If it really works, why not show it?

Posted by Mike Shell on January 18, 2012

Someone sent me an interview with an investment advisor who promotes a passive approach of asset allocation to indices with periodic rebalancing. He claimed that rebalancing index funds and asset allocation does work.

We have run every asset allocation mix you can imagine of all kinds of low cost indices and we have been unable to find any combination of any indices that would even come close to our investors' objectives for total return and within their tolerance for drawdown.

If we could, we would be doing it, too. The fact is, we are more aware of the current state of various world markets and we know that if we pursue asymmetric investment returns, a risk/reward ratio within our objectives, we necessarliy have to actively manage risk and rotate, not allocate. 

The person being interviewed is a writer but also owns an investment advisory firm that offers his passive indexing asset allocation and rebalancing. It's pretty fascinating that they charge a fee for it. If I were going to asset allocate to low cost indices, you can rebalance them at discount brokers like Schwab, Fidelity, or Interactive Brokers for free.

With all that said, my real ponder is: Have you ever wondered why an investment advisory that offers asset allocation models doesn't publicly disclose its performance? I mean, no matter how many asset allocation models they have, they can create performance composites that include every account in each model and present a composite of their performance. Doesn't that seem strange? That an investment advisor managing $1 billion would say that an asset allocation had averaged X% the past several years but doesn't provide evidence that the billion dollars he manages averaged that much?

This same investment advisor routinely criticises "active" mutual fund managers. Much of the time, I actually agree with him: relative performance mutual funds aren't impressive as a group. I too prefer exchange traded funds based on a transparent quantitative systematic index. The difference is that I rotate, instead of allocate. I have nothing against those who promote passive index asset allocation and rebalancing. The academic research they believe supports a buy and hold approach provides me with a reminder that tactical rotation isn't easy, requires real skil, and most will be unable to develop robust systems and execute them. As long as they rant against it, maybe fewer will try, allowing those of us who do to keep doing it. Any good engineer who develops something will spend energy afterwards trying to figure out what may go wrong, break it, with the aim of improving it. To do that, I spend far more time reading those with the opposite view of my own as I do those who think they are doing what I do.

The "passive index asset allocation and rebalancing" mantra would seem to be more believable if we could see their actual performance history. 

 

What is technical analysis and how is it related to quantitative trading systems?

Posted by Mike Shell on January 03, 2012

Technical analysis is the forecasting of market prices by means of analysis of data generated by the process of trading.

Some technical analysts say they don't actually "forecast" market prices with their technical analysis methods. However, I suggest for a buy or sell signal to be useful, it must necessarily have predictive ability. That is, if you buy because your technical indicator defines a price trend as "up", then you necessarily believe the price trend will continue to go up: it has predictive ability. Otherwise, there would be no use in applying the signal. The reason for applying the signal is you believe it has predictive ability: it can signal a rising or falling price trend.

Technical analysis is based on the belief that markets discount everything except information generated by market action itself, therefore, all you need is data generated by market action. In some technical analysis literature, this seems to be in direct conflict with the Efficient Markets Hypothesis (EMH). Efficient Markets Hypothesis is the theory that markets discounts and prices in information, so there is little to know possibility to find mis-priced securities. EMH also theorizes that price trends do not exist- prices are random. This seems in conflict because technical analysis and quantitative analysis studies showing that the market punishes and rewards certain factors. Behavioral Finance studies show that markets under-react and over-react to information - it's not so efficient. Because of under reaction and over-reaction, prices drift in one direction or another over time, or prices overreact and move too far directionally.

Why is technical analysis so popular? In his Introduction to Technical Analysis, Martin Sewell says one reason is Communal Reinforcement. Communal reinforcement is a social construction in which a strong belief is formed when a claim is repeatedly asserted by members of a community, rather than due to the existence of empirical evidence for the validity of the claim. I think that's important to understand about any belief you may have. We could say the same for many different forms of analysis: most of it is based on Communal Reinforcement: people believe it just because they are told to, even though there is no real evidence supporting it. And, if you notice, there are people who believe just as strongly in passive asset allocation as there are people who believe in some form of fundamental or technical analysis. Each of them can bring to the debate hundreds of academic papers, independent studies, back-tests, or even another persons performance history. But at the end of the day, the determining evidence of a real edge is their actual performance - and that's not a perfect glimpse into the future.

I believe there is an advantage to "some" technical analysis, but that is only true for the forms of technical analysis that is quantifiable and objective. Technical analysis that can be quantified are signals that are objective - they can be stated as an equation, programmed with computer code, and tested historically to determine statistical significance to see if the indicator, signal, or complete system has an edge. Many forms of technical analysis do not offer the objective signals that can be quantified into a trading system and tested. The subjective forms of technical analysis, such as some of the "chart patterns" found in Candlestick charting or Point & Figure charting, are too subjective to be testable. The users of chart patters often instead rely on some guru. You'll hear them quote what someone wrote in a book 50 years ago about some price pattern formation. Rather than the solid foundation of mathematical expectation, they rely on blind faith. Some guru said when a price pattern forms a certain pattern, the price usually does this or that. Yet, they don't have any specific data behind it. How often does it go up or down and by how much? Nearly all of these kind of technical analyst don’t have any evidence their methods have a positive mathematical expectation; profits exceeding losses. In fact, many of them can’t use their own performance history as evidence their technical analyst methods have any advantage or that they are able to exploit it. There are few things more fascinating that people following a guru without even knowing the guru’s track record. That is fascinating if you realize the importance of evidence-based systems and decision-making.

I can tell you, with empirical evidence supporting it, that the potential advantage gained from technical analysis is from applying price trend indicators objectively in quantitative trend systems. That is, applying buying and selling tactics with a known probability and expectation based on evidence -not only evidence from back-testing but a forward-walk and actual performance. This is where technical analysis transitions to quantitative trend systems.