Search results

You shouldn't worry about skew or kurtosis in your Kelly calculation. In fact you should probably use fractional Kelly and just stick with a normality assumption.

The person whose work to want to look at for market cycles analysis is John Ehlers. He uses DSP methods applied to finance. The caveat I always give about his approach to technical analysis is that anything from the world of physics applied to finance may be a misapplication of the physics. That...

To paraphrase Churchill, "Capitalism is about the unequal distribution of the wealth, while Socialism is about the equal distribution of the misery."

Correlation is the wrong metric. Cointegration is the metric to use for pairs.

Ehlers is a bona fide DSP engineer. The material he presents is backed by solid science and engineering. As always, a caveat, anything from the world of physics applied to finance is or could be a misapplication of the physics. Still, better than Elliot Wave!

MA's calculated and used in the traditional way don't work because of group delay, commonly called lag. If you are going to use MA's, learn to view them as lowpass filters, and learn how to engineer them for your needs. John Ehlers' work is what you want to look at.

The problem is that there is every way under the sun to define a trend. How many of those can become a positive expectancy system? Who knows, perhaps many of them or perhaps none of them. Unlike the opinion in the "Backtesting is Useless" thread, I think that's what you need to do in order to...

However you choose to define a trend, when it starts and when it ends are randomly distributed through time. Thus the best model of market movement, even with all of it's shortcomings, is still random walk. Even random walks produce patterns that look like trends in hindsight.

Have you ever been to Uranus?

One strategy he has talked about in the past was to buy T-bills, and use the interest income to purchase OTM options. That is an example of his barbell strategy, in other words take low risk and enormous risk at the same time. I do not know what strategies he uses currently.

Have a look at the attached file for example calculations. Of course they are wrong because it assumes a normal distribution and no skew or kurtosis, but you get the basics.

The type and period of the MA doesn't matter. MA's, as traditionally used, don't work, because of group delay, more commonly known as lag. If you are interesting in using moving averages in finance, look up the work of John Ehlers.

What you are thinking about is the volatility surface, and if there are anomalies that you can take advantage of profitably. In theory, if a properly normalized surface is stable over time, then yes you could look for mispricings that way. That is the theory anyway. However, lots of big players...

The true answer is that it doesn't matter. MAs, as traditionally used, do not work. The conventional wisdom is that a 200 day SMA has some value as a self fulfilling prophecy. Do not believe the conventional wisdom. If you insist on going down the rabbit hole of moving averages, at least learn...

Exactly the reason you can't use the DTFT in finance. Look up Ehlers' implementation of the autocorrelation periodogram and MESA algorithms. From ph1l's post, I thought the HHT was noncausal and thus of limited use in trading. I will try to read the article when I can get some time to digest it...

The upside to Ehlers' work is that he tries to apply well understood science and engineering principles to market analysis. The downside (as some see it), anything from the world of physics applied to finance data is a misapplication of the physics. Ehlers models market data as a complex...

I wasn't clear if you do this, but I would recommend that you takes trades based on maximizing payoff, which is: payoff = probability*(return/risk)

This is why I learned to stay the hell away from options, too many parameters to get right in order to make money. Of the five inputs in a BS pricing model, absolute level of the underlying, absolute level of volatility, and time decay are the factors you most need to get correct. That's three...

To answer the OP more directly, the EMH is flawed because of the skew and kurtosis that shows up in essentially all price and return distributions. The causes of skew and kurtosis are due to the complexity of the underlying process. I would argue that financial markets are more complex than even...

I wish markets were perfectly efficient, ie Gaussian. That way I could know exact odds on a particular outcome.