Financial technology, or Fintech for short is a relatively avoided topic among tech enthusiasts, developers, programmers and etc. The reason is very simple actually. Developers don’t necessarily refer to their software as Fintech even though it’s quite literally associated with the financial industry.
Most of them use the term only to help people outside of the industry understand what the software is designed to do. But in the process of helping others understand what the software is being developed for, the term Fintech has accumulated quite a lot of characteristics.
What I mean is that almost everything starting from online banking and ending with the blockchain technology is referred to as Fintech nowadays, which isn’t an easy way to make distinctions among all the available software.
Despite this though, developers continue to create new and innovative solutions to even the smallest issues that both the providers and consumers of financial services face on a daily basis. Nobody seems to be too bothered about the broadening of the whole idea of Fintech, which gives it a larger library of software to choose from and potentially combine in the future.
In this article, I’d like to familiarize you with three distinct “segments” of Fintech and how they’ve managed to revolutionize not only private finances, but the finances of the whole world.
Not only will I simply say what they do, but I’ll make sure to tell you how they do it as well. Let’s begin with my favorite one (as that’s the only thing I can code myself so far).
Sentiment Analysis AI
Although sentiment analysis AI is not necessarily directed to finances, as it can detect the sentiment of almost any topic, it’s still one of the most often used tools in the financial world.
More specifically, it’s commonly used by investors that spend most of their time looking at the financial markets to determine which company share, currency or even cryptocurrency is worth it to invest in.
They did this through explanations of how it works, why it works, and when it’s best to use it. So why don’t we find it out for ourselves as well?
A sentiment analysis AI requires only 3 resources:
- Twitter API
- A lexicon library of almost every word in English or a specific language
- A little knowledge of Python, PHP, Java or various other languages.
The scientist can indicate the range of dates on when these Tweets were issued. The most common date range is any Tweet that was published within the last month or so.
Once the Tweet with the keyword has been identified, the algorithm, which is usually premade breaks it down. This is called tokenizing. What happens here is that the Tweet is broken into separate words with their own attributes of positivity ranging from -1 to 1.
Once all of the attributes have been assigned to these tokens, they are combined as yet another sentence and the algorithm tries to read the most accurate analysis of the Tweet. In most cases, the result is 0.04 or -0.01 on the scale. If it gives a perfect 1 or -1, then there was something wrong with the algorithm as that’s highly unlikely.
This rating is assigned to every single Tweet with that specific keyword. Then it is calculated and an average is determined. The final result is usually the prediction that our algorithm gives us about a sentiment.
Anything below zero means a negative sentiment, and anything above it means positive.
Once the sentiment is determined, it’s up to the trader to factor in his or her own personal knowledge and places a trade he or she feels most comfortable with.
Naturally, this does not have a 100% success rate as thing tends to change quite fast in the financial world as information gets passed down very quickly.
The same thing can be down with the API of pretty much any major network or platform.
Next, I’d like to talk about High-Frequency Trading which is not that new to the world of finance. It’s not that dependent on analysis as well. It’s mostly designed to somehow manipulate the prices on the various exchanges in the world in order to help the “owner” of the system profit as much as possible.
In order to explain it, I need to tell you two main attributes for making this Fintech method work.
These are the location of the servers, and the processing speed of the computer being used.
Jim is an investor that really likes to play video games, which is why he is going to choose a video game studio as his primary investment for stocks. He decides that he wants to invest in Blizzard shares and places an order.
Jim wants to buy 1000 Blizzard shares for a maximum of $55 per share. In order for the market to handle such a large order, it needs to segment it somehow. Usually, this happens through breaking the order down by 500 or 100 shares per trade.
Let’s assume that Jim’s trade just got broken down to ten 100 share orders. The first order is processed immediately as there was somebody selling the shares for $53 apiece. The exchange saw it, matched the two orders and completed them within an instant.
This is exactly when the High-Frequency Trader comes in with his algorithm.
You see, once the first trade had been completed, the HFT algorithm noticed it and decided to try its luck with the next batch of the order. It was programmed to take into account the breaking down of large orders, meaning that it knows there’s still more orders to fill.
It tries its luck with $60 per share, but it doesn’t pass as the maximum that Jim indicated was $55. Then it tries $59, $58 and etc all the way down to $55.
The algorithm is programmed to determine the current market price and try to sell these shares to the buyer for a slightly larger price as people tend to indicate a slightly higher maximum than the market price.
Once the $55 maximum is determined, the algorithm quickly buys up circulating sell orders for Blizzard shares under $55 and sell them to Jim, ultimately filling his order.
Considering the server that was executing the algorithm was located close to the exchange’s server, all of what I just described happened in 0.000001 seconds.
It’s quite a remarkable strategy for trading, but an expensive and unfair one at that. Most exchanges are considering to ban or tax High-Frequency Trading completely as people like Jim don’t get the chance to purchase their desired shares under the maximum price that they’ve indicated, thus concentrating a lot of the profits in a singular location, the HFT user.
More innovation to come in the future
The two examples I’ve showcased here are just the tip of the iceberg of just one sector of the financial industry.
Innovative technology in the banking sector, lending sector and various other parts of finance are being developed as we speak.
But, I highly doubt that any of them can bring revolutionary results compared to what the blockchain is doing right now.
- Speed of transaction
- Security of transaction
- Cost of transaction
Any new project targeting almost anything else is considered much less impactful, but interesting and fascinating nonetheless.
Hopefully the next time you’ll want to determine if pineapple on Pizza is popular among Twitter users, you’ll remember this article and use the sentiment analysis AI. And the next time you buy some company shares, remember that there may be a High-Frequency Trader somewhere in the shadows trying to sell them to you at maximum price.