Discussion in 'Strategy Development' started by ktmexc20, Jul 20, 2004.
I pose this as a question.
Actually my question is... in your opinion to what degree?
why not? bet small, be patient, don't ask for too much. rinse, lather, repeat.
If you trade noise, how do you handle big unidirectional moves?
Go with the flow.
Yes. The problem occurs when you trade what you think is noise and it subsequently turns out to be a signal
bet small, have some insurance. what else is there to do?
If what you mean by noise is a stock that moves randomly, then no way, it can't be done profitably. Where's your edge?
Could you make money betting on the rolls of the dice, if you're only given 6:1 odds for a single dice roll? It can't be done in the long run, and it's the same issue.
What noise? White or gaussian noise offsetting price? That you can trade because by definition the outliers will mean-revert and are profitable to scalp or counter-trend trade.
The more serious type of noise is caused by participants buying and selling randomly. In that case price acts as a random walk, and there is no reversion to the mean or persistence of trends. Of course price isnât always a random walk. There are plenty of trends and rangesâ¦ in hindsight. Itâs just that the random walk component makes trends and ranges difficult to detect until much opportunity is lost.
With a pure random walk there can be no directional edge, no forecasting. Price has an equal probability of going up or down starting from any point. Stops canât work because there can be no expectation of a position getting worse or better, regardless of the PnL.
If you don't have a net-positive system or method, prices will seem identical to a random walk. Your PnL will also be a random walk or worse because of commissions, spreads and human nature.
The challenge is finding and trading patterns despite noise. Sometimes it's impossible to find patterns for a given day, market, time frame or situation. Thus it is best to both survey a range of markets and situations and improve your methods. It is foolish to advance your methods on only a few markets, and foolish to survey many markets with weak methods. Regardless of where you look, patterns must repeat over time to be useful. If they donât repeat over time, there is no way to distinguish them from noise.
Patterns need not forecast direction. You can make money forecasting direction, volatility, volume, inter-market deviations, and derivatives of those and other properties.
I appreciate the attempt at depth in your answer.
My personal belief is that forecasting and edge trading is fallible. I don't go there any more... but that could be a discussion for another thread.
See right now I'm trying to associate the attributes of lissajoux to the PV relation. This is a new concept for me, so I thought by posing the ? about noise, it might help me articulate it's depth with in fine monitoring. If you can help .... I'd be much obliged.
Get me odds like that and you won't have to work again.
Separate names with a comma.