Search results

Exactly, no matter what time frame someone chooses, when the trend starts and when it ends will be randomly distributed. How does one develop a trading method based on those random occurrences? Perhaps do not use TA do generate buy/sell signals, and instead use future value models to determine...

All TA indicators are subjective to one degree or another since you have to choose the parameters to use as input, lookback period for example. Anyhow, have look at John Ehlers work on the Instantaneous Trendline and the Decycler.

Here is a good paper. Just google Robert Tompkins and volatility. He has done lots of interesting work on surfaces. http://thfinance.de/RobertTompkins/EJF2.pdf

So if the machine algorithmns are just trading with each other to foment a market, what are some general methods to "game" the machines? It would seem when the machines are working, price movement is less of a random walk process. So would extreme value methods be most suitable at these times?

Ok, agree with you on slope, that was just the simplest example. I still believe that in the definition and calculation of trend (however one chooses), the time frame determines what the trend will be, and not vice versa. My opinion of course, but that is what makes markets.

Got to disagree. Any method to define or measure a trend should be independent of the scale of the time frame. Example, the simplest definition of a line in xy coordinates is two points. The slope (trend) of that line is dy/dx (rise over run). I can apply that definition and calculation over any...

Defining a trend over a given time period x is easy, and can be done by any number of methods. However, how do you choose the value of x? Choosing x is arbitrary, and trading a trend going forward based off of x will be randomly distributed. The smaller the value of x, the more random the...

There are some good books by Riccardo Rebonato and Alireza Javaheri that go into detail about volatility hedging and putting on skew and kurtosis trades.

Delta can give you a rough idea of probability. For a more quantitative answer, use this formula (assuming a normal return distribution with no skew or kurtosis) X = exp(sigma*t*x)*S where X = future spot sigma = percent volatility t = sqrt(days 'til expiry/365) x = standard deviations S =...

Bullshit, the future cannot be predicted with anything greater than 50% precision.

You have just discovered the fractal nature of financial time series. Congrats!

dom, have you looked at the Autocorrelation Periodogram method of locating the dominant cycle? He prefilters (roofing filter) to stay within the 10-40 cycle bounds. However, remember Ehlers work is all bullshit (if good intentioned) if the market is even close to a random walk process.

John Ehlers is the man for using DSP methods on market data. Lots of his work is available for free if you search. Look at his work on finding the dominant cycle, and using that to tune your indicators. Just don't expect John's work or TA in general to make you into an instant bazillionaire. I...

The reason moving average cross-overs does not work, or more generally TA is because price moves are pretty close to a random walk process. I say pretty close, because the log normal distribution of price always has some amount of skew and kurtosis in it. Thus, while price may not be perfectly...

The attached text file is a couple of lists of tenets concerning speculation attributed to Jesse Livermore. Perhaps the greatest speculator, and one of the worst risk managers ever (looking at you LTCM, Barings, AIG, Bear Stearns etc...)

So then we should all forget about technical analysis because it is BS, and any attempt to validate TA through backtesting is BS because it is all based on hindsight? Maybe we should listen to John Bogle and just buy and hold equities and forget any kind of market timing or trading. Well that...

Fluctuate yes, but if you have a zero or a negative mean, then you don't have a positive expectancy system do you? Of course you would have to have data from enough different market cycles to be reasonably sure, but that is what proper backtesting is for. Scalable is relative, all depends on...

The ATM strike should have a delta closest to 0.50.

If you have a look at John Ehlers latest book "Cycle Analytics for Traders", in chapter 17 there are two sections which may help you called "Evaluating a Trading Strategy" and "Monte Carlo Evaluation." In fact you could Google it and probably find those sections online. Also, I got this...

For options, the delta can give a reasonable approximation of the probability of the price finishing OTM. The probability of price touching the strike is approximately twice the probability of finishing OTM.