On Wall Street and finance, there has been a clear progression that models the use of algorithms and automation more generally in a society. First is that financial exchanges have become largely electronic which enables the possibility of training becoming more algorithmic.
A person can submit an order through a person or use an API. The API is much faster and is the choice for most traders because of things like optimized execution, for example if you take a brokerage firm, one of the lines of business might be on behalf of large institutional clients taking what we might consider a difficult trade so it's not like a family investor where they want to buy 100 shares of Microsoft. It's a large hedge fund and if you want to do it over the span of a day and it's such a large volume that if you're not clever about how you break that trade up you over time you will push prices around in a way that hurts your execution.
This is a control problem; we know how to design algorithms and create a machine that’ll follow these instructions perfectly. We can take volumes of history in real-time data optimizing the schedule in which we trade and have similarly high-frequency trading. Which is closely related but not the same as optimized execution is where you're trying to spot something very temporary therefore you don't miss pricing between exchanges or within an asset itself or just predict directional movement of a stock because of very low-level granular buying and selling data. In the exchange, this is what machines are efficient at doing as opposed to people.
I think we are in an era where there has been successful quant hedge funds that are operated traditionally with statistical arbitrage. Warren Buffett style timescales are better without AI as these are things such as assessing the long-term value of Microsoft. Where we should be looking at what Microsoft is functioning at today, you have to foresee what Microsoft will be doing in the future and be prepared to sit on the stocks for 10-20 years. AI does not come in clutch when you are making decision like this as there isn’t enough historical data in the system to make a solid decision. Something over this sort of time scale will need to hold out through recessions and even wars yet still return a profit in the long run.
Millions of consumers are victims of identity theft or fraud every year and that number since we're going up in the past few years is a pretty shocking test last year there was a larger share of young people between the ages of 20 and 29 that had fraud complaints than there were senior citizens age 78 and above according to report from the FTC. More than half of the people who suffered a fraud-related financial loss had a financial degree in 2016 payment processors and companies both small and large have to protect themselves and their users from identity theft there are a lot of risks to manage and the world of cryptocurrency has opened up a Pandora's box of deception. As you can see there's a mass of solutions to help both consumers and businesses keep themselves safe. Following a chain of criteria when a transaction is made will allow a processor to determine whether it is fraudulent or not, an example could be a transaction made from a newly opened account for a large sum of money. We can label it as fraud the result of this process is generally a binary labelling of the transaction process has a very high occurrence of false positives that means that the average consumer wants to make a purchase is going to get flagged as fraudulent. This is a big problem since they won't shop at a business again since they don't want to be flagged for no reason. Businesses have to invest time and money into an exhaustive training for employees to work on manual review which ends up increasing the time required to put an employee into service.
In its entirety AI can be an extremely useful tool to have in the financial sector, both with trading and business. It helps identify algorithms when trading stocks to help investors make more informed decisions on what will happen to stock prices however as we’ve seen over the past few days with market decline due to Coronavirus it is not a bulletproof solution but works 90% of the time. As with consumer fraud it can detect 90% of fraudulent transactions however also flags up a lot of legitimate transactions which may fit criteria as fraud.
Bình luận