Artificial Intelligence and Investing

Finance & Fury Podcast

Today’s episode of Finance & Fury we’re talking about Artificial Intelligence taking over the ETF and investment market. The discussion was actually started by one of our listeners, Gabriel who recently wrote in, so thanks Gabriel for sending this through.

Introduction; there are a few types of AI ETFs

1. Funds that specifically invest in companies involved in the development of products and services in AI
2. And alternatively, the one that we’ll talk about today: AI is running the ship - funds that use AI methods to select individual investments to buy within the fund

We will talk about:

1. The introduction of AI powered ETFs
2. What this could mean for the market
3. Options on what to do

AI ETF - AIEQ

1. Been around for about almost a year (Oct 2017)
2. Run by IBMs Watson AI - cognitive computing platform capable of answering natural language questions by connecting large amounts of data,
   • both structured (e.g., spreadsheets) and unstructured (e.g., news articles),
   • then, learns from each analysis it conducts to produce a more accurate answer with each subsequent question.
3. Computing called cognitive computing, where systems understand the world the way humans do. That is, learning how humans would think about the world: through senses, learning, and experiences. Watson continuously learns, gaining in value and knowledge over time, from previous interactions.
   • Example – Saw AI beat the world’s best players in complex online strategy and role-playing games
   • At first the characters would stand still, then run around in circles, then run into enemies and die, then after enough deaths they figure out they have to fight back, then figure out the best strategy and style
   • These are complex games and the players can make 300-600APM (almost 10 actions a second)
4. AI is programmed with parameters, that is to say, we tell it what to.
   • Investor demand - Equbot positions investment solutions based on in-depth analysis of investor demands.
   • ‘strive to create products that deliver positive long-term results on a risk-adjusted basis through optimizing proprietary investment research and trading models’
5. How will it do it? - drives an enhanced view of the global investment landscape.
   • The ability to process continually growing volumes of data and thread machine learning throughout our operations enables a truly unique investment process.
   • continually process, grow, and learn, just like our evolving investment technology
   • “We are processing more news information, more data around different countries,”
   • “We also added different capabilities like global macros and country risks, and integrated all the different modules to our platform.” Chida Khatua

How’s it doing so far?

1. Pretty well –
   • 6 months – outperformed the market ETF NAV by 4% - 12.34% vs 8.31%
   • Inceptions (11 months) – 19.3% vs 12.5% - 7% out return
2. Keeping its own at the moment and doing fairly well
3. Slow start – it was learning – matched the index
4. Diverged since then

Features and claims

1. The ETF will create a portfolio of between 80 and 250 stocks, choosing from more than 15,000 companies across the globe.
   • What does it include in the portfolio?
     • Learning to trade other ETFs and take advantage of arbitrage (making money for no risk)
     • Gabriel – Few points from a previous episode
2. ‘In a market cap weighted ETF, if a stock price drops, its market cap drops as well, so the ETF does not need to rebalance’ – Was talking about companies that fall in and out of the index – A company 300 in the index goes to 304 and is sold off (Sorry if I didn’t explain that properly)
3. Amount of market data processed is unmatched - over a million market signals, news articles, and 6,000 US companies analysed daily
   • A lot of noise at first – and there may be some big glitches
   • Can it tell the real from ‘white noise’ news? – Would it react to 40% drop in price news articles?
   • AI programs like this need to see a lot of information to be efficient and learn the best way of doing something
4. Automated data driven investment process that removes significant human bias and errors
   • End to irrational exuberance? - Bubbles
     • Thing of the past? Computers remove humans all together – acts rationally?
     • Worse in the future? What about if it works to manipulate a bubble to profit off it? Happens now – e.g. diamonds
   • Goes beyond the news and looks at funds to buy based on their metrics
5. Active management that combines fundamental, technical, and proprietary investment efficiency analysis to identify companies with high opportunities for long-term growth
6. Artificial intelligence and machine learning capabilities continually build upon the financial knowledge base driving an investment system that perpetually grows in value
   • What if it becomes ‘AI vs AI’? – Efficient market hypothesis becomes real
     • EMP – Market is efficient – Prices rise to high, smart investors will short sell
       • Assumes everyone acts rationally and that prices can’t diverge from their true value for long
       • This is where the basis of not being able to beat the market comes from (which is mostly right but obviously not 100% right)
   • If performance is the driving metric, returns might be gained through positive feedback loops
     • i.e. Buy a share, keep the price driving up to keep the returns going
     • Get enough AI traders working together then that can