Close this search box.
Close this search box.

Artificial Intelligence Deep Learning Machine Learning : How Artificial Intelligence Shaped Algorithmic Trading and Fraud Detection

Artificial Intelligence Deep Learning Machine Learning : How Artificial Intelligence Shaped Algorithmic Trading and Fraud Detection. Nearly ten years ago, the U.S. experienced one of the worst financial crises in history. 

Unemployment doubled to 10% at the peak of the recession, millions lost their homes, jobs,  and savings, and many Americans lost most of their wealth. This was directly attributed to the subprime market collapse. Which is why the S&P 500 (The stock market index of the 500 largest companies) fell by 57 percent between 2007 and 2009, leaving millions of Americans in financial ruin1.

The American people were left feeling cheated by the large banks. The Banks had sold home insurance loans knowing those that took on  the loans could not pay them back. Moreover, the Great Recession had left Americans with the impression that they should not trust the large corporations with their money. Because at the end of the day, they would be the ones that would be the proverbial “suckers”.  

Many young Americans still avoid the stock market nearly a decade after the Great Recession2 due to how much it affected their lives when they were young.

But, what they don’t  realize is that the increase in technology in the financial industry has helped keep the risk of  investing low. Most notably, Artificial Intelligence has helped keep risk low while improving companies investment strategies. Let’s look at how AI has achieved that.

Artificial Intelligence in the 21st century has become a buzz-word that is in every field  today: Medicine, Farming, Retail, etc…. So how has Artificial Intelligence taken over the financial industry? And when did it start? AI was first introduced to finance in the 1980s when  James Simons founded the first quantitative investment firm4. A quantitative investment firm is when an investment firm uses programmed investment strategies to complete trades3. Simons’s work with his firm Renaissance Technologies was revolutionary, legitimizing a  quantitative approach to trading.

By removing emotion from investing and focusing on the  data, Simons was the first to consistently succeed in the market using a data-driven approach6

Simons’s background in code-breaking and mathematics was especially helpful when building  algorithms to “solve the market.”14 Furthermore, this is what lead to Renaissance Technologies unique hiring  strategy, they weren’t hiring MBAs with majors in Finance, but students who excelled in Math, Physics, Biochemistry, Computer Science, etc6. The employees of Renaissance Technologies  weren’t completing trades, but building the algorithms who traded in the market. 

Quantitative finance seems like one of the safest ways to make money, right?

Not so fast. While algorithmic trading may be beneficial, it has caused some of the largest crashes in history (most notably the 1987 meltdown). At that time, algorithmic trading was a fairly novice  practice with little real experience in the market.

While algorithmic trading had been used  before, it had not been tested on such a large scale. With markets expected to dip due to the  expected decrease in the value of the dollar, this was the first time algorithmic trading was  being used at this magnitude. The events that followed led to the largest single day drop in  DIJAs (Dow Jones Industrial Average) history7.

The algorithms performing these actions were not built to deal with this much volatility in the market, because it activated a chain of stop-loss orders.

Stop-loss orders if you don’t know are made to limit an investors loss, by selling stock if  the stock falls below a certain price. It is not only necessary to have a good algorithm, but is  necessary for the algorithm built doesn’t cause mayhem in the market.  

Throughout history, there has only been one other time where algorithmic trading has  caused this massive amount of damage.

Commonly referred to as the flash crash, a firm by the  name of Knight Capital Group lost $440 million in a span of less than 45 minutes.

While similar to the 1987 crash, Knight was trying out a new trading software, but it contained a major flaw that had caused a massive ripple effect through the market. It had sold a massive amount of stock at an absurdly low price in less than 10 minutes. This caused many other algorithms to go  awry and caused the DIJA to drop 1,000 points in 10 minutes.8

A few hours later, the prices of stocks returned to their pre-crash levels, but if HFT could cause these massive fluctuations in the market, was HFT worth the risks? 

Many in the industry have come to understand of the risks of High-Frequency Trading,  which is why in recent years many firms have implemented measures to combat these risks, such as “kill switches” to stop these flash crashes, and regulations on the algorithms, so they  have “pre-trade risk controls”.9 

While quantitative finance does have its risks, it is a vital part of the U.S. market today.

Where 90% of the volume is traded by quantitative means and is estimated to be a $1 trillion market, if we take the right precautions, it can continue to benefit the U.S. market.10

While we understand that algorithmic trading can better predict markets than a typical trader, what exactly makes Quantitative Trading a $1 Trillion market? In addition, two significant benefits stand out  from the rest: speed and accuracy. Algorithmic Traders can make trades in 10 milliseconds or less, to give some scope as to how fast that is, it takes around 300 milliseconds for you to blink  your eye.11

The second major benefit is accuracy, by taking out human emotion and relying solely on data to make trades it makes Algorithmic Trading valuable. 

Until recently, hedge funds and HFT-firms were the sole users of AI in finance. Some other uses include risk-management, optimization, robo-advising, and many more. But, what I am mainly going to focus on is how risk management and fraud detection has changed due to  AI. 

What is a Financial Risk Management Software?
These solutions safeguard an organization’s finances by offering a clear picture of dangers to cash flows and revenue, as well as the ability to manage and mitigate them.

In practice, any application or module that aids in the dissemination of information or the formulation of improved financial risk decisions falls into this category. Approving transactions, fighting chargebacks, verifying compliance, and onboarding new consumers are just a few examples.

There are also solutions for the finance and banking industry, which generally address industry-specific demands and pain issues, such as underwriting and credit rating, payment approvals, and financial fraud protection.

How to Choose a Financial Risk Management Software?
To combat transaction fraud, there is only one way to go: learn more about your clients. However, what you can do with it depends on the risk software you’re using:

Data tools: These programs allow you to collect additional information from a single place, such as an email address or a phone number. Excellent for manual reviews or for managers that require further information. Some generate a risk assessment, while others merely deliver the data in its original form.

End-to-end platform: You’ll get additional data and rules to help you generate a risk score faster. This gives you a lot more power and flexibility when it comes to risk mitigation. The most advanced solutions will utilize machine learning to automatically identify new risk rules.

E-commerce and online financial transactions has become a worldwide phenomenon over the last decade and will only continue to increase. While this has largely benefited consumers, giving access to products that many did not have before, easing travel and cost, all while lowering costs. E-commerce has also increased online fraud throughout the years. Moreover, the ones that have been hit most by this have not been just the consumers, but the large banks as well. Losing over $70 Billion dollars per year the US banks have much reason to combat this  growing problem.2

Artificial Intelligence & Machine Learning – Rebellion Research

One of the easiest ways to detect fraudulent activity is to run an analysis on  the first digits in the given data, commonly known as Benford’s Law. Artificial Intelligence can run this analysis through millions of data points, but the main use of AI in detecting fraud is  through Machine Learning. Furthermore, to give a small definition of what ML(machine learning) is that it is  an application of AI, and focuses on the development of computer programs that can access data and use it to learn for themselves.12

By having ML algorithms analyze millions of data, they  can determine what is defined as real transaction and what is fraudulent. If they find a transaction turns out to be fraudulent, the transaction is then flagged, and the bank  investigates the transaction.  

One part of the financial industry ML has been quite successful in has been credit card  fraud detection. The algorithm that monitors your transaction history uses your previous transaction data to determine whether it is most likely you who is using the card on the current  transaction. If the ML algorithm believes it is a fraudulent transaction. Then it is flagged and looked at by a human representative.2

This process enables the flagging of fraudulent purchases in real-time.  

Unlike Algorithmic Trading, using Artificial Intelligence with Risk Management is still a fairly new idea.

But, it is set to transform the market within the next ten years. Firms are  beginning to quantify everything from someone’s believability to online activity. Commonly referred to as alternative data13, it is beginning both to transform the internal risks in a  company and improve upon the current infrastructure in credit risk decisions.  

Artificial Intelligence in the financial industry, a novice idea a couple of decades ago, has transformed the financial industry and will continue to do so in the future. Furthermore, AI has had a few bumps along the way, most notably the 1987 and 2010 crashes, but due to these crashes, the Financial industry implemented precautionary measures to avoid these crashes in the future. 

So, we can use AI to our benefit by mitigating the risks attributed to it.

In conclusion, for any young Americans that are steering clear of investing due to the 2008 crash, they must understand just how much the industry has changed over the last decade. In addition, the risks in the Industry are much lower, not just AI used in trading, but how it is transforming the workplace as well. Lastly, by quantifying everything, we are more equipped for the unexpected.  

Artificial Intelligence Deep Learning Machine Learning : How Artificial Intelligence Shaped Algorithmic Trading and Fraud Detection By: Matt McManus 

Matt is a current student at MIT with a passion for Ai & Quantitative Investing!

Works Cited: 

1. Duignan, Brian. “Great Recession.” Encyclopædia Britannica, Encyclopædia Britannica,  Inc., 26 Sept. 2019, 

2. Shawncarterm. “Younger Americans Aren’t Investing in the Stock Market-Researchers  Think This Is Why.” CNBC, CNBC, 16 May 2018, why-younger-americans-arent-investing-in-the-stock-market.html. 

3. Chen, James. “Quant Fund.” Investopedia, Investopedia, 29 Jan. 2020, 

4. Buchanan, Bonnie G. Artificial Intelligence in Finance. Apr. 2019, _turing_report_0.pdf. 

5. Zuckerman, Gregory. “How Billionaire Jim Simons Learned To Beat The Market-And  Began Wall Street’s Quant Revolution.” Forbes, Forbes Magazine, 8 Nov. 2019, the-market-gregory-zuckerman-book-excerpt/#6447489d13b6. 

6. Keh. “Billionaire Robots: Machine Learning at Renaissance Technologies.” Technology  and Operations Management, robots-machine-learning-at-renaissance-technologies/. 

7. Segal, Troy. “What Caused Black Monday: The Stock Market Crash of 1987?”  Investopedia, Investopedia, 29 Jan. 2020, crash-1987.asp. 

8. “2010 Flash Crash – Overview, Main Events, Investigation and Aftermath.” Corporate  Finance Institute, investing/2010-flash-crash/. 

9. Zurkus, Kacy, et al. “Cyber Kill Switch: The Good, the Bad and the Potentially Ugly.”  Security Boulevard, 4 Mar. 2020, good-the-bad-and-the-potentially-ugly/. 

10. Henry, Gareth. “The Rise of Quantitative Investing.” Medium, Medium, 27 Nov. 2018, 11. Equedia. “How Fast Is High-Frequency Trading? Faster Than You Think.” Equedia  Investment Research, 26 May 2017, trading/. 

12. “What Is Machine Learning? A Definition.” Expert System, 11 Nov. 2019,

13. “Artificial Intelligence Is a Game Changer for Risk Management in Finance.”, Bloomberg, intelligence-game-changer-risk-management-finance/

14. Stevens, Pippa. “Furthermore, The Secret behind the Greatest Modern Day Moneymaker on Wall  Street: Remove All Emotion.” CNBC, CNBC, 5 Nov. 2019, the-market.html.

Artificial Intelligence Deep Learning Machine Learning

Artificial Intelligence Deep Learning Machine Learning