Apr

4

We have had numerous discussions on this venue regarding stop losses. Part of the surprise from those discussions is that using a stop loss will double your odds of having a loss in the amount of the stop loss.

However the same is true for a profit target. Using a profit target will double your probability of having a gain equal to the target gain. The reason for both phenomena is that in a random walk half of all such trades will get reversed after hitting the target or the stop. The fancy name for this is the Reflection Principle.

Larry Williams writes: 

In a random walk, half of all stops/targets get hit, so if that is not true in several trading systems, does it suggest the market is not random?

Anatoly Veltman writes: 

Electronic markets are far from random. Your broker's HFT frontruns your orders, and non-broker largest HFTs parallel run your orders. Thus your limit (profit-taking?) order is played against by unabling, and your stop-loss order is played against by triggering. Random? Not to your account.

Ralph Vince asks: 

But can non-random ticks, sampled on a bigger time frame, degenerate into randomness?

Anatoly Veltman replies: 

In the sense that all those orders, magnified by HFT mechanism, will carry markets somewhere - sure. The other question is: OK, so 70% of executed trades resulted in robbing the outsider spec - but the HFTs and the brokers have not fully benefited by your loss, because of their high overhead (the arms race, et al). So ok, the wall street salaries, the IT salaries get financed out of your pocket. Then the only way to keep you in the game is to inflate your remaining funds…So the mechanism will continue on…but to what end, if the economy is not picking up? So the result may well be non-random: all prices will go up.

Gary Rogan writes: 

Clearly the natural drift and/or inflation-driven accelerated drift will result in an upward bias that will make a random walk impossible. In addition, if there is an HFT-induced tendency to hit stops and not hit limit orders (by the way are there any objective statistics that prove that?) the question becomes: would an independent observer looking at the data tick by tick, but who is not himself placing limit/stop orders be able to tell that the statistical nature of the tick distribution has changed?

Jeff Rollert says: 

No, HFT is attacking your behavioral biases. Not the academic ones ones. Your bids show your hands.

These are modeled after high yield bond trading patterns.

How would you trade if the book was open and public? That is the point. Trading systems are rational, and your systems are easy prey…seriously, inject the random. To borrow a sports analogy, you can't bore a machine into an error.


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search