Use random walk to make $$$
- Thread starter Thumama
- Start date
Quote from MAESTRO:
If I tossed a coin and it landed 12 times on a row showing heads does that mean I am in a"head trend" ?
I think the mighty marketsurfer coined that phrase.
You should give credit to the source, he has some nasty lawyers.
H"L!J
Yes, brilliant.
well stated.
HLJ!
I had a hard time trying to follow what you are actually suggesting, since you seem to be mixing terminology IMO, and your thesis is not very clear.
However, I will attempt to decipher and humbly comment from my interpretation.
2. If the value of the variance ratio test indicates a nonrandom behavior of the time series, extend the time series by adding values ( future prices) that will force the variance ratio test to indicate a random walk. (revert to its expected value)
If the value of the VR indicates it is a non-random walk process, generally that implies it is a mean reverting process (VR<1) or a trending process (VR>1). Intentionally adding time series data to force it to revert to its mean, (if your VR test has already classified it is a mean reverting process) is not the same as forcing the VR back to one (random walk state) . If your VR test indicates it is mean reverting, and you believe this process will continue into the near future, then you should simply expect it to revert to its expected value (and bet according to your expectation).
3. Try this for different timeframes and on average you get some abnormal returns.
In theory, you could come up with many hypothetical scenarios that would have outperforming returns. The problem is that conjecture does not equate to fact.
However, if your research demonstrates to you that prices are mean-reverting, than rigorously (emphasis on rigorous) back-test and forward-test that hypothesis on the instruments that demonstrated that behavior, and devise a strategy to profit from it.
There are plenty of ways to argue that markets are random or not; which further lead to many more discussions about the actual definition of random (unfortunately, there are also many classifications of randomness, which leads to proliferation of confusion on these types of threads). But, generally speaking, I tend to agree with Brandon and Maestro's comments on approaching the markets; It's best to assume that markets are random in the simple sense that you should not try to predict the future outcome with any type of certainty, and devise a way to profit under that assumption.
Cheers
... and there you have it.
dtrader98
Sorry if my statement was not clear.. English is NOT my first language
My proposed idea is based on the assumption that returns (with different time frames) follow a random walk process ( VR=1 )
What I meant is to add data points (future returns) to the time series so that the new VR equals 1.
Thumama
Using the word random leads to confusion, random means absolutely unpredictable; but the market seems to be 50+ in favor of predictability for long term direction; but for the price itself at a particular point, as ProfitTakgFool mentioned at support/resistance, we all agree that it is absolutely unpredictable. In brief, we can say something like "I think the price will be going up/down but I don't really know exactly where it will be closed."
Quote from MAESTRO:
If I tossed a coin and it landed 12 times on a row showing heads does that mean I am in a"head trend" ?
Yes, it actually does mean we are in a heads trend, if and ONLY IF, other people PERCEIVE it to be a trend. You make money from people entering at various stages of the perceived trend. Eventually someone gets left holding the bag as we revert to the mean, and the process starts again, but not necessarily in the same direction.
When you toss a coin obviously the odds are 50-50, but when you have a market it is not 50-50, because there are high percentage areas where momentum investors will enter the market. There are high percentage areas where stochastic investors will enter the market. There are high percentage areas at s/r etc. etc. etc.
The trader that can trade in THESE better than 50-50 scenarios will typically be more consistent.