[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Testing Exits (Wanring: Long Dull Post)



PureBytes Links

Trading Reference Links

Using Dave DeLuca's Random Trader code, I've been testing various exit 
strategies and their properties. I'm using daily data for IBM from 
1/4/84 thru 3/17/99. Here are the exits I've tested:

After a random entry:

1. Exit each trade after 11 days
2. Exit using Chandelier Standard (using 3 ATRs)
3. Exit using Chandelier Thrust (using 3 ATRs)
4. Exit using Parabolic (as coded in TradeStation)

Here are some of the test results:

Exit No.     	   1         2          3        4
No. Trades  	 4068	   3909       4840      2288
Avg Trade Len      11        17         8         11
Avg MAE (pts)    -2.39     -2.17     -2.01     -2.17
Avg MFE (pts)     2.36      3.99      2.08      3.01
Max Equity (k$)   1.6      195       11.2      250
Min Equity (k$)  (181)     (18)      (70)        (9)
End Equity (k$)  (152)     176       (51)       245
Profit Eff (%)   63.4      42.4      80.3       48.6
Loss Eff (%)     63.2      75.0      79.6       68.8
Win %            48.2      37.6      50.8       40.3      

Each trade includes comm/skid of $60.

The Chandelier exit is Chuck LeBeau's posted here previously. It 
basically exits a long trade after price has retraced 3 ATRs from the  
highest H since entry. The Chandelier "Thrust" reverses the logic 
somewhat (attempting to minimize profit giveback) by exiting (longs) at 
the lowest L since entry PLUS 3 ATRs. The parabolic exit uses the canned 
TS indicator to exit (longs) when price drops thru the parabolic price 
calculated using the standard accelfactor of 0.02. 

MFE and MAE are maximum favorable excursion and maximum adverse 
excursion as written about by John Sweeney.

Profit Eff is the profit efficiency calculated as (for winning longs):
(Exit Price - Entry Price)/(Highest H since entry - Entry Price) * 100%
Higher numbers indicate less profit giveback.

The Loss Eff is loss efficiency calculated (for losing longs):
(Entry price - Exit Price)/(Entry Price - Lowest L since entry) * 100%
Lower numbers indicate the exit is making the best of a bad situation.

These results tell me:
* in terms of maximizing equity, the best exit appears to be the   
  parabolic.

* the Chandeleir exit does do what it's supposed to ie. catch the big
  moves. It's MFE was by far the largest of the group. It's also very 
  profitable.

* the Chandelier Thrust does do what it's supposed to ie. try to     
  minimize profit giveback. HOWEVER, as can be seen by the ending
  equity (poorer than standard Chandelier), minimizing profit giveback
  doesn't do much for the bottom line. It's nice psychologically to 
  avoid profit giveback, it much more important to stay with the big
  moves. This is what the big boys always preach, eh?

* winning % isn't important. The exit w/ the highest win % lost $.

* the parabolic is able to minimize losses as well as the Chandelier. 
  Additionally, it captures profit better (higher profit eff) and gets
  out of losers sooner (smaller loss efficiency).

Comments/criticisms are welcome.

As I get time, I'll code the parabolic exit with Wong's entry and post 
the results. Since Wong's system only entered long trades, I'll include 
a short entry. I'll also want to test the resulting system on other 
equities.

Regards,
Bill Vedder