Is trend your friend?

[oxusmorouz]

For metastock users, I'm posting the equity curve indicator:

Equity Long $:
a:= Buy condition;
b:= Exit Condition;

d:= BarsSince(a) = 2;
f:= BarsSince(b) = 1;

g:= BarsSince(d) < BarsSince(f);

i:= ROC(C,1,$);
j:= If(g,i,0);
k:= Cum(j);
k

Equity Short $:
a:= Short condition;
b:= Exit Short condition;

d:= BarsSince(a) = 2;
f:= BarsSince(b) = 1;

g:= BarsSince(d) < BarsSince(f);

i:= ROC(C,1,$);
j:= If(g,-i,0);
k:= Cum(j);
k

Equity total $:
a:= Fml("Equity Long $");
b:= Fml("Equity Short $");
a + b

I coded this formula myself, after being quite upset with metastock's backtester. You can incorporate brokerages by using the "valuewhen"
function. Longs, shorts and exits are based on next day's close.

 

[beginner_av]

Hi Oxy,
Very good work done. Yesterday I was chatting with asishda who was stressing on statistics and edge for systems. Since you are an expert in systems development now, can you tell me what will you consider to be a good system?
Remember Sir, I am not looking at basic stuff like edge, statistics, T-tests, Z-score for significance/confidence, runs for dependency etc. I need to know what constitutes a good system and how do we take that into account wihile developing.

 

[aroymkp]

Well according to weekly charts - One would have entered Dow around 2875 in 1990s and exited around 10294 in 2001

Rentry 9337 in 2003 and long till now.

Can't understand why you feel it is not rewarding

 

[kkseal]

Asishda, adding to your statement of 'seeing the yellow (early) before the red' i'd say not only before the red but before the green as well. I think it's the power of anticipation that separates the best from the average in this trade (Don't know if that can be statistically tested). Our methods/systems need to be honed in a manner that allows us to anticipate better & better.

Regards,
Kalyan.

 

[beginner_av]

Asishda, adding to your statement of 'seeing the yellow (early) before the red' i'd say not only before the red but before the green as well. I think it's the power of anticipation that separates the best from the average in this trade (Don't know if that can be statistically tested). Our methods/systems need to be honed in a manner that allows us to anticipate better & better.

Regards,
Kalyan.

Ireally envy you guys...the way you can see red before yellow and even green too. I'll never be able to compete with you on that front. particularly after reading mark douglus - the one thing that i taught myself was *not to predict the markets*


btw there's a nice article in TASC as to what happened to the Cycle bears who shorted the Dow based on their traditional cycle theory.

 

[oxusmorouz]

Hi Oxy,
Very good work done. Yesterday I was chatting with asishda who was stressing on statistics and edge for systems. Since you are an expert in systems development now, can you tell me what will you consider to be a good system?
Remember Sir, I am not looking at basic stuff like edge, statistics, T-tests, Z-score for significance/confidence, runs for dependency etc. I need to know what constitutes a good system and how do we take that into account wihile developing.

LOL, I don't believe I even qualify as an amateur system developer.

I have formed my own (probably lousy) observations about system development BAV. A lot of importance is given to ex-post numerical factors such as profit factor, drawdowns, % wins etc. However, personally, I feel (with my limited knowledge) that any system which gives a positive "utility" (in quantifiable terms) is anything worth considering.
Let's take an example of a trader who trades for a living.
He incurs the following costs:
Software Depreciation = 5,000Rs
Data = 7,000Rs
Hardware Depreciation = 3,000Rs
Opportunity cost of lost salary = 50,000Rs (depends)
Internet costs and misc = 3,000Rs
Brokerages (fixed, assuming no STT) = 2,000Rs
Opportunity cost of lost interest = .5% * Capital (Let's say capital = 10lakhs, so interest forgone by trading = 5,000Rs a month)

Total cost incurred = 75,000Rs

Now, deduct 75,000Rs at the beginning of each month from your capital. At the end of the period, determine if the closing capital is greater than starting capital, and determine the number of months when closing capital is greater than opening capital. A positive figure in the former and a "satisfactory" % in the latter is a justifiable evidence that the system has yielded a positive utility. A poor performing system will go bankrupt, net of costs, very soon.
I haven't used much statistics in this analysis and approached it from survival terms. Hope there isn't any blunder committed.
Looking for expert comments.

 

[oxusmorouz]

Well according to weekly charts - One would have entered Dow around 2875 in 1990s and exited around 10294 in 2001

Rentry 9337 in 2003 and long till now.

Can't understand why you feel it is not rewarding

Please justify.

 

[beginner_av]

Well according to weekly charts - One would have entered Dow around 2875 in 1990s and exited around 10294 in 2001

Rentry 9337 in 2003 and long till now.

Can't understand why you feel it is not rewarding

The whole premise here is that you do not have the perfect (ideal == impossible) equity curve of buy low and sell high, but a TF system that stays in the market all the time and tries to take any breakout/beginning of trend based on popular public methods available. And then compare the same with B&H.
The whole idea is to see what kind of outcome is there. Once you have historical data and hindsight you can beat it to death till you have the best returns that even Eckhardt, Seykota and Tudor Jones will also envy.

 

[alokdaga]

Well - I agree that on weekly timeframes and proper money management - atleast for entry stops - the longer term trends are better to follow. Some scrips like have been buliish since 2001 till now and have never given exit signals - advancing from 10-15 bucks to 1000+ now. Dow charts have not seen but i am sure weekly/ monthly time frame can be very rewarding

 

[beginner_av]

The problem is how do you know that the trend will be longer term when you enter? If your determination is incorrect, with weekly charts, you lose lot of money on your stops. And we are talking of active trading. IF you want position trading for over 6 months etc where you want to buy out of a long term basing pattern, you can definitely use weekly or even monthly charts.