This is a mobile optimized page that loads fast, if you want to load the real page, click this text.

198% CAR - too good to be true?

Jump to new
G

gooloo

Joined
22 July 2008
Posts
10
Reactions
0
so I've had 1000's of "eureka!" moments over the past 12 months of playing with trading strategies and backtesting... all of them have ended in disappointment.

I've recently come up with a system that returns -

start - 01/01/1995
end - 01/06/2008
total trades - 16718
CAR - 198%
Max Sys DD - 11.3%
% winners - 29%

This is with no optimisation (there really aren't that many paramaters to optimise).

I'm not posting this as a "look how good I am"... rather as an exercise in what can I possibly have done to fudge these results???? I have spent the last month trying to disprove these figures with no success.

I'm using 10% position sizes, 1% max loss stops(intraday), and 45percent trailing profit stops(intraday). entry and exits are on next days open. I'm not forward looking in any of my code. Universe is entire ASX. Data is from premiumdata and I assume clean. What else should I be checking???? How can I break this thing?

Am I too pessimistic in thinking that returns such as this are completely ridiculous and I've fudged something, somewhere? I would have considered any kind of CAR above 30% with drawdown of less than 15% to be exceptional. What % values do people consider a "good" system?

Help me crush my dreams
 
yep... daily bars. can you clarify your red flag? is it based on a "if it seems too be good to be true it IS too good to be true?".... because that's what I'm basing mine on.
 
gooloo said:
yep... daily bars. can you clarify your red flag? is it based on a "if it seems too be good to be true it IS too good to be true?".... because that's what I'm basing mine on.
Click to expand...

The two figures seem to be incongruent in the context of a mechanical system.

But not an expert in mech systems, I'll leave the in depth stuff to others.
 
what commission/spread are you using on the 16718 trades??

Cheers
...........Kauri
 
those figures are based on zero trade costs...
If I use $15 per trade the figures end up -

total trades - 16703
CAR - 185%
Max Sys DD - 11.3%
% winners - 29%

I'm not sure if that's what you mean by commission / spread?
 
How will 198% capital adequacy ratio make you money.What has that got to do with how profitable your system will be?Could you explain this?

CAR = tier 1 capital + tier 2 capital
risk weighted assets​
 
I`ve been reading Tharp and he says that R multiples are what is important in determining the expectancy oy each trade.

For example ..

Average profit = net profit ÷ number of trades.

Expectancy = average profit per trade ÷ average loss. (___R)

Your system has a high loss % so the winners would have to be real beauties with your position size of 10% cap. and 1% stop loss.
 
That works out at an average of about 5 trades per day. Pretty high I would have thought for an EOD system. And I would expect brokerage to make a bigger dent than that with so many trades.

My suspicion would be look-ahead code, even though you say there isn't any. Also, just as a comparison, try changing to buy on high and sell on low.

GP
 
I have been doing some reading and came across an interesting point about how ROA % are calculated.

Q. How is the return calculated with your system?
 
Have you set your trade delays adequately so that you enter and exit your positions the day after you get your signals and NOT on the same day ??
 
thanks everyone for your thoughts - i'll try and answer all questions below....

by CAR i'm referring to cumulative average return.
the universe is the entire asx
i have a volume filter set - the 20 day volume average must be greater than 50,000

trade delays are set to enter and exit on next days open, stops are intraday
with no brokerage the results are -

start - 01/01/1995
end - 01/06/2008
total trades - 16718
CAR - 198%
Max Sys DD - 11.3%
% winners - 29%

with brokerage at $15 per trade the results are -

start - 01/01/1995
end - 01/06/2008
total trades - 16703
CAR - 185%
Max Sys DD - 11.3%
% winners - 29%

with trades set to buy on next days HIGH (rather than on open) and sell on next days LOW (rather than on open) i hit 100% drawdown.

for interests sake.... some other stats from amibroker's backtest report -

Recovery Factor 17.15
CAR/MaxDD 17.58
RAR/MaxDD 21.22
Profit Factor 2.27
Payoff Ratio 5.56
Standard Error 23865153571.62
Risk-Reward Ratio 0.14
Ulcer Index 1.94
Ulcer Performance Index 99.72
Sharpe Ratio of trades 2.17
K-Ratio 0.0092

the standard error seems a big number - i'm struggling to find a definition as to how this is calculated and how to interpret it.
 
A few thoughts:

(1) Are these results using In-sample data? How do they look using Out-of-Sample data? ie. Try a walk-forward test on out of sample data.

(2) Check the min. tick size you have set in the AA settings. Having a min. tick size of say 1c can greatly skew results if you are exiting intraday on penny stocks.

(3)
i have a volume filter set - the 20 day volume average must be greater than 50,000
Click to expand...
50,000 is quite low imho if you are trading the entire ASX. ie. 50000 shares of a 5c stock isn't much volume....actually 50000 shares of a $2 stock isn't much either! Try it with C*V>250000. This works well for me using a breakout system. If you prefer higher longer term volumes then MA(C*V,20)>250000 would be worth a try.
 
gooloo said:
i have a volume filter set - the 20 day volume average must be greater than 50,000
Click to expand...

What have you set under Backtester settings - Portfolio - "Limit trade size as a % of entry bar volume"?

Also, is $15 fixed brokerage realistic?

Also, if you change to buy on the high of the day and sell on the low then the system sounds sensitive slippage.

I'm not sure how advanced your Amibroker skills are, but you could try writing code the introduces slippage as a % of the distance between open and high (on entry) and open and low (on exit). Then you can optimise on this variable and see how sensitive the system is to slippage.
 
thanks guys...

As I haven't optimised anything I don't get much difference on in-sample or out of sample data. The results average around the same no matter the time period (from 1995 onwards).

the tick size is set to 0 (means no minimum)

If I change the volume filter to be MA(C*V,20) > 250,000 the results are -

start - 01/01/1995
end - 01/06/2008
total trades - 13336
CAR - 21.7%
Max Sys DD - 17%
% winners - 31%

If I set the volume filter to C*V > 250000 the results are -

start - 01/01/1995
end - 01/06/2008
total trades - 13736
CAR - 30%
Max Sys DD - 16%
% winners - 31%

An improvement on percentage of winners but a massive reduction in CAR and a jump in maxDD. I can play around with optimising the volume threshold but I'm guessing optimisation is always going to push this number lower.

limit trade size as a % of entry bar volume is set to 0 (no limit)

$15 per trade may be a touch low - once I confirm brokerage rates for IB I'll change this.

I agree it does seem very sensitive to slippage.... Given that around 30% of profit comes from the trailing profit stop I think this is definitely a factor. Also add to this the average trade is only 3 days, AND the max loss is a measly 1%.
an interesting idea on coding in slippage calculation - need to think some more about how I code this.

thanks everyone for your responses - giving me a lot to think about and a lot of ways to break my system

keep it coming!
 
My guess would be slippage. As suggested before, filter out by dollar volume traded (at C*V at least 100,00) and try run it over a different interval.
 
You must log in or register to reply here.
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more...