Feb

12

 Specs. I am trying to write a little something about technical analysis. Here's what I came up with. (by the way the first one to show that random charts and stock market charts look similar was Harry Roberts I think in 1956 or so but Holbrook working may have done it 20 years before). Anyway, how would you improve on what I wrote.

Most traders in the markets use charts and technical analysis to establish and exit their positions. Academicians and skeptics point to the random nature of many technical patterns. Here's a typical chart generated by random numbers. If you don't tell a trader it's randomly generated, they'll come up with all sorts of predictions and patterns that the chart generates. And if you dare to suggest that what they're doing is mumbo jumbo, they take great offense and beat you on the head with examples of great traders who follow charts, and examples of others who consistently make a fortune by using charts.

There's a trader from Harvard who uses charts and has made 20 billion who says "using a chart is like a Dr. taking your temperature before a diagnosis." Another one says that if charts are so useless how come everyone including you looks at it before making a trade. One of the most respected and successful traders, a friend, puts the debate in focus: "There are lots of great tools in technical analysis (some of them in his book like trader's positions, and breakouts, open interest and spreads). They're very useful as part of a bigger trading process. There are good saws and hammers but it takes a good carpenter to make them work."

There's a guy in Japan who calls himself the Japanese Victor Niederhoffer who has turned $ 10,000 into 5 million by using charts. I hope to meet him in Japan when I visit there for a talk arranged by one who believes in charts, an estimable fellow who combines charts with anthropology, life extension and sports, and perhaps I will become the American Matsohita-Masamichi.

Options values are determined by using random numbers with the same standard deviation and distribution of prices as would be generated with the random number generators I just mentioned. Every trader on the floor uses such generators to predict the price that an option should trade at, and they do very well with this model– until something like the 1987 crash occurs and they go broke.

A famous former academic big options trader and head of the exchange said that almost all the scientific options traders he knew found that when you apply the random walk model to options, it turns out that puts are priced much too highly. He said that he's watched every last one of them go broke. The problem here is that extreme events tend to occur much more frequently than the random walk model would predict.

As I write, the Swiss franc recently jumped about 100 standard deviations above its last price in a few minutes, a one in a trillion shot, and billions were lost by option writers who used defective models to place their bets.

Andrew Goodwin comments: 

The error made is in the actual coin flip method versus the computer generated random flips. If you flip a real coin an infinite amount of times, then the side that is heavier because of a greater extruding feature weight will land more often on the bottom excluding unknown aerodynamic effects.

I hereby wish to debunk weighted coin tosses as fair. That includes the wear and tear on the coin that changes the weight. Over time, the side with the extruding images on the same coin wears down and you get closer to random results.

With the computer generated flips you get no advantage betting either way unless you game the random seed.

If you get Monopoly style game dice that are indented 6 times on one side and just one time on the other, then you are better off betting on the heavier 1 dot side landing on the bottom on a given role over large numbers of roles even if the edge is tiny.

I officially quite playing indented dice board games now.

anonymous writes:

One approach I have taken is to identify some price formation of interest that can be defined by quantitative rules (perhaps "inside days" or "upside breakouts") and then analyze that formation in an actual data series. If the occurrence and distribution of the formation in the actual data series is consistent with randomness, then I make the assumption that it is highly unlikely that the formation contains any additional predictive information.

Sushil Kedia writes: 

"A synthetic price series if generated using some function incorporating random numbers looks similar to a real stock chart"

Q1. Is every variable in that function taking random numbers as inputs? Q2. If A1 is no, then is any such function using random numbers akin to the error terms in the assumptions of a good regression model? Q3. If A2 is yes, then how does such a random number generated chart conclude that real markets are random? Q4. If A2 is no, then which parts of the synthetic price generating function are significant enough to conclude that the final outcome is really random?

Please allow a surmise to be placed on this table, before you tear it off:

The outcome of prices is a joint function of the random reaction to new information at that instant as well as a function of sensitivity of all participants to trigger or not to trigger actions on such moment by moment information updates. Sensitivity is again a multi-variable function comprising of but not limited to factors such as existing position (bias), risk perception (capacity to add or reduce risk at that instant), time horizon and so on and so forth.

Please allow just one more surmise on this table, for the moment, where I will unabashedly borrow from the Palindrome's famous idea of reflexivity. Markets have a feedback loop.

My arguments supporting the surmises:

Cause & effect thinking that is the cause celebre and raison d'etre of known forms of sciences has yet not evolved into modelling, evaluating or concluding enough about phenomena that have feedback loops as well as random reactions.

That's where art steps in.

Eventually as the long held and commonly accepted belief (derived from philosphical arguments) of this list has been that there is no possibility of any reward without some risk, since at zero risk the other side of the trade does not exist, all workable methods will have approximations and estimates.

If one method may or may not be better or inferior than the other, having a method is better than no method. If even in the illusion of forecasting better than randomness one can use a chart, any form of art, or any other mechanism to stay actionable in the face of risk and prevent oneself from ruin, then too randomness will allow one to get closer to being rich enough.

Finally I will quote two giants from this list itself:

Ever Changing Cycles as espoused by the Chair himself, refute any scope for any one method to remain superior or inferior to any other.

The Senator having said once to me that every Cigarette packet comes with the statutory warning that smoking kills and yet it is the user of that information who ignores it. So any method is not the bigger factor in performance, it is the user of that method.

Whether Technical Analysis appears archaic, has refused to involve beyond the simplistic and lacks the sex appeal of rigorous numerics, so long as it triggers a trader to be adaptive, manage his risk and makes one pay one's bills, it's ok. There are enough systematic quant funds that have blown up and the biggest blowout did happen when Genius Failed, since it refused to recognize the ever changing cycles. 

anonymous writes: 

A price chart is an attempt to model relevant aspects of price change. Price change is not linear displacement, whether vertical, horizontal or oblique. Nonetheless, price change can be represented as vertical displacement and time elapsed as horizontal displacement. Such a model, however, invariably supports relationships that does not correspond to anything in the original process.The angular inclination of a trend on a price chart is a visually striking feature of this representation. Such angles have no intrinsic meaning for the price series, but this is one of the many factors (along with our facility for pattern recognition and wishful thinking) that contributes to our interpreting more from price charts than rigorous testing reveals is there.

- William Eckhardt


Comments

WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '10041' AND comment_approved = '1' ORDER BY comment_date

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search