Gamestudio Links
Zorro Links
Newest Posts
hold key code returns very high value instead of 0 or 1
by 3run
Today at 16:08
oldschool fps project
by 3run
Today at 15:46
Sam Foster Sound | Experienced Indie Game Composer for Hire
by titanicpiano14
Today at 15:02
Project Aum - Winter Contest 2017 - Update
by Ayumi
Yesterday at 22:34
[ALL STYLES] 2D Videogame Artist (pixel, vector, painted)
by DesixStudios
Yesterday at 14:45
Easy question
by jcl
Yesterday at 13:47
Cheap bouncing shells/gibs
by 3run
Yesterday at 00:36
AUM Magazine
Latest Screens
RPG PARTY
Triton Wing now available on Steam
AckCon'17: Lotter vs The World
Triton Wing
Who's Online
17 registered (Ch40zzC0d3r, Ezzett, Lobo, Hredot, AndrewAMD, 3run, alibaba, ChrstphFr, firecrest, 2 invisible), 454 Guests and 5 Spiders online.
Key: Admin, Global Mod, Mod
Newest Members
Amocfx, cme, pataba, byano, Tuguldur717
17834 Registered Users
Topic Options
Rate This Topic
#466158 - 05/28/17 02:23 1.58 Z3 Margin slider results volatility
DdlV Offline
Serious User

Registered: 06/10/13
Posts: 1296
Hi jcl.

I am updating to 1.58 this weekend. As part of my update process I Test with various Margin slider settings (Risk slider doesn't seem to matter any longer). I've noticed that as new versions of the strategy have come out that volatility has increased.

1.58 Z3 seems to exhibit much more volatility at different Margin than prior versions. At both Margin 49 and 51, for example, DD is > 1400 and AR is more like 70%. The other measures (PF, SR, UI, etc.) also bounce around similarly. Graphing these measures for various Margin slider settings does show a trend, but there is much more volatility around the trend than in past versions. Looking at the straight-line values at Margin 50 rather than the Margin=50 point values gives AR more like 100%, Cap Req'd more like 1200, etc.; indicating that using the Margin=50 point values of 130%, ~900, etc. in the Manual is rather misleading...

Do you see the same thing? If so, any idea why Z3's results now vary so much with different Margin slider setting? Past versions were substantially smoother...

Thanks.

Top
#466444 - 06/14/17 15:32 Re: 1.58 Z3 Margin slider results volatility [Re: DdlV]
DdlV Offline
Serious User

Registered: 06/10/13
Posts: 1296
Hi jcl. Welcome back! Would appreciate your thoughts on this and the other Z3 question when you have a sec. Thanks.

Top
#466480 - 06/16/17 16:31 Re: 1.58 Z3 Margin slider results volatility [Re: DdlV]
jcl Online

Chief Engineer

Registered: 07/22/00
Posts: 26002
Loc: Frankfurt
Thanks! The relation of Margin and profit is not linear. There are two reasons for this, lot size granularity and skipping trades when the margin is too low. Especially the last effect can cause a different AR at different Margin settings.

Top
#466500 - 06/18/17 16:57 Re: 1.58 Z3 Margin slider results volatility [Re: jcl]
DdlV Offline
Serious User

Registered: 06/10/13
Posts: 1296
Thanks jcl.

I understand the non-linearity. I was rather trying to get at 2 other points:

a) The volatility of Z3's Tests at different Margin values is greater now than it has been in the past. (I.e., its "VIX" as opposed to its trend.)

b) Because of a), using the Test results of a single Margin point (50) does not truly reflect how Z3 is expected to perform. In the case of the 1.58 Z3, the expected performance is actually lower than the Margin=50 Test results; but in future versions it could conceivably be higher.

Is this a correct perception & understanding?

Or are you saying that in Real Trading of 1.58 Z3 one should expect substantially better results using Margin=50 than if one used Margin=49 or Margin=51?

Thanks.

Top
#466515 - 06/19/17 10:05 Re: 1.58 Z3 Margin slider results volatility [Re: DdlV]
jcl Online

Chief Engineer

Registered: 07/22/00
Posts: 26002
Loc: Frankfurt
Since any update comes with different parameters and possibly other modifications to the Z systems, the margin volatility is unlikely remain constant.

Top
#466534 - 06/19/17 21:12 Re: 1.58 Z3 Margin slider results volatility [Re: jcl]
DdlV Offline
Serious User

Registered: 06/10/13
Posts: 1296
Thanks jcl. Would appreciate also your comments on the other aspects. To wit, is the manual's reporting of a point Margin's results OK, or should the (in this case lower) fit values have been reported? Probably more importantly, in Real Trading using the point Margin value should one expect the point results, or the (lower) fit results?

Thanks.

Top
#466552 - 06/20/17 19:48 Re: 1.58 Z3 Margin slider results volatility [Re: DdlV]
jcl Online

Chief Engineer

Registered: 07/22/00
Posts: 26002
Loc: Frankfurt
I assume that the data in the manual is ok as long as I have no information to the contrary. And in real trading, I would generally not expect to get the same results and volatility as in a backtest - unrelated to what margin you're setting.

Top



Gamestudio download | chip programmers | Zorro platform | shop | resources | magazine | manual | support faq | bugs

oP group Germany GmbH | Birkenstr. 25-27 | 63549 Ronneburg / Germany | info (at) 3dgamestudio.net