System33_v5 (divergence signal only)
Systems home page Support06BUGS New ideas page Latest trades Contact Bookshop Trivia
Fixing the bug in the stop by replacing it with the 07ATR_Shortexit and 07ATR_Longexit produced the following results.
SYSTEM33 TEST 5 original ENTRY with new stops
07ATR_Shortexit and 07ATR_Longexit
| Score | Interval | Trades | Profit | Win-lose | Risk-R | Max DD | Ave win | Ave loss | Best win | Worst loss | Profit/DD | Equitycurve |
| A++ | 1250 | 153 | 3804 | 41.1 | 2.79 | -512 | 123 | 44 | 559 | -114 | 7.4 | COOL!! |
BUG1 (that were corrected with new stop)
While the pj stops are generally good, they are 15 instances where they are exiting short 2 bars after entry
without having moved anything like 4 atrs away from the price action.
The stop has 2 features which are perhaps redundant, and these include the acc (acceleration factor) which is allowed to
tighten the stop by a maximum of 1 atr in 20 steps. Its never been tested by me if a stop is best to grow tighter, grow wider
or to stay the same number of points or atrs away as the position improves.
Also the stop has a standard deviation of atrs added to it, which again can probably be just replicated by increasing atrs
by 0.25 or 1 or so.
So in this study, I shall rebuild the stop from new, and remove the acceleration factor and the std devs from the code, re writing
it in my own language. REPAIRED (see 07ATR_Shortstop tested above)
BUG2
Which is not actually a bug, but rather a logical error whereby the long position is closed out at the same time as a divergence
bar is signalled to buy it again.
This adds 3 points commission on each occurrence and is detrimental to overall profits. Unresolved
A maximum loss stop needs to be very intelligent, and as of yet I still haven't found a solution to this
The first bug will be addressed first and the new stops shall be called 07ATR_Longexit and 07ATR_Shortexit.
Simplicity is the main cornerstone of the design, and I will start with ATRS set to 3.5 which seems to be around the optimum
for a 1250 tick FT-SE Chart.
Settings to start will be as follows
ENTRY SIGNAL = SYSTEM33VARIAB, WHICH HAS THE NEW SYSTEM33 SETTINGS;
SYSTEM33 TEST 5 original ENTRY with new stop 07ATR_Shortexit and 07ATR_Longexit
and allows trades to run out before new trade is taken, massive improvement with R-R
by doing this.
Settings
|
1. System33 the original system without modification will retain this name, any modifications to this will carry a version number such as system33.2 for ease of understanding, and all changes will be noted on these pages.
SIGNALS included in this system are :-
For the ENTRY
1. 06system33divergence (using divergences of the tradevolumeindex indicator) (DEMBUY)
For the EXITS
07ATR_Longexit 07ATR_Shortexit
Commissions for FT-SE and DAX will be set to 3 points for ALL TESTING, 1 Contract only.
DATA Used
H.S FT-SE100 Futures (with many gaps) so this test is for preliminary testing only
DATA-type TICKCOUNT!
First date November 13th 2001 Last date August 17th 2007
TESTING WITH EACH SIGNAL PRODUCED THE FOLLOWING RESULTS
TEST 5 Improvement ideas.
Consider using a more intelligent stop method, similar to spoostops?
Optimise stop in the most promising time frames TRY A NEW STOP!. TRY A 20-60 POINT ROLLING STOP WITH WIDE ENTRY, LIKE MY CURVED STOP!!!
Also have problem of exiting longs when a new long is triggered straight away.
|