/ milton

Who Is Milton Built For?

"The combination of man and machine is wonderful.  The process of man’s mind working with technology is what elevates us"
- Ray Dalio, Principles

Executive Summary

Milton is an AI driven tool that was designed for every investor interested in the US stock market to access 100% independent, data-driven, and objective views to help make better decisions.  This viewpoint can add value to the research process of any investor, both professional (folks on Wall Street) & non-professional (folks who don't work on Wall Street).  Why?  We believe every investor increasingly deals with the same two problems: information overload & information bias.  Simply put, if you love investing, independently spend time researching new stock ideas or trends, or generally do homework before buying or selling a stock, Milton can add value to your research.

If you have not done so already, we suggest you read Milton's white paper to learn how he works.  It's a little technical, but hopefully it provides you background on how Milton was designed to approach investing.

Background: Different Perspectives

The Non-Professional Investor

At Apteo, we understand no two investors are alike.  To illustrate, let's think of some very simple differences between three hypothetical individual investors.  Let's name them "A", "B", "C" below.

Obviously, these three investors couldn't be more different: their age, financial goals, and future obligations contrast wildly from one another.  As a consequence, their approach to investing should, and will, differ dramatically.  For example, Investor "A" may allocate most of her pre-tax 401K contribution towards stocks given her age, Investor "B" may prefer the peace of mind of a fixed-rate CD with a child heading to college soon, and Investor "C" may have a higher allocation to bonds to earn fixed income based on a well-planned retirement.  And just like all non-professional investors are unique, so are all professional investors.

The Professional Investor (aka "Wall Street")

Not only are professional investors very different from one another, it feels like the world of professional investing itself has become more nuanced.  We have long passed the day where machines account for more trading volume on the NYSE than humans (WSJ May 2017), and well-documented changes in investor behavior such as the move to passive investing (e.g., buy the S&P 500 index versus individual stocks) has expanded the demand for a host of new products such such as ETFs (WSJ Oct 2016).

Simply put, the business models, objectives, incentives, and inherent biases across professional investors differ dramatically, and feels more complicated than ever.  To put this in context, there are about 17,000 investment companies in the United States spanning asset classes (stocks, bonds, etc), including investors that solely invest in other investment companies (Winton Dec 2017; ICI 2017).  But that number vastly under-estimates the number of unique fund structures and bespoke investment products managed by these very different participants.  

To examine this further, let's use the same framework to look at another very simple example between three hypothetical professional investors that all focus on stocks (or more commonly referred to as "equities" in Wall Street parlance).  Let's name these three professional investors "X", "Y", "Z" below.  

While the language might seem a little more intimidating in this exhibit versus our last example, the concept is just as simple.  At the most basic level, these money-managers are all "stock market" investors, but their business models & incentives are quite different.  Investor "X" may be a "market neutral" fund (AQR Apr 2015) that focuses on small-cap stocks with market caps below $500 million, whereas Investor "Y" is a "long-only" investor that holds no short positions, and focuses solely on US and international stocks with market caps above $10 billion.  As a consequence, Investor "X" and Investor "Y" probably spend time researching totally different types of stock ideas and have different benchmark indices for future investors to assess whether they've performed "good" or "bad" in the past.  Meanwhile, Investor "Z" has a totally different business model, since they are solely assessed on how close their performance mirrors the underlying performance of their benchmark index (the S&P 500 in this example), irrespective of the benchmark's direction (MarketWatch Apr 2018).

The Common Thread Between Investors

So we've laid out two very simple examples of how investors are different.  But just because you are a professional investor versus a non-professional investor, that does not preclude the possibility of real similarities between these unique market participants.  At the most fundamental level, especially in the context of speaking about Milton, we know that most investors have a substantial allocation to the US stock market.  In fact, whether you are the pension system of California (CalPERS) who manages nearly $325 billion of public retirement benefits and dedicates nearly 50% of its assets towards equities (Pension & Investments Dec 2017), or you are a 25 year-old that uses the old rule of thumb of taking 100 minus your age to determine the correct equity allocation (75%) for your pre-tax 401k contribution (Investopedia Jun 2014), almost every investor has exposure to the US stock market in one way, shape, or form.

But let's not stop at that fundamental similarity: anyone who has read a financial publication knows there are well-defined colloquial terms we've created to bucket investors with different investment preferences.  For example, we've come to refer to 'income' investors as those who love dividends paying stocks, 'growth' investors as those who love the prospect of finding the next big technology stock, 'value' investors as those who love finding the unsexy hidden gem that others have written off, 'experiential' investors as those who invest solely on their own purchasing habits, and 'technical' investors who rely on charts.  In addition to these different preferences, investors have totally different holding periods.  While Warren Buffett ideally likes to think about a stock over "forever" (WSJ Feb 2017), its important to note the average holding period for a stock on the NYSE is less than 4 months, according to some estimates (Politifact Jul 2016).  While people have different interpretations of the true "definition" of these preferences and holding periods, these "schools of thought" generally cross-over between professional & non-professional investors.

But the similarities go further than these more traditional classifications.  In fact, some of the largest investment companies in the world, such as Blackrock and Vanguard, and new entrants such as Betterment, have developed widely used and relatively low-cost products to cater to non-professional investors who want to emulate some of the practices used by professional investors to invest in financial markets.  These products include the development of structured products for people to invest by sector (e.g., prefer tech stocks versus autos), invest by "market factors" (e.g., prefer "momentum" stocks versus "value" stocks), invest solely via low-cost ETFs (or "robo-advisors"), invest based on geography (e.g., US versus Europe), or company size (e.g., large cap versus small cap).  So while there are distinct differences between professional and non-professional investors, there are also distinct similarities.  This is what creates the beauty of the stock market: so many different participants, each with a unique set of objectives, all working in tandem to achieve their goals.

Our North Star and What Milton Can't Do

So ironically, the common thread between all investors is that we all have a truly different perspective, based on our own set of unique circumstances and objectives. The team at Apteo believes this is a very simple but extremely important guiding North Star when we build AI tools for investors, like Milton.  Milton is not built to tell you how to invest money on behalf of yourself, or your clients.  To that end, we are not an investment advisor and make no recommendations to invest in any security.  Rather, we built Milton to help give you a powerful tool that can help you make better decisions based on your own set of unique circumstances.  Since we don't invest capital on behalf of ourselves, our employees, or clients, we believe Milton's perspective is 100% independent and gives him more flexibility in delivering objective & data-driven insights to you without conflict.

Let's expand on this, because it is an important core principle when we developed Milton.  Milton wasn't designed to evaluate whether its better to invest in individual stocks or passive vehicles like ETFs (FT Oct 2017), evaluate whether certain investment styles may or may not see brighter days in the future (CNBC Oct 2017), evaluate the performance of a class of professional investors (AQR May 2018), or make recommendations about the optimal asset allocation strategy (Schwab 2014).  These are all extremely complex questions that have been explored – via academic research and real-world practical experiments – by some of the largest, most prominent, and well-respected investors and institutions globally.

These debates will likely continue for years to come, and some may likely never be resolved since markets are so dynamic.  While we believe all of these questions merit lots of thinking and applied research, Milton was not built to answer these questions.  Ultimately, it goes back to our guiding North Star principle when developing AI tools: investing is a very personal decision, and we want to help you invest better.  To that end, we believe every investor will increasingly need access to data-driven tools like Milton to deal with information overload and bias.  

Who Is Milton Built For?

So now that we've laid out what Milton can't do, let's clearly state what he was designed to do: Milton was developed to provide sophisticated AI-driven insights to any investor – professional or non-professional – to help make better investment decisions.  As of today, Milton focuses on providing insights exclusively for the US stock market.

As reviewed in our white paper, Milton generates lots of insights on a daily basis (to be more specific, he generates about 30,000 daily forecasts over different time intervals for a large universe of publicly-traded US stocks).  And we are constantly trying to improve Milton: he reads new things every day, ingests more data and new types of data, and we work hard to make him better based on a continual process of research and development.  But based on nearly 2 years of research & development, and over 12 months of real-world testing of his stock price forecasts, we have identified three core areas where Milton can help every investor:

  1. Screening for new stock ideas
  2. Better understanding sector trends
  3. Doing confirmatory due diligence when researching an idea

Based on our work thus far, we believe Milton is better suited to evaluate certain sectors and stocks than others.  Therefore, we are taking a data science approach to evaluate Milton's insights, integrate our learning back into Milton's design & engineering, and repeating this process.  However, we have found many instances where Milton is able to generate a valuable insight that may be helpful to any one of the investors we described earlier.

Case Study: How Milton Can Help

To help illustrate, let's use an actual case study from September 2017.  As written in the white paper, Milton generates a daily report with stock price forecasts for a universe of tickers.  To simplify this even further, think about this report as a simple list of stock tickers rank-ordered based on Milton's view of his favorite stocks to his least favorite stocks.  We, the team at Apteo, then spend time tracking how Milton does at ranking stocks over time to understand his strengths and weaknesses, with the hope of continually making him better.  One effective way to evaluate Milton's work is to look at how his "ranking" of stocks changes over time.  So let's look at a real example from a few months ago (Q3 2017).

Situation: Best Buy (NYSE: BBY)

August 2017:  Best Buy (NYSE: BBY), the publicly-traded US-based consumer electronics retailer, saw a significant decline in its stock price between the time it reported Q2 2017 earnings (August 27, 2017) and late September (September 20, 2017).  During this time frame, Best Buy stock declined from $63 to $53 while the broader S&P 500 rose and a broad barometer of US retail stocks (XRT) also rose.  To put this in perspective, Best Buy underperformed the S&P 500 by nearly 18% during this time frame and underperformed the US retail ETF (XRT) by over 20% during this time frame.  This is fairly significant underperformance, and we spent time evaluating what happened to Best Buy between this time period as part of our ongoing data science efforts.  

Based on our evaluation, we came up with three general explanations for this stock price decline: (1) despite meeting analyst expectations for Q2 2017 results, management downplayed future performance on the conference call with investors (Seeking Alpha Aug 2017); (2) management signaled that margins could be pressured into the important holiday 2017 season (CNBC Aug 2017); (3) and concerns about Apple's new slate of products (scheduled to launch in September 2017) was a negative over-hang on Best Buy (CNBC Sep 2017), which obviously relies on suppliers like Apple to come up with exciting new products to drive customers.

What Did Milton Think?

Before we evaluate Milton's results, we look at how Milton has done forecasting a stock in the past, and take a data-driven approach to understand if his insight may be statistically more probable or less probable of being accurate versus Milton's entire universe of stock forecasts.  This is done in parallel with the work process described earlier.  In this instance, we saw that our prior forecasts on Best Buy had been promising, and therefore, we wanted to understand how Milton would react to this significant stock price decline.  While stocks go up and down every day across every sector, we see Milton react differently to different situations.  For example, sometimes a stock will go down and Milton's ranking for that stock will go up, and sometimes the converse is true.  But during this period, we saw Milton become more bullish on Best Buy and his rankings for Best Buy consistently moved higher relative to his entire universe of stocks, and by late September 2017, Best Buy was a top 10 ranked stock (top 1% of forecasts) in Milton's universe of stocks where he looks for opportunities that have significant absolute upside in the US stock market.

What Happened?

Ultimately, Best Buy stock more than recovered its losses from this period as it rallied from $53 to $69, or nearly 27%, from September 20, 2017, into year-end 2017 (December 31, 2017).  This validated Milton's forecasts and by any measure, this was a large absolute rally in the stock with Best Buy outperforming the S&P 500 by 20% and the retail ETF (XRT) by nearly 13% during this time frame.  This rally, while multi-faceted, was driven by a few different dynamics: (a) continued earnings revisions higher by Wall Street analysts despite the move lower in the stock, (b) improvements in investor sentiment towards Best Buy as concerns about Apple's new product cycle abated, (c) broad sector rotational forces drove fund flows into consumer stocks, and (d) the  2017 US Federal corporate tax cuts that benefited relative high-tax payers in the US such as Best Buy (BBY 2016 & 2017 effective tax rates north of 30% versus approximately 24% for S&P 500 prior to tax legislation).

Why Does It Matter?

We did not give you this example to flatter Milton, although we agree it shows him in a pretty positive light.  Rather, we wanted to show an example of how Milton's insights can be used by every investor, irrespective of your individual perspective.

To illustrate, (a) if you are a non-professional "income" investor who loves dividends, you might have wanted to look at Best Buy at $53 since it had nearly a 4% forward dividend yield, (b) if you are a professional "value" investor, you might have wanted to look at Best Buy at $53 since it was trading at north of a 10% free cash flow yield on 2017 estimates while still growing pre-tax earnings, and (c) if you are a large hedge fund that was short Best Buy stock into Q2 earnings, you might have wanted to lock-in profits and move onto the next relative short idea.

This example is powerful because it shows that you can use Milton's insights to help you invest better, even if you agree or disagree with Milton's insights.  But to be clear: Milton is certainly not always right, and is often wrong.  But that is why we work hard to curate Milton's insights for you based on the data-driven approach described earlier.  Simply put, Milton is another tool in your investment toolkit to help you make better decisions.  On our end, we are working very hard to make Milton better every day.

But Wait... Why Was Milton Right In The First Place?

We will walk through some more detailed examples like this one in future posts to help illustrate ways Milton can help you.  We will also dedicate significantly more time trying to explain how Milton comes to a conclusion that leads to a valuable insight, like in the example above.  As Warren Buffett famously said, "in the business world, the rearview mirror is always clearer than the windshield."  Simply put, its easy to understand why Best Buy's stock rallied during Q4 2017 in hindsight, but it is difficult to truly understand why Milton's forecasts for Best Buy were so bullish.  This is a real challenge for a small AI technology company like ours, but we are working hard to improve our research in this area specifically.  Understanding how AI technologies "think" is a massive challenge and topic in of itself.  Therefore, we want to dedicate an appropriate amount of time to help you understand how we approach this valid question.


"The combination of man and machine is wonderful.  The process of man’s mind working with technology is what elevates us"
- Ray Dalio, Principles

For our conclusion, we will go back to the start.  Simply put, as investors, we can all benefit from a different perspective.  Especially in a world where data growth is exponential and we all face information overload and information bias, we think AI driven tools can help informed investors make better decisions.

If you have any questions or comments, feel free to reach out directly.

Thanks for the time.

-Manan (manan@apteo.co)

About Apteo
Apteo, the company behind Milton, is made up of curious data scientists, engineers, and financial analysts based in the Flatiron neighborhood in New York City.  We have a passion for technology and investing, and we strongly believe that investing is one of the most reliable and effective ways to build long-term wealth.  We build AI tools to help informed investors make better decisions.  

To learn more about us, please reach out to us at info@apteo.co, join our mailing list at milton.ai, or subscribe to Milton’s blog at blog.milton.ai.

Apteo, Inc. is not an investment advisor and makes no representation or recommendation regarding investment in any fund or investment vehicle.

Who Is Milton Built For?
Share this

Subscribe to Milton's Blog