Exits not more important than entries?

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
babelproofreader
Roundtable Knight
Roundtable Knight
Posts: 138
Joined: Wed Nov 10, 2004 4:36 pm

Exits not more important than entries?

Post by babelproofreader » Sun Aug 23, 2009 1:46 pm

I am sure many are familiar with the adage that good exits are more important than entries, even to the extent that a random entry coupled with a good exit can be a viable entry/exit system. Personally I have always thought that a good entry is something worth working towards in the belief that a good entry is also a good exit in the stop and reverse sense. As a result I have spent some time and effort on developing what I think is a good entry and post a chart of EOD cumulative tick returns for the S & P 500 resulting from this entry and invite any comment.

The entry rule is quite simple - enter at the open the day following a signal, exiting any position in the contrary direction. The signal is simply that two indicator lines are both greater than their previous levels for a long, or both less than their previous levels for a short, otherwise remain in the current position. A simple stop and reverse rule set.

For comparative purposes I also tested with a simple protective entry stop GTC which does not move, the entry in this case being the same as above except that no entry will be taken if the open is at or below this stop level for a long or at/above this stop level for a short, in which case any current position is maintained.

The chart below shows the cumulative tick returns for the back adjusted S & P 500 futures contract with assumed costs of 8 ticks per position, 4 ticks being taken off the tick return for the day of entry and another 4 taken off on the exit day. For comparative purposes the cumulative tick return for a "buy and hold" strategy is also plotted.

Some simple summary statistics
Long only
Total tick return 69782
Max DD in ticks 1135
Trade Win percentage 81%
R Multiple expectancy 0.83
Average win size 1.12 R
Average loss size 0.42 R
"Van Tharp's" SQN 7.9

Short only
Total tick return 69147
Max DD in ticks 770
Trade Win percentage 72%
R Multiple expectancy 0.89
Average win size 1.43 R
Average loss size 0.52 R
"Van Tharp's" SQN 5.5

Long and Short combined
Total tick return 138929
Max DD in ticks 1513
Trade Win percentage 76%
R Multiple expectancy 0.86
Average win size 1.27 R
Average loss size 0.47 R
"Van Tharp's" SQN 6.3
Attachments
S&P.png
S&P.png (118.11 KiB) Viewed 4368 times
Last edited by babelproofreader on Thu Aug 27, 2009 12:45 pm, edited 2 times in total.

Roger Rines
Roundtable Knight
Roundtable Knight
Posts: 1946
Joined: Wed Oct 06, 2004 10:52 am
Location: San Jose, CA

Re: Exits not more important than entries?

Post by Roger Rines » Sun Aug 23, 2009 2:03 pm

Interesting example. If there is a spreadsheet, consider adding that to your posting so the details can be understood.
babelproofreader wrote:Personally I have always thought that a good entry is something worth working towards in the belief that a good entry is also a good exit in the stop and reverse sense.
I never understood the need to highlight an Exit strategy by depreciation the Entry. It isn't possible to have a reasonable system without both components, plus all the other logic strategies a solid system requires, so why project an impression that an Entry isn't a meaningful component. Further more, why wouldn't someone want to do their best with all the various logic strategies, including the Entry process, in a system to ensure that the results produced were the best possible way of supporting their speculative endeavors?

It is good to see support appear for the Entry aspect of trading system.

babelproofreader
Roundtable Knight
Roundtable Knight
Posts: 138
Joined: Wed Nov 10, 2004 4:36 pm

Post by babelproofreader » Sun Aug 23, 2009 2:52 pm

Roger,

Unfortunately there is no spreadsheet available as I do all my development/coding etc. in Octave. I gave up using spreadsheets (Excel) for trading purposes some time ago.

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Sun Aug 23, 2009 2:53 pm

I'm not so sure that stop-and-reverse systems are obviously and unambiguously exit-free. There is equal justification to claim that they are entry-free instead, since these systems exit at the same time they enter, and enter at the same time they exit. They are symmetric: entries and exits are one and the same thing. Resembling Schrodinger's half-dead cat, these systems are halfway entry-free, and halfway exit-free (in my opinion).

I myself would be more convinced that "X doesn't matter at all" if lots of different choices for X were tried, giving the same result every time. It doesn't matter what X you choose, you always get the same result.

Or, if one wished to demonstrate the opposite, namely that "X doesn't matter at all" is completely untrue, I myself would be more convinced if lots of different choices for X were tried, giving wildly different results. I.e., some X's are dramatically better than others, some X's are dramatically worse; therefore it does matter a great deal which X you select.

babelproofreader
Roundtable Knight
Roundtable Knight
Posts: 138
Joined: Wed Nov 10, 2004 4:36 pm

Post by babelproofreader » Tue Sep 01, 2009 8:17 pm

As a follow up to my original post I thought I would put up another graph with a slightly different entry. The signal to enter is exactly the same (the indicator lines both being > or < previous levels) but instead of entering at the open the next day the entry is on a stop placed 1 tick above the 2 bar high for longs and 1 tick below the 2 bar low for shorts. Personally I am quite surprised at the difference that a seemingly small change makes, and this reinforces my belief that the search for an optimal entry is a worthwhile pursuit.

p.s. The title may be slightly misleading - to be read as entry as described above, with and without a protective stop.
Attachments
S&P2.png
S&P2.png (175.8 KiB) Viewed 4161 times

Asamat
Roundtable Knight
Roundtable Knight
Posts: 175
Joined: Fri Jun 03, 2005 7:50 am
Location: Walldorf, Germany

Post by Asamat » Thu Sep 03, 2009 2:49 pm

In the end, I think it is useless to try to disassemble a system in separate pieces like entries, exits, portfolio choice, MM, ... The pieces are not sufficiently independent in order to learn much from that. The system as a whole works, but asking whether it works because of the entries, because of the exits, or because of this or that leads to nothing useful. It is like asking whether a car is working because of the wheels or because of the engine. It works if you have both, and gas and oil as well, but not otherwise. You may adjust the slider of horsepower, you may play around with the size of the wheels, but you still need a certain type of engine and a certain type of wheels, in order for the car as a whole to work.

Not every entry works within the context of a certain system or system type. You can not use a rocket engine in a car simply because it is also called an engine. And you can not use a sufficiently different entry simply because it is also an entry.

This does not mean it's useless to compare different entries, or useless to optimize certain aspects of the system. But I don't think the components are sufficiently independent to have a meaning on their own without the other ingredients of a system present.

RedRock
Roundtable Knight
Roundtable Knight
Posts: 939
Joined: Fri Jan 30, 2004 3:54 pm
Location: Chicago

Post by RedRock » Thu Sep 03, 2009 5:29 pm

Asamat wrote:I You can not use a rocket engine in a car simply because it is also called an engine. .
Depends on your goal... eh
http://tinyurl.com/The-Blue-Flame
:)

cryder
Roundtable Knight
Roundtable Knight
Posts: 101
Joined: Tue Jul 22, 2008 5:02 am
Location: Melbourne, Australia

Post by cryder » Thu Sep 03, 2009 5:50 pm

An adaptive system. :D

Post Reply