Strategies | AI - Artificial Intelligence Investing

21st Jan '16 by roddy10

3 comments 3147 reads

Following are the opening paras from an article in the FT today:

Nearly three decades ago, Warren Buffett penned an annual letter to his adoring investors mocking an insidious new trend beginning to make its mark in the money management industry: the computer.

“In my opinion, investment success will not be produced by arcane formulas, computer programs or signals flashed by the price behaviour of stocks and markets,” he scoffed in his 1987 annual letter. But the chairman of Berkshire Hathaway, famous for his long-term strategic thinking, may have short-changed one of the great forces reshaping investment.

(Source:http://www.ft.com/cms/s/0/5eb91614-bee5-11e5-846f-79b0e3d20eaf.html?ftcamp=crm/email/2016121/nbe/TradingRoom/product#axzz3xsiIL7FL )

I thought it would encourage some discussions.

I am sure Ed would make the point that Stockopedia ranks should really be a launchpad to do more work but I think it is worth discussing how people use the system....

Unlock with Email

Unlock with Google

Already have an account?
Login here

Novice Investor

21st Jan '16

No, Ed wasn't making that point at all! He wasn't specifically saying you shouldn't, but he has been suggesting you can pick shares for your portfolio by using his stockranking system. The only input is designed to ensure diversity.

NI

Edward Croft

22nd Jan '16

Novice - what I have always maintained is that portfolios constructed to be exposed to certain factors (especially quality, value and momentum) have historically outperformed over almost every 2 or 3 year cycle in most markets. Constructing portfolios exposed to the same factors may outperform but only if the historic correlation continues.

Now I've thought and read very deeply about whether the correlation will continue - if it doesn't then the StockRanks will be ineffective. Obviously the whole world has known about Value Investing since biblical times (or at least since Ben Graham) but it still works. Surely if markets were efficient it should have been arbitraged away?

I think the reason it hasn't is the fact that the value effect is a form of risk premium. There are reasons why some investors don't want to own value shares, so those that are willing to own them are justified in earning a premium for doing so. The reason for that premium may be due to greater risk (e.g. bankruptcy or credit risk), behavioural reasons (nobody likes owning out of favour stocks) or structural reasons (e.g. institutional factors) or some combination of all three. What we do know is that the premium has continued to exist in spite of everyone knowing about it. Just as there's an equity risk premium over bonds, there's a value premium & a momentum premium over general equities.

The StockRanks are structured to be exposed to these risk premiums. What subscribers using them should be aware of is that there are other subscribers using the same system. There may be more crowding in certain shares as a result and more volatility in those shares than previously - especially in small caps. On this note I recommend reading this from Cliff Asness. The StockRanks have had a fantastic run to date, but I've always maintained there will be long periods of underperformance. It's to be expected. I think it will be fascinating to see how these factors perform over the next 18 months given we're in a serious correction.

Regarding computational investing. I wouldn't take Buffett's 1980s words as gospel. We're 20 years on and our understanding of risk and markets has improved. Insurers can make consistent returns using computational models of risk and using diversification. Investors have proven that they can do the same - earning the risk premiums in the market using similar strategies.

The future will not be the same as the past, but it ought to rhyme.Warren Buffett picked up on the tendency of good, cheap shares (QV) to outperform long before everyone else. Buffett-esque returns are likely unachievable nowadays because literally every finance graduate has been drilled in how markets work. That doesn't mean that investors can't outperform the market by investing with the same principles, it's just the returns may be more moderate.

lucien2k

There are some interesting things going on in this space.

Google has recently released an open source product called Tensorflow, which they are pushing as their machine learning engine that powers search, ads etc for them. Hard to argue with a system that powers $66bn in revenue.

There are also sites like Quantopian that allow you to create backtest trading algorithms using real datasets.

The potential is certainly out there, and it doesn't require massive backing. I just wish I had more time to work on it myself!

AI - Artificial Intelligence Investing

Unlock the rest of this article with a 14 day trial

About the Author

Roddy10

3 comments

Unlock the metrics for all these stocks and more...

Unlock the metrics for all these stocks and more...