jptann
Member
Posts: 101
Registered: 13-6-2017
Member Is Offline
Mood: No Mood
|
|
Some minor issues with GSB 30.0
Peter:
If I understand the methodology, you use commissions and slippage in the back testing optimizer to ensure that we don't get too many trades with small
wins and losses. I certainly agree with this, but it looks like the calculation of the performance chart and the performance metrics do not include
any slippage and commissions.
In my opinion, this will give you a false sense of security. I will go through the various studies by looking at the optimization charts, and without
S&C they could change dramatically.
Since in the day trading and even swing systems I notice that you don't use limit trades (that's a good thing) but you have to have a slippage
inclusion to make market trades realistic. Buy on open or close, etc.
This is the major issue so far with the program and makes it almost unusable for trading systems. Once I add S&C into TS or MC and review the
Performance and detailed charts, the performance is dramatically less than what I require and also does not reflect reality.
I think that if S&C are indicated they should be used for each trade or contract throughout the study.
One other "nit". A number of us use a SSD as part of our computer set up. It would be nice that we could change the default directory where we save
projects, programs, and other notes to a different drive than GSB. that is, I like my GSB on my SSD (I drive) but would like to save history files to
my E drive which is a normal HD drive. Some global settings for drives where we can save stuff
My next approach is to incorporate custom filters
Peter, you do need to take some time to do another on line tutorial on WF and the most recent set ups or have an updated manual.
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Working on the docs today. GSB updates have been so rapid, the docs are lagging.
Slippage and commission is very much a user preference. The 3 systems I provided with GSB have high profit per trade, so its not so critical.
ie GSBsys6ES = $185 profit less slippage and commission.
This is not at all what you are saying. "performance is dramatically less than what I require and also does not reflect reality."
You need explain your metrics in much more detail.
I think its best practice to use 1 tick in report + your brokerage. However my personal practice is to not do this, and its what Im use to.
I acknowledge that my personal practice is not a good one, but I look closely at the predictable years where slippage becomes an issue.
So I dont get the false sense of security, but I agree novice uses will. So best to use 1 tick each side in reports too.
GSB is also a bit faster if the commission and reports figures are identical.
However $15 per side commission is what I use by default on fitness slippage. The purpose of this is to make GSB produce systems with higher trades.
If you are building systems without secondary filter on, and then use secondary filter for WF only, its vip to have slippage set to zero on reports
and fitness.
(or at least until you do WF)
Most users are not doing this, or even aware you can do this.
Most apps dont allow change of paths. ie ts & mc. I will consider it, but you can also move GSB to any drive you like. GSB also doesnt use much disk
space.
GSB is not disk intensive at all, no no need to have ssd. Custom indicators is the only thing where TS writing files is very slow. Thats a once off
event though.
Is it just the systems that are saved path that you want to change?
|
|
|
jptann
Member
Posts: 101
Registered: 13-6-2017
Member Is Offline
Mood: No Mood
|
|
Peter:
Thanks for your reply. To expand on my explanation, be careful of average trade win. If the ratio of ATW to ATL (average trade loss) is close to 1
or lower, then slippage will matter in the long run. I have been trading ES daily since 1999 and over the last 18 years (including 2008,9) the
slippage has averaged 1/2 a tick per side or $6.25. My TS commissions are lower than $2.50 per side so I just use that. Should be sufficient for
most brokerages, at least TS and IB.
So, I include slippage and commissions in the fitness, do I put the same fitness and slippage in the report area also. This would then give me what I
want I think?
One other issue, I assume that the validation data is not used in any of the backtesting and walk forward. Is that correct? I want to make sure that
data is virgin and unseen during the development process.
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
I would keep the report fitness at $6.25 +$2.50 This is a very good slippage figure. Should be no need to adjust this.
You can then tweak fitness commission. I still like the $25 figure so I get less trades with higher average profit & PF
You are correct on validation. This is vip
Awesome you survived so long & hard years like 2005.
Its a different topic but has slippage changed over the last few years?
|
|
|
jptann
Member
Posts: 101
Registered: 13-6-2017
Member Is Offline
Mood: No Mood
|
|
Peter:
Actually, in the lower volatility environment the slippage is less. In trading a program I am running now in real time, I rarely get any slippage.
All orders are "next bar at open" and usually get the open price. Of course, like you I don't trade high frequency day trading and this is on a swing
system that is averaging about 2 trades per week or around 100 trades per year.
I will state that slippage is not an issue during the summer (your winter) but if volatility ever picks up, then slippage will come back. As an
example, the last 4 trading days in ES the range is about 10 points except for right after the opening. Hard to make any money in this environment.
I will make a suggestion in the "new feature subject". I have an idea that may be of interest to all of your readers. By the way, started trading
commodities in 1967 with my Dad. Have been trading ever since, but not full time till 2000 when I retired from the semiconductor world. Was trading
silver against the Hunt brothers in the 80's and did very well.
My late Dad started trading in 1933 as part of his metal business so I guess you would say that I have been following along for almost 100 years with
my Dad's work. Still have his "by hand" daily calculations back to 1950. Interesting to review them.
I have traded almost everything including silver, gold, copper, eggs, sugar, potatoes, pork bellies, soy bean complex, wheat, corn, oats, barley etc.
Now only trade electronic instruments and don't trade anything with low volume. My suggestion that I will include in the other forum works for all
commodities.
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Interesting bio Jim, Im sure you have lots to contribute to the forum. Im amazed your slippage is so low. I found a few years ago that it had
increased more than expected, so ended up doing smart orders to improve things.
|
|
|
![[*]](images/davis/default_icon.gif){kind=icon}
