systems that work

What you say is basically true, however the historical record shows that reality does not correspond to statistical theory. In theory, events that are classified as "6 sigma" events should happen only about once every 5 or 6 decades. In reality, these events happen much more often. This discrepancy between theory and the historical record is described by the term "fat tails". This is one example of the limitations of statistical theory. Unfortunately there are other examples. I would never risk a third of my account on any single trade. In fact, I have specific rules for money management that limit my risk on a per trade basis and on a daily risk basis. I know from experience that others would laugh at my risk protocols calling me ultra conservative. On the other hand, after my first year trading, I never again experienced a drawdown greater than 8%.

Again, I have to repeat my comment, and I expect others to pass this right by. When price exhibits random behavior, your edge disappears. You may as well go to Vegas and play slots. The significance is that this limits the absolute number of opportunities (tradeable setups) that you have in a day. Also it means that you cannot risk big, because you cannot know when this will occur (until you are in the middle of it). I think I have said this a couple of ways now.

Good luck tomorrow
Lefty
 
Lefty,
I am intrigued. I presume you are detrending the data so as to eliminate any trend/cycle components, leaving pure intraday volatility/noise from which you are trading some form of range breakout.

Or perhaps the corollary. Discard the vol and trade with the prevailing trend or cycle tops and bottoms.

I'm not meaning to pry, just curious about your strategy at a conceptual level.

Also, which detrending method do you prefer?

TIA.

Colin
 
Hi Bolter:

The "classical" reason for detrending is to remove cyclical effects from serial data. There are other important benefits to doing this, especially for traders who are trying to "learn" a market.

First, if you do the detrending and analysis (distribution of moves by size and by temporal window) long enough, you get a "feel" for how your market acts. I do this for intraday data, and over time I have developed a feel for changes in volatility (the size of moves changes as you would expect) and a sense of the flow of the market during the day (the size of the moves implies changes in the order flow). As I have stated in earlier posts, with proper processing and analysis, you can see specific points in time that I call "temporal windows", where price moves in a non-random fashion. The example I gave is where I have found a temporal window whose identifying "landmark" is a move of x ticks followed by a pullback. Within this temporal window, if price resumes it movement, I knew there was an 80% probability that it would continue for at least 4 additional ticks. Once again (I think this is like 4 times) you have to understand that this edge disappears IF price cycles from stationarity to random distribution.
THIS is why a trader cannot (in my opinion) risk a third of their account on a single trade.

Good luck guys
Lefty.

Edit:

From a technical point of view, it doesn't matter what method of detrending you use. Practically speaking I like to subtract one price from the next. It gives me a feel for the way price moves during the day.
Because vol is a more reliable phenomena on the macro scale, I like to "embrace" it. I look for contractions intraday, and I wait. Just as a surfer learns the tidal fluctuation at his favorite surf spot, I have learned to identify tidal variation in vol.
 
At least few of us has the "skills" to develop profitable systems consistantly and constantly, than just owning a specific system worrying about the system fading. (Oh... along with it the skills to actually manage them)

First and fore-most, it's "skills"...

:D

:D

:D
 
Quote from Lefty62151:

Hi Bolter:

The "classical" reason for detrending is to remove cyclical effects from serial data. There are other important benefits to doing this, especially for traders who are trying to "learn" a market.

First, if you do the detrending and analysis (distribution of moves by size and by temporal window) long enough, you get a "feel" for how your market acts. I do this for intraday data, and over time I have developed a feel for changes in volatility (the size of moves changes as you would expect) and a sense of the flow of the market during the day (the size of the moves implies changes in the order flow). As I have stated in earlier posts, with proper processing and analysis, you can see specific points in time that I call "temporal windows", where price moves in a non-random fashion. The example I gave is where I have found a temporal window whose identifying "landmark" is a move of x ticks followed by a pullback. Within this temporal window, if price resumes it movement, I knew there was an 80% probability that it would continue for at least 4 additional ticks. Once again (I think this is like 4 times) you have to understand that this edge disappears IF price cycles from stationarity to random distribution.
THIS is why a trader cannot (in my opinion) risk a third of their account on a single trade.

Good luck guys
Lefty.

Edit:

From a technical point of view, it doesn't matter what method of detrending you use. Practically speaking I like to subtract one price from the next. It gives me a feel for the way price moves during the day.
Because vol is a more reliable phenomena on the macro scale, I like to "embrace" it. I look for contractions intraday, and I wait. Just as a surfer learns the tidal fluctuation at his favorite surf spot, I have learned to identify tidal variation in vol.
More...

Just found this thread. Interesting ideas Lefty. Thanks.

A question for you on the back of your post: would you expect a systematic trading system's returns to have more useful / predictive qualities than a market's returns?

AM
 
Lefty,
I find your post about "random behavior = lost of edge" very spectacular and thought provoking,

I Do however do not understand what you mean by detrending, what is this?

I often just look if we are high higher or low lower to see if we are on outside days, if we are ,then trend day, if not not.
 
You must log in or register to reply here.
Back
Top