomes of the LIST Smart Order strategy.

Market Abuse, Recommendation Systems and Anomaly Detection

RecSys AD MA

As the first thread of our blog, we start with a rather ambitious project: applying Recommendation Systems techniques to create an anomaly detection tool. The approach is quite general, but here we will tell you specifically about our first experiment in the Market Abuse field. In this first post of the series we will briefly describe the terms of the matter.