LOUIS BACHELIER

Let's kick off our review of the evidence by going way back to 1900. That's when a young French mathematician, Louis Bachelier, was undertaking a doctorate in mathematical sciences at Paris's Sorbonne University. Bachelier was studying how the market prices of options on government bonds behaved. After gathering his findings he cut and diced them mathematically and then presented them in his doctoral thesis. These words are from the introduction of the thesis:

The influences that determine the movements of the Exchange are innumerable; past, current and even anticipated events that often have no obvious connection with its changes have repercussions for the price … The determination of these movements depends upon an infinite number of factors; it is thus impossible to hope for mathematical predictability.²⁰

Bachelier concluded that the market can only be beaten by luck. Here's another quote from his thesis: ‘At a given instant the market believes neither in a rise nor a fall of the true price.' And another: ‘Evidently a player will be neither advantaged nor disadvantaged if his total mathematical expectation is zero.'

So he pretty much kicked the game off by describing the financial markets as a gamble. As with roulette, you can't pick the next spin of the wheel. He's saying price trends are a myth because security prices move in a random, unpredictable manner.

Bachelier argued that for each quoted price there existed both buyers and sellers. In any trade, who's right and who's wrong? And he has a point. You may be full of conviction, but so too is the guy on the other side of the trade.

ALFRED COWLES

Alfred Cowles III was born in 1891, the grandson of the founder of the Chicago Tribune — which means he was born into money. In an effort to find this money a home, Cowles developed a strong interest in the financial markets. He subscribed to several investment newsletters that, as today, claimed to provide their readers with an investing edge. From 1928 to 1932 Cowles tracked the tips provided by the two dozen most popular sheets. It was an interesting period since it covered the overheated bull market preceding the 1929 Crash, the Crash itself and the crushing bear market following it. Cowles felt, in retrospect, that any tipping sheet worth its salt should have been able to flag the big market moves before they occurred.

Over the four years of Cowles' review, the Dow reached its giddying heights of September 1929, sustained a 23 per cent plunge over two days (28 and 29 October 1929) and fell by an unprecedented, and yet to be surpassed, 89 per cent between September 1929 and July 1932. These were massive events but, just like the general public, the tipping sheets didn't see them coming.

Cowles became intrigued by the question of whether stock prices are predictable. He wrote to Yale professor Irving Fisher, President of the Econometric Society and an old friend of his father. There was a certain irony here since Fisher will be forever remembered for making what proved to be one of the most ill-timed market predictions of the 20th century. Two weeks before the 1929 Crash he stated that stock prices ‘have reached what looks like a permanently high plateau'. With egg still running down his face, it's likely Fisher was also interested in the answer to Cowles' question. In January 1932 Cowles established the Cowles Commission for Research in Economics. The commission spawned the respected journal Econometrica, which carried an article in July 1933 entitled ‘Can Stock Market Forecasters Forecast?' Its conclusion: ‘doubtful'.

In 1950 Cowles published a very interesting study in Econometrica. He was looking at the potential for markets to trend. It made intuitive sense to him that if stock prices trended then there would be a greater proportion of price sequences than reversals. He had published a study along similar lines in 1937, but in response to criticism he revised his statistical method for the 1950 paper. He found that there was a slight tendency for prices to sequence, but that it could easily be explained away. For example, there is a tendency for general stock market prices to change over time anyway.^b Alternatively, a move in interest rates is sufficient to explain the degree of change Cowles observed. Stock yields adjust to changing interest rates, and shifting yield expectations deliver changes in underlying stock prices.

More importantly, even if a correlation did explain the results observed in Cowles' study, it wasn't significant enough to deliver a trading profit.

HOLBROOK WORKING

In 1934 agricultural economist Holbrook Working published the results of a novel study that looked at price changes in commodity markets.²¹ He started by acknowledging that patterns existed in historical price charts. And clearly he wasn't alone in thinking so — for years chartists had been drawing conclusions from the paths these charts traced out. But Working's question wasn't how the charts appeared but whether there was any justification in drawing conclusions from their appearance.

Traditional price charts show consecutive market prices wandering across the page, but Working used these price charts to produce a different type of chart. He measured and then showed graphically the quantum of successive price changes, and he found that, instead of being pattern-like, they took on a totally random appearance.

He tested the concept further by producing graphs ‘in reverse'. First he generated a series of random numbers; these he treated as the quantum of successive price change. Then he used them to construct what appeared to be traditional price charts. He compared these randomly generated charts with real commodity price charts, and couldn't tell the difference. He even showed the charts to professional commodities traders down at the Chicago trading pits — they couldn't tell the difference either.

Holbrook's findings were a blow to traders who were confident that historical price charts delivered insightful information.

RALPH ELLIOTT AND MAURICE KENDALL

Ralph Elliott told the world in 1938 that market prices unfold in clearly discernible patterns. His concepts have been distilled into a stock trading technique called Elliott Wave Theory. Elliott stated in his book The Wave Principle that:

… because man is subject to rhythmical procedure, calculations having to do with his activities can be projected far into the future with a justification and certainty heretofore unattainable.²²

Let's look at how Elliott's concept stacks up against the evidence.

Fifteen years after The Wave Principle was published, London statistician Maurice Kendall put Elliott's claims under the microscope. His article, published in the Journal of the Royal Statistical Society, shot holes in Elliott's claims. After analysing copious volumes of data covering 19 different stock groupings over a 10-year period, as well as 50 years of price movements on wheat and 135 years of price movements on cotton, he found no evidence of association between consecutive commodity prices. And he had this to say about stocks:

[Trends were] so weak as to dispose at once of any possibility of being able to use them for prediction. The Stock Exchange, it would appear, has a memory lasting less than a week.²³

In other words, in debunking Elliott, Kendall found himself agreeing with Working.

HARRY ROBERTS

Six years after Kendall's paper, the Journal of Finance published a paper by Chicago statistician Harry Roberts. Like Working and Kendall, Roberts felt that proponents of technical analysis were being fooled into believing price charts were supplying information that simply wasn't there. Technical analysts had long argued that embedded within price and volume data lay everything you needed to know about a stock — market sentiment, future prospects, value and risk.

But, like Working and Kendall before him, Roberts felt that the shortcoming of traditional price charts was that they displayed successive price levels rather than quantum changes in price. This, he felt, gave the artificial appearance of a pattern or trend. Roberts' methodology was similar to Kendall's but he based his study on US companies. And, like Kendall, he concluded that chance outcomes produce patterns that invite spurious interpretations. Roberts made particular reference in his study to the appearance of one randomly generated chart — it showed a pattern that any technical analyst would have clearly identified as a ‘head-and-shoulders' top.

Roberts declared that the main reason for producing his paper was to call to the attention of financial analysts empirical results that seemed to have been ignored. He reminded them of Working's and Kendall's earlier findings as well as describing his own. When you read Roberts' paper you can sense his frustration. That was 1959. Roberts lived until 2004 and, despite his pleas, few people listened. They continued to trust what the charts were ‘telling' them.

Roberts felt that the appearance of a trend was an easy concept to challenge, because chance outcomes are often successive in nature. A coin tossed a hundred times doesn't come up heads, tails, heads, tails, and so on for 100 tosses in a row (in fact this outcome is so unlikely you should feel comfortable betting your life savings against it happening). But successions of heads or of tails will occur, and if you plot the outcome it will look like a trend — even though it's not.

Yet how often do you hear someone playing a game of chance claiming to be on a ‘hot streak'? The hot streak is merely their interpretation of a string of chance events. Or the person who has tossed heads four times in a row and announces there's a greater chance that tails will come up next because ‘it's time it did'. In both cases chance events carry no predictive power whatsoever, yet people dearly believe that they do.

The reality is that the interpretation of past stock price movements can be equally deceptive.

WHAT ABOUT BULL AND BEAR MARKETS?

Let's now consider this issue from another angle — one that isn't research based and that appeals to intuition.

Market prices are driven by perceptions of value — many people's perceptions of value. Some base their perceptions on estimates of future earnings, but not everyone reaches their value judgement so analytically. For many, sentiment alone drives their perceptions of value. So what happens if sentiment becomes the predominant force shaping perceptions, such as occurs in bull or bear markets? Sentiment can become self-fulfilling. Isn't it possible that if most investors believe a market should rise, and it does, that this will reinforce their belief and the market will rise further? And while that view can't be held forever, isn't this causing the market to trend for a while at least? In other words, are the findings of the researchers — that market prices don't trend — inapplicable when sentiment is polarised, as in bull and bear markets?

People get drawn into crazes — bike riding, Pilates, yoyos, Sudoku puzzles, Pokemon cards, the stock market. It's simply part of being human. A bull market is just another craze, and it's tempting to believe that bull markets trend. As more and more people get caught up in the craze, the price movement gains momentum. Increasing prices attract more participants. Anticipated price rises become self-fulfilling as more people join the craze.

One of the problems here is that few people recognise a bull market, which means that, even if it can be shown subsequently that bull markets trend, most participants don't know when to start looking for that trend. And there's a bigger problem: even assuming the capacity to recognise a bull market, how is one supposed to know when it's going to end? A failure to recognise that is like walking blindfolded towards a cliff.

Outside of a bull market it's more difficult to argue the case for trends. What about a bear market? Bear markets typically start with a crash or a significant correction, which is then followed by a period of enhanced price volatility. Attempting to find a trend among that lot is unlikely to be either productive or profitable.

PUTTING RANDOMNESS INTO PERSPECTIVE

I want to put this whole randomness issue into perspective, because it's easy to come to the wrong conclusion. It's not that share prices are driven solely by chance; that would be nonsense. A stock consistently earning $2 per share is clearly worth more than one earning 50 cents. Real forces do drive share prices over time. But the forces guiding current prices are based on perceptions of the future, and these perceptions are constantly changing. It's this that leads to the observed randomness.

To help explain, let's consider a hypothetical company. Like most companies it generates profits. It pays a proportion of these profits to its shareholders in the form of dividends and retains the rest to reinvest into the business. Over time the underlying value of the business increases and so its share price follows — but not in a straight line. If fundamentals justify an increase in the underlying value of the company of 6 per cent over a 12-month period, you won't see the share price increase by a geometric average of 0.112 per cent every week over the course of a year. No stock ever has or ever will trace out a perfectly straight share price graph. Instead, market prices jump around in an unfolding long-term revaluation.

Given the multitude of other factors that also drive share prices —
including investor sentiment, interest rates and economic conditions (and changing perceptions of all these factors) — market prices must be viewed as a reflection of the underlying value of a business rather than an exact measure. Share price movements map out a course like a drunk walking from point A to point B. For significant periods of time he'll probably be heading in the wrong direction but eventually he'll arrive at his destination. So it's wrong to judge where the drunk's final destination is simply by observing the direction he's walking at any point in time. Similarly, it's wrong to judge a stock's true value by either the current market price or the direction the stock price is heading at any particular time.

Using another analogy, the relationship between a stock's real value and its market price can be likened to a rubber band. Sometimes market price and underlying value are close; sometimes extremes of human emotion and misperception pull them far apart. But, like the rubber band in various stages of extension, the connection is self-correcting. Misplaced sentiment and misjudgement can cause prices to move in the wrong direction, but not forever.

This underpins one of the fundamental problems I see in relying solely on technical analysis. Technical analysts draw conclusions from market price and volume data, but this information is provided without any distinction between rational and emotional influences on the data. This all-inclusive method of measurement — the supposed strength of technical analysis — I see as a weakness. To take your cue solely from the actions of the market means you'll probably find yourself following the drunk as he's trying to find his way home.