by Brian DeChesare Comments (6)

Artificial Intelligence-Based Fintech Companies: The Best Startup Exit for Bankers?

Artificial Intelligence Fintech Companies

If you go into finance, will you be replaced by machines?

It depends on your specific role, but quite a few positions lend themselves to automation.

You could fight back by finding roles that are less likely to be automated, but another approach is to embrace these changes and teach yourself the skills – programming – that let you automate your job.

And one of the best ways to do that is to move from a bank to a fintech startup focused on artificial intelligence (AI).

If you have a business and math/science/programming background, you might be the ideal fit for this exit opportunity – as our reader today found out:

The Rise of the Machines: Getting Interested in AI

Q: How did you become interested in AI and fintech?

A: I came from a family of programmers, so I always knew something about technology.

But I never thought of tech as a “career path,” so I was lured into Wall Street right out of school and worked at a boutique investment bank in London.

I had always thought that artificial intelligence might revolutionize industries and automate our daily tasks, but my banking experience made me even more convinced of that.

I saw firsthand how many of the accounting, finance, and legal tasks at banks might be automated, and I decided to reach out to data scientists and AI/fintech startups to enter the industry.

I won an offer at a well-known fintech startup and joined the sales & marketing team there.

It wasn’t the traditional exit opportunity, but I also wasn’t the traditional banker.

Q: I can relate to that one.

For those who aren’t familiar with the terminology, can you explain what “artificial intelligence” and “machine learning” are and how they’re related to automation?

A: Sure. To explain these concepts, I’ll use the example of finding undervalued companies.

Automation means using a computer to handle tasks that humans might have done before.

For example, a program such as Capital IQ that scans public companies’ filings and retrieves data based on keywords and rules is “automation.”

It saves you time because it can retrieve each company’s revenue, EBITDA, and other financial stats, and you don’t have to comb through the filing manually.

Artificial intelligence (AI) goes beyond that by making more “human-like” decisions. Programs based on AI learn and improve their performance based on market events and results.

For example, an AI-based program could look at a set of public company filings, share prices, and other data, and make a recommendation on which companies might be undervalued.

It would do that by looking at metrics such as the growth rates, margins, cash flow, and valuation multiples, and applying “rules of thumb” to determine if the current valuation is off.

It tries to replicate the same thought process humans use when making this decision.

Algorithms can complete this process much faster than humans, so they can scan huge sets of data and make recommendations in seconds.

But humans can still spot things that algorithms struggle with, so this exercise won’t always produce the correct results.

Based on the performance, programmers would tweak the algorithm to account for other factors.

For example, it might turn out that positive sentiment around a stock was incorrect because some key commentators had a strong bias.

Based on that, programmers might add some “bias detection” when the program scans for news and commentary.

Machine learning is similar, but it’s a subset of AI where the program is “trained” using large amounts of data.

So, it’s less about manually coding specific algorithms and more about giving a program a lot of data and teaching it how to perform a certain task.

Nvidia has a good summary of the differences here.

Q: Great, thanks for that overview.

How are finance firms using AI and machine learning in day-to-day tasks?

A: Currently, automation is far more common than actual AI and machine learning, at least in roles such as investment banking and private equity.

You see more advanced technology with quantitative investment funds that make decisions based on math and statistics.

Quant funds already have teams of scientists and mathematicians, so they don’t use the products/services of AI-based companies for “finished products,” but rather for the data feed.

These funds need to analyze a huge amount of data, so AI is essential because it lets them determine what’s relevant and what’s not.

For example, if the fund is using a strategy based on the correlation between commodity prices and stock performance in emerging markets, they need to sift through a massive amount of data, large portions of which might be useless.

But an AI solution might sift through the entire data set and say, “Only the trends for gold and silver prices relative to stock performance over these 30-day periods in Markets A, B, and C seem relevant; here’s the smaller, relevant data set for you to analyze.”

And then the fund would apply their statistical filters and algorithms to this filtered data.

If you were to compare AI/fintech to the oil & gas industry, it would look something like this:

  • Oil: The data feed or “raw materials.”
  • Drillers/Refineries/Gas Stations: AI/fintech companies that turn the raw materials into something useful.
  • Gasoline: The investment recommendation or trading strategy that results from finding the right data and processing it appropriately.

Many quant funds are their own “refineries” and do not use much fintech, but other funds do find value in accessing filtered and thoroughly back-tested data feeds.

How to Join the Machines: The Top AI Fintech Companies

Q: Thanks for that analogy and explanation.

What are the most well-known fintech companies in this space, and what products/services do they offer?

A: There are dozens, or maybe even hundreds, of companies in this space, and it’s still very early (as of 2017-2018), but the broad division is between the huge tech companies and private startups valued at less than $1 billion (and often below $100 million).

The big tech companies like Google, Facebook, Microsoft, IBM (Watson), and Nvidia all have elite AI research teams that work on a wide range of problems, some of which are finance-related and some of which are not (e.g., self-driving cars).

Among tech startups, here’s a quick summary of a few companies that might be poised to disrupt industries:

Sentient Technologies – This is an SF-based AI company that works across industries and created an investment subsidiary (Sentient Investment Management), with an equity fund led by CIO Jeff Holman, formerly of Highbridge.

Vicarious – This one is more of a general AI company, but the CEO has said the goal is “actual human intelligence.”

Quantopian – This company wants to open source data and let users create algorithms to compete against those of other quants; it’s like “gamified stock trading.”

They’re also going to provide backtests of the winning algorithms so that investors can put their money to work with the best ones.

If the algorithms do well, the developers will receive commissions.

It’s free to developers, and there are no barriers to entry, so even high-school students could learn how to code and create winning strategies.

Steven Cohen from SAC Capital / Point72 fame committed $250 million of his capital to be managed by this company. The results have been mixed so far, but it’s still quite early.

Trade Ideas – This one has been around for ~15 years. The company created an AI bot called “Holly,” which uses 35+ different concepts and strategies to create new algorithms, test the markets, and learn from its performance.

The company caters to small-cap hedge funds and a few larger ones, and it has seen exponential growth because Holly was up by 89% in 2016 (and it outperformed the market in 2017 and 2018 YTD as well).

It correctly reacted to “black swan” events in real time, including the Brexit vote, when she decided not to make any trades.

Accern – This company focuses on news analysis and pulls data from over 300 million sources, including both verified and non-verified ones.

The company’s AI bot “Mommy” identifies and categorizes tweets and news stories related to public companies based on exposure, impact, and reliability.

It’s able to analyze random tweets and obscure news sites and determine how reliable the information is, often predicting with a 90% success rate the chances of the story being published by a legitimate media company.

Running Alpha – This company was founded by a former physicist who had studied tornadoes and weather patterns before becoming a trader in Chicago.

He uses a physics-based approach to beat markets, and it is so technical that I won’t even attempt to explain it here, but it involves crowd physics and feedback loops.

Dataminr – This is one of the most well-known companies in the space, partially because of government interest (CIA investment arm In-Q-Tel invested) and its quick move to a Fidelity-backed startup with a $1 billion+ valuation.

Early on, the company secured a data deal with Twitter that let them analyze all tweets in real time; government agencies were also interested because of the potential to create terrorist watch lists.

The company also serves other industries with a similar analysis, and it offers real-time alerts on public companies.

Q: Thanks for that run-down.

If you want to work at one of these companies, what should you expect in the recruiting process?

A: The best path to executive leadership in fintech/AI is to pursue a degree in business, particularly finance, and also take courses in programming, data science, data management, and artificial intelligence.

Work experience in the finance industry helps, as you’ll understand the needs of traders/bankers/research analysts in more detail, but you also need a basic knowledge of coding.

People in pure tech roles skew heavily toward Ph.D.’s in data science, computer science, neuroscience, physics, and engineering from Ivy League universities and other “target schools.”

Everyone has a strong foundation in coding, and Python expertise is heavily favored.

There is a huge demand for talent and limited supply, so it’s not like applying for IB/PE roles where you’re competing against 500 other candidates, and it’s impossible to stand out.

It’s so informal that you could go on Angel’s List, look up AI/fintech companies, and connect directly with investors and founders to ask about internships and jobs.

Interviews tend to be based on your industry knowledge and side projects.

If you can walk in and show them evidence of AI-related programs you’ve worked on, especially if they’re related to finance, you immediately have a huge leg up.

Open source contributions on Github and similar sites are also helpful.

If you’re interviewing for non-tech roles (e.g., business development or sales), your coding knowledge is less important, but you’ll need to know the space in-depth and explain how you’d go about contacting prospects, winning deals, and so on.

Q: Great, thanks for explaining that.

What has the experience at your firm been like so far?

A: It has been very positive so far. I focus on client on-ramping (i.e., helping new customers set up our software and start using it).

That usually happens on a monthly or bi-weekly basis, and my job is to figure out what the clients’ requirements are and make sure our services deliver.

I’d say it’s a 50% / 50% split between client acquisition/on-ramping and “internal business issues” that scale up or down over time depending on what’s going on.

The relationship between the business and tech sides varies at different firms; sometimes they’re well-integrated, but at other firms, they might be separate, with interdepartmental exchanges on a daily, weekly, or bi-weekly basis.

That happens because the developers need significant time to build the solution, even if the business team can speak with customers and quickly determine their requirements.

The culture varies quite a bit as well, with more informal workplaces in California and more of a “corporate casual” feel in NYC. But even there, it’s still far more casual than traditional finance firms.

Q: Great. Thanks for that overview, and for your time!

A: My pleasure.

M&I - Brian

About the Author

Brian DeChesare is the Founder of Mergers & Inquisitions and Breaking Into Wall Street. In his spare time, he enjoys memorizing obscure Excel functions, editing resumes, obsessing over TV shows, traveling like a drug dealer, and defeating Sauron.

Break Into Investment Banking

Free Exclusive Report: 57-page guide with the action plan you need to break into investment banking - how to tell your story, network, craft a winning resume, and dominate your interviews

We respect your privacy. Please refer to our full privacy policy.


Read below or Add a comment

  1. Hey Brian,

    Followed your blog for a bit and this AI/Fintech post is really interesting. I’m an undergrad student at a target school looking for a summer internship as my summer class just got cancelled. I have previous internship experience in wealth management & previously worked for Google, along with AI laboratory and economics research under my belt from my time at school. I saw the list of firms you posted, but they seem to be sparse regarding contact details. Regarding reaching out for a quick gig this late do you have any suggestions or know of any startups or experimental hedge funds that would be interested? If you have any advice feel free to reach out via email, it would be greatly appreciated. Also you have quite the niche with this website, great job mate!

    1. I don’t, sorry, but if you can find their contact information, you should definitely reach out and use one of the templates on this site in the networking articles if you need an example to follow. One tip is to track down these companies at events or conferences, if you can get in, rather than applying online since the signal:noise ratio is much better.

  2. Hi Brian,

    I have the goal to start my own distressed boutique private equity firm some day down the road. I’m a little unsure about what direction to go. I am currently a public accounting auditor with experience in a variety of industries and my CPA.

    I have several job opportunities to get out of public accounting. One is an accounting/financial fp&a role at a large manufacturing firm for aerospace and defense. the other is a private market advisory role as an analyst. I would gain a lot of exposure to pe funds, but it isnt really a role that would be due diligence for potential investments. i believe that would be a small part of it. I think a lot of would be operational stuff for PE funds.

    would either one of these two work? I imagine from there I would spend 18-24 months, maybe try to transition to a restructure/turnaround advisory firm, go to a top 20 MBA, and see what opportunity is out there after MBA to help get me to my goal. would I be better off taking a role like one of these or wait for a more advisory/due diligence type role?

    1. I don’t think either role would be that helpful. FP&A is pretty far removed from PE, and so is market advisory if you don’t work on deals or potential investments. You need to get some non-accounting experience ASAP, so the best bet is to move into the valuation team if your firm has one, or move to an independent valuation firm, or do something else where you’re working with forward-looking numbers. Then, either move into IB directly from there, or go back for an MBA and move into IB/PE afterward (along with a pre-MBA internship).

      Another option might be to aim for roles in restructuring consulting ( because they do hire people with accounting/corporate finance backgrounds as well, and you can move into distressed PE from there.

  3. This is very interesting.

    How long until the capital markets are nothing but dueling computers? And how long until low(er) level programmers are replaced?

    Do you know of any organizations that have applied an AI/Machine Learning approach to value investing?

    Sounds like the path to riches is ownership of a proprietary data feed…

    1. I think we’re already there. Computers already make up 50-60% of all market trades in the U.S., sometimes up to 90% in volatile periods. That’s why there are so many fast swings based on momentum now.

      RE: AI/machine learning applied to value investing, maybe this firm:

Leave a Reply

Your email address will not be published. Required fields are marked *