Do you see patterns in Random Walks?
- Thread starter atlTrader666
- Start date
Video of a massive starling flock turning and twisting over a river in Ireland has gone viral, and with good reason. Flocking starlings are one of natureâs most extraordinary sights: Just a few hundred birds moving as one is enough to convey a sense of suspended reality, and the flock filmed above the River Shannon contained thousands.
http://www.wired.com/wiredscience/2011/11/starling-flock/
http://www.wired.com/wiredscience/2011/11/starling-flock/
Yes, this is one of the first things I also noticed when I started looking at the markets. I found this not only applies to price/volume or volume, but to the profits per trade for pretty much any algorithm. This told me that you can try to bet on the high occurrence of a lot of small position trades (e.g., scalping) or you can try to bet on getting those occasional home run trades to make up for all the small losses (let your profits run, cut your losses short). This is essentially the trade-off between entropy and enthalpy: you can bet on high probability (entropy) but small gains, or relatively rare but high individual profits (enthalpy). The age-old problem of course is if you cut your losses short, you also potentially cut those extremely important home run trades since nothing goes up straight. The balance is quite exquisite and depends heavily on things like the recent volatility, and the commission and slippage -- perhaps this might be the advantage of prop firm vs retail, as retail commission is often too high to make that balance profitable using scalping or high-number of trades.
Regarding random walks and patterns, just my opinion, I believe the market is not a pure random walk, but it is more like the Langevin equation or Brownian dynamics, i.e., there is a fundamental deterministic component, and there is a random component. At times the relative sizes of each component varies, i.e., the random component can dominate, or the deterministic component can dominate, i.e., chop vs trend.
The purpose of this thread is to stimulate thinking. Who cares how long the thread gets... or whether market are random walks.
Do you think index components exhibit flocking behaviour (i.e. the 4 flocking characteristics) around the index?
Do you think index components exhibit flocking behaviour (i.e. the 4 flocking characteristics) around the index?
I've read ALL of the posts within this thread from the start. There have been some great interesting debates, minus the pissing match somewhere down the middle. Anyway, I have a question which pretty much makes the original question obsolete and should answer the most important thing
Here it is, especially for all you PhD guys.
Question:
In mathematical theory, is it possible to create an automated strategy that will be profitable on a random walk chart?
Assumptions:
We can assume we are running the random walk into infinity, and that the randomness of it cannot be predicted and doesn't follow any known model of random number generation (for example we use some unknown noise patterns for the value generation - like background space microwave radiation). By profitable I mean that if we run the random walk into infinity, the profits just keep increasing.
I think answering this questions answers a lot of others that have been raised during the discussions within this interesting thread.
Here it is, especially for all you PhD guys.
Question:
In mathematical theory, is it possible to create an automated strategy that will be profitable on a random walk chart?
Assumptions:
We can assume we are running the random walk into infinity, and that the randomness of it cannot be predicted and doesn't follow any known model of random number generation (for example we use some unknown noise patterns for the value generation - like background space microwave radiation). By profitable I mean that if we run the random walk into infinity, the profits just keep increasing.
I think answering this questions answers a lot of others that have been raised during the discussions within this interesting thread.
Yes, assuming you are trading against humans, and not machines. And assuming you don't plan to live forever, and will exit the market entirely at some point.
Different definitions of "random" may need different strategies, of course.
Yes and no. There are no games that could be constructed on the 50/50 coin toss type of a process that have positive (or negative) expectations in the long (indefinite) run. The expectation for any strategy based on a 50/50 coin toss game is always zero (providing of course that we are not dealing with a bounded walk or have limited budget etc.) However, there are unlimited number of games that one could construct on BIASed (non 50/50) games as well on any random walk type of a process that have other than normal distribution. Thankfully, markets are not normally distributed; they are log-normal at best. It creates the opportunities to create strategies that exploit the market's âabnormalityâ in a very stable fashion.
