AI Tech Renaissance…impacting the bottom line
Machine Learning Technology Maturity Model and how it is helping firms increase margins and improving compliance
No matter what you call it — “machine learning(ML),” “cognitive computing,” “Internet of Things (IoT),” “smart machines” or “intelligent automation” — Artificial Intelligence (AI) is basking in a renaissance now. It is not simply because of the promise AI holds for the future but because of the impact it is having on every facet of many businesses today, especially the Financial Services and Compliance industry.
The AI promise for Financial Markets Compliance (FMC)
Wall Street’s future will be dominated by firms that embrace technology from electronic fixed-income/equity trading systems to predictive and prescriptive algorithmic systems, reducing the need for people in the front and back office. Global investment banks that successfully adopt automated trading and other measures can increase profit by 30 percent. On the sales and trading front, digital channels will become the defacto medium for interacting and transacting with clients. Coverage models will shift radically as banks ruthlessly examine every front-office routine and manual activity for its automation potential. Financial institutions must overcome the shortcomings of existing compliance pipelines that do not live up to the standards of expanding new regulations. Capital markets tends to be a fast-follower industry. Things like machine learning are automating and resolving complex challenges faced by the industry. Everyone will very quickly move to it…and you surely do not want to be left out!
Machine Learning in Financial Markets — A Primer
What is this thing called Machine Learning and how can you benefit from it?
So, what is AI/Machine Learning (ML)? ML is based on algorithms that can learn from data without relying on rules-based programming. It came into its own as a scientific discipline in the late 1990s as steady advances in digitization and low cost computing power enabled data scientists to stop building finished models and instead train computers to do so. The unmanageable volume and complexity of the big data that the world is now swimming in have increased the potential of machine learning — and the need for it.
ML in FMC
ML can transform the way that transaction monitoring technologies identify suspicious activity. Machine Learning can be used to learn transaction behavior for similar customers, discover transaction activity of customers with similar traits, pinpoint customers with similar transactions behavior, identify outlier transactions and outlier customers, learn money laundering typologies and identify typology specific risks, dynamically learn correlations between alerts which produced verified suspicious activity reports, and continuously analyze false-positive alerts and learn common predictors.
Complex business processes and layers of compliance pose a challenge for traditional rule-based programmatic solutions. ML delivers an adaptive learning process where the technology self-learns traits and provides anomalies and alerts in a dynamic fashion, reducing human intervention and the need for constant heavy lifting programming.
The C-Suite Quandary
Addressing the Pain Points faced by C-Level Execs
It is not just about Dodd-Frank. Other emerging rules issued by the SEC and the SROs have added to the already heavy compliance burden in this industry. Compliance functions at financial services firms continue to face diverse pressures with broadening compliance remits, ever-increasing volume of regulatory changes, rising personal liability, resource staffing challenges, and myriads of communication channels with clients and partners that serve as potential conduits to fraud and non-compliance. Many mid-to-large size firms do not have sufficient resources to manage their compliance programs effectively, with over half of these firms reporting a status-quo or decline in compliance budgets. The compliance department is also being exposed with the risk of being deemed supervisor.
From Boardroom to C-Suite to the compliance head, executives face the challenge of how to do more with less, how to remain fully compliant in an evolving regulatory environ, and how to innovate to stay ahead of competition. The tangible and intangible risks of compliance — fines, personal liability, litigation, brand reputation — needless to say, can make or break a firm.
Traditional and Innovative Solutions that fit the bill
A plethora of solutions crowd the market. These range from tack-on solutions to rule-based systems, to Enterprise Data Management solutions to Big Data platforms. A majority of the solutions are focused on audit trail and compliance — after-the-fact. In other words, they are track-&-trace technologies that provide audit trails to identify fraud and misuse. These are the descriptive technologies.
Big Data Analytics Cloud Solutions
With Big Data, you can embrace predictive and prescriptive tools to identify potential threats before they occur. In many sectors, including FMC, there’s a much more urgent need to embrace the prediction stage, which is happening right now. Today’s cutting-edge technology already allows your businesses not only to look at the historical data but also to predict behavior or outcomes in the future — for example, by helping credit-risk officers at banks to assess which customers are most likely to default or by enabling communications compliance to anticipate which customers are especially prone to “churn” in the near term.
A frequent concern for the C-suite when it embarks on the prediction stage is the quality of the data. That concern often paralyzes executives. Hence Big data analytics is an elixir for sure, but the data needs to be cleansed, merged, deduped for starters and you need a small army of data scientists to ensure you are gleaning the right information from the data deluge.
Link Analysis Solutions
In its most basic form Link Analysis establishes relationships between entities using a variety of techniques. The criteria can be as simple as exact matching (e.g., phone numbers and email addresses match) to complex fuzzy matching techniques that use complex match codes traditionally found in data quality tools for partial record matching. Statistical techniques have also been employed to identify relationships between events and entities, as well as to prune relationships to “relations of interest.” Link analysis is most often used to define an entity as well as identify who may be related, how they are related and, finally, what the entity’s behavior is or is likely to be. Often, link analysis is used to augment investigations by identifying nonobvious relationships and is used to extract features that can be used to enhance other detection techniques. One of the major benefits of link analysis is that it is relatively easy to combine structured content (transactions) and unstructured content (comments) into a single graph that tells a story.
But where is the automated Prescriptive Analytics approach?…
The Aha! Moment: An Effective platform at the Forefront
The Hyper-effective solution for Financial Market Compliance
The missing piece in most innovative solutions is “Prescriptive Analytics”- the most advanced stage of Machine Learning. Voila! — The Next-Gen FMC Platform
Prescriptive Analytics, the third and most advanced stage of machine learning — is the opportunity of the future and must therefore command your attention. It is, after all, not enough just to predict what customers are going to do; only by understanding why they are going to do it, can companies encourage or deter that behavior in the future. Technically, today’s machine-learning algorithms, aided by human translators, can already do this. For example, a global bank concerned about the scale of defaults in its retail business, recently identified a group of customers who had suddenly switched from using credit cards during the day to using them in the middle of the night. That pattern was accompanied by a steep decrease in their savings rate. After consulting branch managers, the bank further discovered that the people behaving in this way were also coping with some recent stressful event. As a result, all customers tagged by the algorithm as members of that micro-segment were automatically given a new limit on their credit cards and offered financial advice The prescription stage of machine learning, ushering in a new era of man–machine collaboration, will require the biggest change in the way we work. While the machine identifies patterns, the human translator’s responsibility will be to interpret them for different micro-segments and to recommend a course of action. Here you C-suite folks must be directly involved in the crafting and formulation of the objectives that such algorithms attempt to optimize.
The real-time analytics pipeline is the heart of this next-gen solution. It determines whether a given trade or communication item violates regulation or not. The platform supports traditional e-discovery methods, such as search, but, more importantly, it features a complete machine-learning pipeline with multiple predictive models and modeling techniques like:
· Classification — To identify irrelevant messages such as automated notifications, newsletters, out-of-office messages, and print job notifications.
· Graph Analysis — To build communication profiles of individuals. This technique is often used in security analytics and malware detection to identify anomalous behavior. Graph analysis can establish hot spots of fraudulent activity based on who is talking to whom.
· Text Analytics — To identify the language behind fraud, determine the sentiment and certainty in the language of a trader before and after executing trades. Semantically, it can interpret if too much information (e.g. deal coloring) was involved with communication partners.
Reaping the Benefits
ROI driven. Proven Solutions. Real Results.
In 2014, a significant number of regulatory enforcement cases against banks cited the lack of a systemized approach to FMC supervision and the lack of surveillance on a daily basis. Increasing regulatory expectations are forcing banks and other financial institutions to integrate external data into their operational risk framework. This is driving the need to use machine learning to pull external data and set filters to get relevant and scaled data on an automated basis.
Machine learning and neural analysis can cross-check banking data with public record databases to determine bad actors or suspicious transactions. However, the advanced solutions are best suited for solving the problem of entity resolution related to financial crime not only incorporate machine learning techniques and processes but do so with a principles vs. rules approach.
When tackling the problem of entity resolution, AI and machine learning techniques can be utilized to compare two records and measure the probability and risk associated with a match. This is representative of classification, a desired output of a machine learned system. Inputs are divided into two or more classes and a model is produced that assigns unseen inputs to one or more of these classes. Ordering these results in a two-dimensional observation space enables the compliance analyst to determine if two records represent the same entity. It also allows the analyst to establish thresholds of what to review and what to exclude from review. Other advanced analytic tools such as social network analysis and scatter plot visualizations can facilitate the detection of relationships and anomalies.
A Financial Compliance solution designed to support machine-learning tasks and advanced analytics is an effective model to identify and monitor risk across the enterprise. It helps institutions to:
• Identify hidden risk
• Prioritize and order alerts
• Improve operational efficiency
• Optimize resources for more targeted investigation efforts
• Eliminate false negatives and significantly reduce false positives
Machine learning applied to Financial markets provides a principled approach and framework that handles entity resolution in a formal but flexible and extensible way. It explicitly factors in the scale, inherent uncertainty and rapid change involved in large real-world tasks. Whereas fragile systems crumble at Internet scale, robust systems can withstand increasing stress and anti-fragile systems actually improve with stress. Robust, anti-fragile solutions make better inferences by bottom-up learning from data, top-down knowledge engineering or horizontal discovery of new evidence artifacts
The need of the hour is a proven solution that has been field tested and addresses critical challenges faced by compliance departments along with a quantifiable ROI.
What is the Next Step?
This innovative approach focuses the compliance team’s attention on actual fraud cases. To enable effective compliance reviews for analysts, a dynamic user interface is an absolute must. The user interface provides the opportunity to make the system smarter as a whole. For example, a properly designed UI can solicit decision-making information from compliance analysts that can be automatically integrated into a feedback loop for analytics — a continuous learning system that gets smarter over time. Such feedback is instrumental to the system for the following reasons:
• Keeps the system up to date, for example, in the face of changing regulations.
• Provides more algorithmic training information in the form of fraudulent trades or communications.
• Injects additional domain knowledge such as new expressions used by traders or new types of fraudulent transactions.
So …no place for agony in the long term?
It’s always hard to be sure, but a unified enterprise risk framework and machine learning should be a top priority on the C-suite agenda. We anticipate a time when the discussion of what intelligence, “artificial” or otherwise, might be, will end, because they would be processes in every-day life. If all financial services organizations and autonomous bodies act intelligently, perform intelligently, and respond intelligently, we will cease to debate whether high-level intelligence other than the human intelligence exists. A next-gen Financial machine learning solution can leap-frog your business.