# AnalyticBridge

A Data Science Central Community

The stock market is much more random than what most investors think. Indeed, it looks almost like a random walk: even if it has been going down 7 days in a row, the probability that it will go dow tomorrow is still 50%: random walks are memory-less processes. The chart below illustrates a simulated, perfectly random / unpredictable stock market.

• Why can't you consistently win? Events that can boost or kill the stock markets are quickly exploited by professional traders. This explains the gap between yesterday close price, and today open price. Indeed many traders don't keep open position overnight because of this.
• For large indices (QQQQ etc.), the correlation with today's price and the price from yesterday or from k days ago (k = 1, 2, 3, etc.) is essentially zero. If it was significantly different from zero (from a statistical point of view), then this pattern would have been exploited by professional traders - many holding a MIT or Princeton PhD in mathematics - and eventually the pattern would die. Note that the time series correlogram (some call it the spectral signature) associated with stock prices uniquely identifies the nature of the underlying stochastic process. Time series with a flat correlogram represent pure random walks, that is, processes where you can not predict the future.
• So is the stock market truly random? No. It is true that when a large number of independent algorithms (run by hedge funds) compete in real time to extract money from Wall Street, eventually nobody wins or lose - and the stock market becomes as random as a lottery: some can win for some time, but on the long term they can't. The return becomes zero. Think of your daily commute when you are stuck on highways driving at 5 miles an hour: can you find a pattern to beat the rat race, maybe a new road that you can use to beat your fellow commuters? If you do, that road will be found by thousands of commuters fighting the same battle to minimize commute time, and your miracle road will turn into rat race again, in a matter of days. The same mathematical principles (arbitrage) apply to the stock market. However, even if you exclude insider trading and after hours trading, the market, while almost random, is not entirely random. Most of Wall Street trading algorithms have been designed by mathematicians trained in the same universities, and are not independent: it is like all search engine algorithms providing the same search results for most keyword searches. This indeed explains why the stock market can experience flash crashes or flash spikes.
• This discussion about randomness applies to short-term trading. But is there a long-term trend that can be leveraged? In my opinion, since year 2000, the stock market has no up or down trends: while highly volatile, it is mostly flat. I think there's a good chance for a slow downward trend given the fact that baby boomers are retiring, demographic pressures, and the young generation having lost trust in 401K plans. Yet government might impose some regulations to artificially keep the market from collapsing - otherwise long term shorting would be a great option for the smart people who are currently "all cash".
• Does it mean that there's no money to be made for the average investor? Well, it is becoming more like a lottery, and a few people win money, sometimes big money, in all lotteries. Look at my above chart showing a purely random (simulated) stock market: there are local trends, patterns such as short squeeze, "head and shoulder", market crash, steady growth, runs (going up or down for 7 days in a row). Every kind of pattern can be found in this simulated, random, trend-less stock market, just like in the real stock market. It is clear that you could make money in this simulated market - what is less clear is that you have no way to predict if your algorithm will or will not make money. Note that to increase your odds, you have to back test your strategies and perform sound cross-validation or "walk-forward" (http://en.wikipedia.org/wiki/Walk_forward_optimization) to predict what your return might be.
• Despite the increasingly random nature of the stock market, it is still possible, for the average educated trader, to exploit patterns. Typically, any pattern that Wall Street is reluctant to exploit - such as staying all cash for 5 years in a row and trading like crazy when the right opportunity presents itself. In a nutshell, Wall Street traders must think very short term or otherwise risk losing their job. This provides an opportunity for the amateur trader. As far a short-term patterns are concerned, we've found that they occur no more than three times, and can be exploited only during the third occurrence (the first two occurrences being used for pattern detection and confirmation): after 3 occurences, Wall Street gurus have exploited and killed the pattern (the patten - e.g. a short squeeze on a particular stock - might still be there, but it's parameters have shifted on the fourth occurence, due to heavy exploitation).
• The future could be rosier: with more traders leaving this extraordinary competitive environment (e.g. they leave because they are laid off due to poor performance) and less money circulating in the stock market due to baby boomers retiring, the nature of the stock market will become less random, whether the general market trend is up, down or neutral. This will provide new opportunities for traders who haven't been killed in the 2000-2010 "wild fire".

Views: 14011

### Replies to This Discussion

I have found tradeable patterns in the market in 10 years of data, but the predictor variables are not price related. The market in the short term is highly exposed to event risk, but there are inefficiencies in pricing and not always solely the price of the stock.  Due to the extrodinarily numerous ways to make money, there are methods to make money.  However, as the patterns become exposed, they become exploited and neutralized.  The question is which way is your way  to overcome the numerous cognitive biases in a non linear counter intuitive futuristic system.

Thomas:

Very intellectual argument.

So much so, it's hard to figure out what point you are trying to make with all those "big" words.

"numerous cognitive biases"

"non linear counter intuitive futuristic system"

"as patterns get exposed"

I find your comments edifying in a translucent manner such that your non linear variables are based on a non parametric synergy with the socio-economic political efficiencies of the leading third world nation-states leadership's psychic preferences.

Also, if you liked that explanation, you are really in need of a brain transplant.

You can make money in the stock market, I know people who do. There are tradeable patterns, they are called price inefficiencies.  there are fundamentals, which means that the investor has a better understanding of the stock's product and services and management team than the other portfolio managers.

They don't limit themselves to just trading or investing in stocks, but they use many different methods and instruments to trade.  How they manage the process is with automation to remove the human cognititve biases, which has the potential to derail the sucessful signal if a human is required to interpret the signal and place the trade.

the simpliest observation is that its MUCH easier for management to screw up a company and make mistakes with their ever optimistic all powerful control beliefs, than it is to run an always growing always profitable company for 20 plus years.  in other words, shorting has larger downward movement in percentage terms and is an easier way to make money if you know the signs.

The only problem with (shorting) a poorly managed company is that there are so many of them.

ATT is a perfect case in point.

They started out with a monopoly on both local and long distance phone service.  They blew that by pissing off their customer base enough so that legislation was passed to break them up into "Baby Bells".  However, they still had long distance left (right)?

Then, they managed to screw that up when they failed to take account of an obvious technology that would eventually destroy their business model - VOIP.

Then, they made a bunch of idiotic purchases to include buying a cable company that didn't have a digital (or reliable) footprint.

Then, they also bought NCR AFTER their computer line became (after about ten/10 years) no longer an industry leader.  They ended up taking a loss on that one by selling the company back to it's original employees.

Then, they sold off their subsidiary operations like Bell Labs.  That brought tears to my eyes when such a premier technical research organization got exported (to the French or was it the Italians?).

The whole time the corporation still survived and generated "some" cash flow.  However, ANYBODY dumb enough to think that ATT would survive intact AFTER PROVING that their bloated management systems were completing dysfunctional was really brain dead.

Of course, in the short run, it was probably possible to make money on ATT by going long (on those rare occasions that they "accidentially" made the right decision).

In my opinion, it is a lot easier to "predict" when a company's price is going to get a boost than when a company is going into the dumper.  Of course, there are exceptions to all rules - especially yours.

It's the exceptions in life (and trading) that will really give you a haircut.

High tech and large companies are played differently than say, small retailers in a deflationary environment or when disposable income is NOT growing.    Each large industry gruop has its own characteristics, but sometimes its mgmt and sometimes its macro and sometimes its both when a particular stock moves. . . . I have made most of my money being short and lost most of my money being long, which is to my original point:

The question is which way is your way  to overcome the numerous cognitive biases in a non linear counter intuitive futuristic system.

And a programmer i know just programmed "Reversion to the mean", and made plenty of money while he was asleep.

Currency trading, which used to be considered one of the more safer trading vehicles, (since government policy couldn't be manipulated), is now manipulated.

Just happened to read today in NY Times:

# A Primer on How Currency Manipulation Worked

http://dealbook.nytimes.com/2014/11/12/a-primer-on-how-currency-man...

"In one instance, the British regulator said a trader at Citigroup made a \$99,000 profit based on trades executed in a 33-second period ahead of the E.C.B. fix after colluding with traders at four other firms".

How is the average trader supposed to compete against that?

Making money in stocks has always been more an exercise in money management than finding and exploiting patterns. Knowing how to allocate money and setting stops and exit strategies are more important to success than. Using simple breakouts is still profitable. That being said, there is a huge difference between managing accounts in the thousands of dollars and running a fund. It is still possible to make money in stocks without being a quant or understanding machine learning.

It is sometimes better to be lucky than good, but, you are correct, there is no way to count on luck over the long term.

Nevertheless, we have to wonder at just how rich Warren Buffett had become since 1960, when he started value investing, as promulgated by the great Benjamin Graham, if he could afford to make his notorious \$200 billion mistakes and still remain one of the world's richest people.

The "stock market" behaves much more like a school of fish swimming in the ocean or like a flock of birds flying overhead than fundmental models and random walk theory can describe. Other similes can apply, but these are apt. We do know that order does exist within the apparent chaos, but mathematical models as well as we have been able to develop them just have not been capable of reliably describing it or predicting it. At any one time different fish in the same school are swimming in radically different directions. So how does a fisherman catch one? Must he catch the entire school of fish to get the one he wants?

Nevertheless, the most successful traders and investors would admit that they are not right with their decisions 100% of the time. They do not even comprehend everything they could know that bears on the price and value of their selected securities. However, they do practice the art of reducing their risks and capitalizing on opportunities as they arise. Practicing this art successfully over the long term requires data and metrics, yes, but above all it requires human intelligence and skill.

While I agree with the well-known random-walk hypothesis, this assessment of whether one should invest in the stock market as a whole presumes a particular style of investing with an emphasis on statistical pattern analysis of some sort.  There are a variety of investment styles.  Here is a more optimistic view:  I've been investing in mutual funds and later (2000) stocks on my own since 1974, starting with very little savings from a military salary, and have done well. I've experimented with a variety of mutual fund and stock selection techniques, with some complete failures, some tremendous gains, and mostly middle-of-the-road returns. FWIW, I advise my children to save and invest 25% of their income, focus on value investing in stocks that pay a dividend, long term buy & hold, favoring 'dividend aristocrats', reinvesting dividends (DRIP).  I recommended that they learn how to read a 10K, with an emphasis on ratios such as debt/equity and current. I am not a 'trader', but also do some limited option trading now, which takes a lot more time and a higher degree of sophistication and discipline. I agree with other comments in this thread that being an insignificant individual player affords more flexibility than institutional investors, and have in the past used trailing stop loss orders that institutional investors could not employ because of their size.  My opinion, based on a very limited  40-year sample size of one,  is that it is possible for individuals with limited assets to invest successfully if they are thoughtful, selective, and patient, with a long term perspective emphasizing buying stocks that are undervalued and generally not worrying about the market as a whole. However, that just is one of many possible investment styles. I agree that trading is very high risk for the individual investor, and if done at all, it should be with a small percentage of one's portfolio with the expectation of hard lessons. Nearing retirement, my focus now is on income.  This is not meant to be the start of a lengthy investing discussion, just a counterpoint more hopeful view. Warning: I am not a certified professional investment adviser; the foregoing is a very limited commentary and the opinion of only one person.