The Human Compulsion to Seek Short-Term Patterns When Investing

One of the details I always comment on when I am reading articles on Seeking Alpha is the inclusion of 'technical analysis' or other types of short-term price prediction based on patterns. It takes a lot of different forms, some as simple as merely looking at charts (i.e. past prices of the stock graphed with time on the x-axis), and others slightly more sophisticated (read: 'mathier'). I have seen citations including Fibonacci retracement, Bollinger Bands, and of course the common 50 day and 200 day moving averages, all interpreted to derive many a different conclusion.

Typical technical analysis mumbo-jumbo

Moving averages can be informative for one's first cold look at a stock: comparing them to the security's current price more or less gives you the market's impression of the company or security. This information can also be generally gleaned from the price relation to 52 week highs and lows and stock analyst buy/sell ratings.

Beyond that, however, I see technical analysis as market tomfoolery. It is an attempt to see patterns in short-term price movements, that depending on which theory you subscribe to, can be more or less random.

The random walk theory is the classic economic perception that because markets are efficient, prices are going to follow a random walk. Here is some analysis that attempts to separate the efficient market theory from the random walk theory. In essence, the idea is that prices are explained by successive random steps in any given direction. The third link above is important because I think it reconciles the fact that a random walk can only coincide with efficient market theory if the random walk is to be short-term noise while eventually leading to the efficient market price. This of course would not support a very rigorous version of the efficient market hypothesis (EMH).

While I subscribe to a weak version of EMH, I do not necessarily believe in the random walk theory, at least in the long term. I do believe that it could explain bubbles and short term market mispricings, but in my opinion these could be better explained with behavioral finance or simply by variance around an 'accurate' price. I find behavioral finance to have more explaining power due to a presumption by a great deal of the statistical analysis in investing that markets are 'cold' and that decisions are being made by efficient automatons and not by humans (Nudge: Improving Decisions About Health, Wealth, and Happiness provides an excellent framework for understanding the difference, except where I use 'efficient automatons' they use 'econs'). Behavioral finance provides a way of explaining mispricing in the market based on human emotion and the way we perceive.

Does that mean that market participants using statistics have not been successful? Of course not. One of the best success stories I have heard of is the hedge fund Renaissance Technologies, which according to their wikipedia page has averaged a 35% annual return after expenses and as far as I know has never had a losing year. Because hedge funds are such black boxes, it is difficult to understand how Renaissance is truly making money and I am not sure how to address them as a phenomena in the short-term trading sphere. Note: there are also a large number of high-frequency trading (HFT) rigs that use some sort of fundamental or technical indicator, but I have yet to see conclusive evidence that technical trading like this can make money over a longer time period.

The question will always be whether it is not the pattern but rather some fundamental idea being observed through the pattern that is what is making money. Instead of simply finding more complicated mathematical techniques to observe the pattern as it is seen through market prices, perhaps a more effective methodology would be to understand what is causing that event.

On the frontier of simply statistical analysis, Bruce Babcock suggests that over the longer term markets trend after you have looked past short-term noise, and suggests using chaos theory to understand it. This to me seems more plausible than patterns in the short-term, however once again I think it could be better explained by behavioral finance.

Considering the relatively small amount of market data we have to pull from to make statistical assertions lends one to reject a 'patterns for the sake of patterns' investment style. Sample size limitations and a lack of fundamental reasoning for why prices should behave in any given pattern leads me in the end to reject the notions provided by Babcock. While it could be successful as a trading strategy, without any underlying reasoning why prices should adhere to a given pattern might suggest that any given 'trend' he is observing could be better explained and modeled using some other methodology. I am of the persuasion that behavioral finance, while still in its infancy, offers the best methodology for explaining long-term price variance and that the way human emotions interact with capital markets would be the only effective way of attempting to predict future market prices.

As human beings, we are equipped with an innate ability and compulsion to see patterns. It is a very effective way to attain survival in the natural world. I will agree that there are some situations in investing where patterns can be informative, but not the geometric ones used in day trading. Without any fundamental reason why prices should behave in a certain way other than observation of historical data, it would seem ludicrous to go with the patterns.

The Selection Bias in Goldman's Argument Against the Obama Regulation

Subscribe in a reader

The proposed Obama regulations that I'm referring to are the statements that came out today that he would:

"seek to prevent banks that have special access to low-cost Fed funding from operating or investing in hedge funds, private equity funds, or trading purely for their own benefit in a way that’s unrelated to serving customers." [Bloomberg]

While this is sure to be contentious, I think it's important to take a step back and look at what especially discounted Fed funding is intended for. I'm not talking about the funding that the Fed provides during a "business as usual" scenario, but in scenarios such as this recent credit crisis, when the discount window's possible term length was raised to 90 days and interest rates slashed, the goal was presumably to get some emergency liquidity out there to try and unfreeze credit markets and stimulate lending again.

I won't try to quantify the potential for ruin that hedge funds, private equity, and proprietary trading might potentially represent to a bank, although this is certainly possible. If the goal is to stimulate lending, however, having all these components as parts of banks that have the discount window available to them would make it difficult to monitor where these emergency funds might be going once they leave the Fed. While these other activities no doubt have the potential to be economically important, I don't think there would be significant objection that the Fed's primary objective in a crisis, is to funnel funds to firms that are going to use that money to lend and keep credit flowing, might be hindered if these funds are instead going to any number of other activities available to some of the firms that received emergency Fed funding.

This leads to the Goldman comments coming from Chief Financial Officer David Viniar. He argued:

"If people are focused on things that caused, or were real contributors to the crisis, it wasn’t trading...Most trading results were actually pretty good, not just at Goldman Sachs but at most firms and that’s not really where the problems were." [Bloomberg]

Looking at the results of trading from the crisis as a sample of typical trading is a vivid example of selection bias. It seems that Viniar would like us to think that because most trading was profitable during the crisis, it will be profitable in the future and doesn't doesn't pose systemic risk to banks. While on average trading profits might be positive, that says nothing for the skewness and kurtosis of trading profits/losses. By that I mean that proprietary trading could be negatively skewed or with fat tails (positive kurtosis) in the negative end of the distribution.

This is an example of a positively skewed, positive kurtosis, positive mean distribution. While it will be on average positive most of the time, the long negative tail creates the possibility for financial ruin when returns end up in that part of the distribution.

I'm not going to take sides on this issue, since I can see the validity in both. That being said, I think that Vinair's argument regarding proprietary trading is mistaken. For example, while fund managers can on average beat the market by holding a portfolio of the S&P 500 and writing puts (insurance against a fall) on the S&P 500, when it does go bad, it's ruinous. This is the "picking up nickels in front of a steamroller" scenario.

While I don't claim to know the distribution of trading profits or the absolute level of risk these firms are taking on (they might not know either), I can understand why Obama and his administration would want to prevent artificially-lowered interest rates from fueling these types of activities. Especially since it's meant to stimulate lending, allowing firms that have access to the Fed's discount window to engage in hedge funds, private equity and proprietary trading could be a usage of funds not best for unfreezing financial markets.

That being said, Obama needs to be careful when he talks about "curbing excessive risk taking". Lending to small businesses is one of the riskiest types of loan a consumer bank can do, and since stimulating the flow of credit to areas of the economy such as this seems to be an Obama administration priority, curbing risk taking shouldn't be his goal.