OmniTrader Forum OmniTrader Forum
forums calendars search
today this week
 
register logon control panel Forum Rules
You are currently browsing as a guest.
You should logon to access more features
A Self-Moderated Community - ALL MEMBERS, PLEASE READ!
Vote for Members who contribute the most to your trading, and help us moderate content within the Forums.


  Current location        Thread information  
OmniTrader 2017 Upgrade Forums
Portfolio Simulation Mode
Zero losses for 2008 yields 45% drawdown
Last Activity 9/2/2018 1:56 PM
21 replies, 1389 viewings

Jump to page : 1
Now viewing page 1 [25 messages per page]
 
back reply
Printer friendly version

^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : Zero losses for 2008 yields 45% drawdown
Posted : 4/21/2011 11:49 AM
Post #23798

I have been working on a longer term daily EOD strat lately and I'm somehow getting a huge disconnect from the trade-by-trade "view trades" report in the Port Sim vs. the Port Sim equity curve.

I'm now managing an absurd 84% HR over 88 trades and nine years over the NASDAQ 100. Win-loss ratio is an an equally absurd 5.8(average win: 7.9; average loss: 1.4). Yeah, right. In my dreams).

But beyond the nonsensical numbers, I am attaching "proof" that there were no losses for the period(all of 2008) where the Port Sim shows a drawdown of 45%. I am at a loss to explain any of it.

More wasted time and effort


[Edited by c on 4/21/2011 12:02 PM]

Attached file : zero losses yields 45% drawdown.gif (240KB - 190 downloads)
Attached file : zero losses port sim.gif (312KB - 213 downloads)

^ Top
Jim Dean

Idol
2000500100100100252525
Posts: 2884

Joined: 9/21/2006
Location: L'ville, GA

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/21/2011 12:08 PM
Post #23799 - In reply to #23798

Hi, Charles:

This sounds "gazillionly" familiar ;~)

Unfortunately, without any ability to duplicate it, I don't see how Nirvana can help.
^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/21/2011 12:55 PM
Post #23800 - In reply to #23799

I agree, Jim. If Nirvana wants to look at it, I will send it to them--the whole strat. If they don't want to look at it, coming from me, I understand.

This strat was just one of a "gazillion" other experiments I run hoping to bit-by-bit improve my overall strat building competence.

I am doing something different here that, up until now, I have worked hard to avoid: check-marking the individual parameters of an orders block. Because, as you have wisely stated a number of times, don't do something unless you understand it. My brain is painfully limited and therefore I cannot grasp the interrelationship between optimizing orders block parameters AND either legacy optimization or DO. In other words what is the interrelationship between when orders block parameters are optimized AND when DO is turned on or when it is turned off when running the Port Sim.

Don't waste your time trying to explain it all for the "gazillionth" time. Unless you want to try to explain again for others. It's just something I will never get, among many other things.

The point is I try to limit trying things I have no hope of ever fully grasping no matter how much I study it. So maybe my first ever use of an orders block with optimized parameters is doing something here I can't fathom.

You might laugh at what I'm using for my orders block. It was not intended to be a polished effort. I was just trying out a new idea. But it yielded a wild 93%(yes, that's right)HR in the FT report. I will attach it. I only selected one and only one stop: custom target. I then adjusted some of the default parameters and checked them. That's my entire exit "strategy" in this case.

BTW, I am not even using WC(DWC/PB/FB) at all in this case. Just current, daily EOD for every block. No DO or legacy optimization. No other optimization in any block.

It's beyond me.




[Edited by c on 4/21/2011 11:04 PM]

Attached file : 93% FT HR Report.gif (10KB - 199 downloads)
Attached file : One stop selected--Cust Target.gif (40KB - 179 downloads)

^ Top
Jon Grasty

Elite
5001002525
Posts: 660

Joined: 10/9/2006

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/21/2011 1:40 PM
Post #23801 - In reply to #23800


C,

I sent an email requesting the files needed to look into this for you. We will be more than happy to help in anyway we can.

Jon Grasty
Nirvana Support
^ Top
bvandyke

Member
25
Posts: 36

Joined: 4/12/2011
Location: Dallas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/21/2011 5:05 PM
Post #23802 - In reply to #23798

A bit off topic but a question from an new EOD user on the Port. Simulator.

I have my Trade Selection Tab set to Strategy and using just One strategy.

Looking at my trades report from the given strategy it appears that on a given day where multiple trades were entered...Symbols selected for entry that day were taken alphabetically from those Stocks available for entry on that day.

Sorry that's a bit wordy.....but i'm assuming that if on a given day my PS settings allow more than one entry the entries will be taken alphabetecially from Symbols that had entry signals that given day?





^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/21/2011 10:02 PM
Post #23803 - In reply to #23798

Yes, Jon, thank you. Of course I will send whatever files you require. Maybe a day or two.

I have a correction to make. I said that I did not use WC in this strat. I confused myself and did not double-check before I wrote that. I do indeed use WC in this strat. I use current+1(weekly) DWC. I am careful not to use the FB option. Only DWC. I don't think it matters in this case as to what is happening here, but I wanted to be certain I corrected myself for the record.

BTW, I have been tediously attempting to try out EVERY system I have for DWC. I was inspired by Jim's group project. On a much smaller scale, I wanted to attempt to see how each individual system, with DWC applied to them, affects a single strat. Not at all scientific or comprehensive. It would take a large group and many man-hours. However, I noticed that a particular system, the KBA-C system, of all systems, does, for whatever reason, not only work well, generally speaking, but strikes a particular "sweet spot" in terms of high HR's and a fairly large number of trades. I'm speaking of Breakout-type opportunities. That is the DWC system I'm using in this case.

I mention it because I cannot contribute like the experts can but when I can contribute, even a little bit, I want to do it. For those who are just starting out with DWC, I would suggest you at least try the KBA-C system for DWC if you are working with breakouts. I don't know why but it seems to work very well in my limited testing.

Bvandyke, I cannot answer your question. I hope the experts will help you.

[Edited by c on 4/21/2011 11:02 PM]

^ Top
Jim Dean

Idol
2000500100100100252525
Posts: 2884

Joined: 9/21/2006
Location: L'ville, GA

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/22/2011 5:25 AM
Post #23804 - In reply to #23803

Over in the NClub, a possible Weekly Confirmation bug was still found to be crawling around.

It appears to occur when the Confirm block is in a SECOND strat-string (I think Nirvana calls them "child" strings). A simple run was done with WC used in the Parent string, then a second run was done with WC block separate, in a child string, feeding into an empty Confirm block in the parent string. I did not see a full copy of the strat or profile settings - just the block diagram and the results - but the two runs should have had identical results.

The analysis results were VERY revealing. The second run with the WC in the child string showed 92% hit rate (50% ANP) but the single-string one showed around 66% HR (14% ANP).

QED - do NOT use Weekly Confirm on a Child String, until Nirvana gets it fixed.

For NC members, here is the link to Mel Dickover's post:
HTTP://www.nirvanasystems.com/nirvana/nirvanaclub/members/NClubForum/thread-view.asp?threadid=1065&posts=71#43993

[Edited by Jim Dean on 4/22/2011 5:45 AM]

^ Top
Jeff Drake

Veteran
100100252525
Posts: 287

Joined: 11/17/2003
Location: Austin, TX

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/22/2011 9:45 AM
Post #23805 - In reply to #23803

C,

Jon will give you more detail when he receives your files, but a quick glance at the first scree shots show nothing out of the ordinary. Some of those trades lasted a year or more. While they may have ended in a profit, you were holding positions through turbulent times - hence the equity curve.

If you go look at the trades on the charts of the individual stocks, I think you will see why your equity curve looks the way it does. Remember that hit rate and profit per trade only tell you the direct distance from point A to point B, but they tell you nothing about the actual path you took.
^ Top
Barry Cohen

Sage
50001001001001002525
Posts: 5467
Online

Joined: 1/19/2004

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/22/2011 9:56 AM
Post #23806 - In reply to #23802

Looking at my trades report from the given strategy it appears that on a given day where multiple trades were entered...Symbols selected for entry that day were taken alphabetically from those Stocks available for entry on that day.

Sorry that's a bit wordy.....but i'm assuming that if on a given day my PS settings allow more than one entry the entries will be taken alphabetecially from Symbols that had entry signals that given day?


Bill, trades in the report are selected based on this order:

1. Date(earliest)
2. Advisor Rating(highest)
3. Symbol(alphabetical)
4. Exchange(alphabetical)

^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/22/2011 12:35 PM
Post #23807 - In reply to #23798

Hi Jon,

Files were zipped and sent.

Jeff Drake's post was very interesting for what it did NOT say as well as what it did say.

It's quite extraordinary to me that Jeff says that, upon initial glance, there is nothing particularly out of place here. I mean that even though a number of trades were held a year or more, a 93% FT HR was not at all surprising to him. To be fair, he has not seen my strat yet.

So the fact that a 93% HR is even possible, whatever the circumstances, as long as Jim's caveat for the DWC "child string" issue is not in play here, is incredibly good news for all of us. It expands the range of what may be possible. After all, Nirvana thinks an 80% HR is possible for their UTM.

It does imply that in skilled hands(more skilled than mine), some incredible HR's are possible and should not be dismissed out-of-hand as wildly impossible.

So the fact that Jeff did not dismiss a 93% HR as being too good-to-be-true is a remarkably good non-commentary. Good stuff.

I want to re-emphasize an earlier point. Not even I would use a single stop, Custom Target, as my sole TP. I was merely testing out some filter combinations and was not paying particular attention to my exits. I was just trying out new ideas. But when I see a 93% HR spread out over nine years and more than 100 trades, I have to stop and see what is going on or get help to see what is going on.

As to Jeff's comments about the equity curve reflecting the PATH of the trade vs. the straight-line, A-B HR and PPT numbers, I think I should have known that but somehow it escaped me. What threw me off what that even though the sometimes year-long trades eventually ended up in a profit, I still find it hard to understand why, according to the View Trades report, the Max Draw is 45% when NOT A SINGLE LOSS was recorded from late 2007 to early 2010. I get the path-thing. I will eventually get it but I'm having a hard time absorbing how you get 100% profitable trades for over two years and yet have a nasty drawdown. But on some level I get it. I just have to digest it.

I look forward to the details of your analysis.

Thanks for the help.


[Edited by c on 4/22/2011 12:38 PM]

^ Top
Jeff Drake

Veteran
100100252525
Posts: 287

Joined: 11/17/2003
Location: Austin, TX

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/22/2011 3:36 PM
Post #23815 - In reply to #23807

C,

I've seen a number of systems/strategies on the market that post very high hit rates, but they are prone to substantial drawdowns. In fact, you could take just about any basic system and use a 1 ATR profit target and 4 ATR loss stop and see similar numbers. It's defintely possible. The question is how big of a potentail loss are you willing to risk in order to achieve that high hit rate.

I've attached a chart of the SPY to illustrate the point. It shows that we buy in mid 2008 and sell in early 2011 for a profit. In this case, we have a 100% hit rate and a 9% profit per trade. However, due to the recession that occured while in trade, we incurred a 44% drawdown. This is similar to what you saw on your strategy run.

Many times I've worked on a stratey and the initial numbers from a performance report were exciting, but when I got additional information from the equity curve, I saw extensive drawdowns. Your example is a good illustration of why Portfolio Simulation is essential for proper strategy evaluation.

I hope that helps. Jon will check out your files and let you know if there are any other issues. Enjoy your weekend and good luck with your research.
Attached file : SPY.png (25KB - 203 downloads)

^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/23/2011 2:32 PM
Post #23816 - In reply to #23798

Wow, Jeff, that explanation and chart you provided for SPY makes it all clear. I appreciate the response.

Something else I think that Jeffs' post indirectly points out is that my particular strat, as I have set it up, is very much a slave of the overall market. Had the overall market been declining instead of climbing for this past decade, my strat would have largely reflected the market and it would have incurred tremendous losses. The only thing that allows the strat to show an eventual profit is overall long-term market direction.

This strat or something like it is likely more suited to very long term buy and hold investors who would prefer to place their faith in the overall rise of the market measured in years, not minutes/days/weeks/months as many of us(me included) are inclined to do. Perhaps dividends investors who are less concerned with large drawdowns are interested in this approach. That's fine.

The work and art comes in when we want to be in and out of trades in minutes/hours/days/weeks.

So this exercise has been helpful for helping me put into perspective the relationships of high HR's vs. length of trade and large drawdowns. It would seem like this would be intuitively obvious but unless mentors like Jeff and Jim help us, even the obvious often escapes us. Or at least me. Thank you.

I want to move this very helpful thread along and also circle back to a point that Jim was helping all of us with concerning the new potential bug in DWC.

So we can all get on the same page, I'm going to do two things:

1. I'm going to show how to set up the long-term strat we have been discussing. It's very simple. Nothing brilliant about it. But that will answer Jim's question about DWC and my possible use of child strings, I think.

2. In addition, I'm going to show a PS run and report where I am using a strat much different than the one we have been discussing. But there are two common elements between them. 1. A very high HR(85% FT). 2. I'm using the same KBA-C system for DWC as I was in the long-term strat. This new strat is also a call-option strat whereas the long-term strat was just stocks. Much shorter term, too. Average trade is 26 days. The new results are very interesting. Teaser: With this new strat, we're not quite back to the "gazillion" Twilight Zone but the PS run kind of gives hints of that. If the DWC child string bug Jim describes is in play here, then my strat is an illusion. But if not, I think the strat is on solid ground.

Out of time for now. I will continue this thread later.
^ Top
julesrulesny

Elite
500100100100100
Posts: 900

Joined: 8/15/2007
Location: NYC

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/24/2011 1:03 PM
Post #23823 - In reply to #23815

Jeff,

Thanks for your chart on SPY and the description. But, I got confused on something.

You are showing a 100% HR where only 1 signal was fired tho it was eventually profitable.. If I remember, HR means profitable signals.. If I had tested a strategy over the course of X amount of days and fired 10 signals, of which 4 were profitable over X amount of days and, 6 were losses over X amount of days.. Does this mean my HR was 40%?
^ Top
Barry Cohen

Sage
50001001001001002525
Posts: 5467
Online

Joined: 1/19/2004

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/24/2011 4:30 PM
Post #23825 - In reply to #23823

If I remember, HR means profitable signals.. If I had tested a strategy over the course of X amount of days and fired 10 signals, of which 4 were profitable over X amount of days and, 6 were losses over X amount of days.. Does this mean my HR was 40%?


Bingo!
^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/25/2011 7:15 PM
Post #23840 - In reply to #23798

I want to show a PS run on a strat that uses the KBA-C system, default parameters, as DWC. I also want to circle back to Jim's comment about a possible DWC bug.

There is one large oddity about the results.

First, the FL is NASDAQ 100.

The ROI over 215 trades and nine years seems almost gazillion-like: 2,500%.

Maybe this is what the experts get or probably much better. I have no idea how to tell if any of this is legit.

But the oddity is that the HR is quite low: 46% in the report and about 50% in the PS run.

Jim said that the bug drove HR's above 90%. So my low HR might suggest that DWC is not being affected by the bug. But I don't know.

But it's the combination of relatively low HR and high ROI that I don't understand. I have run many PS tests over the years and have never seen this combination. I have always assumed a high correlation between HR and ROI. But this strat seems to defy that. I have no idea what to make of it. Maybe it's within the realm of the possible. But who knows.

My TP could be the answer. It is an optionized version of the Hurst TP with my own little additions. But even that doesn't seem to explain what I'm seeing.

If Nirvana would like to take a look at this, I would like to send it to them.

Attachment 1: PS run.
Attachment 2: Report.
Attachment 3: View Trades.
Attachment 4: TP.












[Edited by c on 4/25/2011 7:35 PM]

Attached file : KBA-C DWC PS.gif (108KB - 186 downloads)
Attached file : KBA-C DWC Report.gif (5KB - 163 downloads)
Attached file : KBA-C DWC trades.gif (38KB - 172 downloads)
Attached file : KBA-C DWC TP.gif (34KB - 181 downloads)

^ Top
Jim Dean

Idol
2000500100100100252525
Posts: 2884

Joined: 9/21/2006
Location: L'ville, GA

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/25/2011 7:34 PM
Post #23841 - In reply to #23840

Maybe I missed it.

Did you say whether or not you have a second thread that uses dwc in a systems block?

In another thread Barry said that dwc was not fixed in the systems block.

This contradicts what many of us heard earlier. But as yet, followup questions have not been answered.

If you are not using dwc in a systems block or in a child thread then you're probably ok

I'm less sure ofthat now than I used to be. Hopefully some more clarifications will be provided.
^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/25/2011 7:47 PM
Post #23842 - In reply to #23798

I only use DWC in either a Confirm block or Filter block.

I only use one primary line of blocks. No secondary lines at all.

If by using DWC in a systems block, you mean that we select Current+1(weekly) and move that block to child/secondary line, as a confirmer for the primary line of blocks, then no. I don't do it that way.

Here is my setup on one primary line of blocks:

1. Systems(current, meaning daily).
2. Vote.
3. Confirm(Current, daily).
4. Confirm(Current+1, KBA-C, DWC checked).
5. TP.

I think I know what what you are referring to as far as using WC in a systems block. I did do that a while back and I think I said so in an earlier thread. But I don't think I said so in this thread. If I implied that, I apologize. The above list is how I'm setting things up.

Thanks.





[Edited by c on 4/25/2011 7:51 PM]

^ Top
Jim Dean

Idol
2000500100100100252525
Posts: 2884

Joined: 9/21/2006
Location: L'ville, GA

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/25/2011 8:07 PM
Post #23843 - In reply to #23842

Hmm

Two confirm blocks in a row is out of the ordinary.

Try putting the daily confirm logic into a second string Systems block that feeds the main string confirm block, maybe.
^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/26/2011 1:22 PM
Post #23849 - In reply to #23798

Interesting that a separate daily confirm block and DWC confirm block on the same line is unusual. That's fine. It's not obvious to me why it could be a potential problem.

A strat that comes to mind that "requires" a daily confirm block(if trading in the daily timeframe) is the classic Darvas breakout.

1. Systems block--DAN-B.
2. Vote block.
3. Confirm block--Gap and/or Vty system.
4. Confirm block--DWC checked.
5. TP.

All on one primary line of blocks.

If this is unusual, that's different from saying there is some kind of internal computing issue that prevents this setup from working.

I'll try your setup and see what happens.

Thanks.
^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/26/2011 1:48 PM
Post #23851 - In reply to #23798

Let me correct myself. You said two confirm blocks IN A ROW. You did not say on the same line. My misread. So I could add a filter block in between the two confirm blocks or try your other suggestion. Got it.

Something else. I promised I would lay out the strat that started this thread. The one with the 45% draw yet had zero losses for two years or so.

Nothing special, really. Very simple.

1. Systems block: Hammer(came from the OT pro forum. I did not write this.)

#System

If C > L+ 2*((H-L)/3) and O > L+2*((H-L)/3) and _
STO(5,2,3) < 50 then
Signal=LongSignal

ElseIf C < H-2*((H-L)/3) and O < H-2*((H-L)/3) and _
STO(5,2,3) > 50 then
Signal=ShortSignal

End If

2. Vote. LB--1, any signal, fixed--0.

3. Filter. C >= MZ_Lower2(1,1,20,1,0) AND C <= VL2(1,1,20,1,0).

4. Confirm. KBA-C, DWC, Current +1. Before--4(I probably should use 3), After-- 0.

5. Orders block. Custom Target. Optimized. Some default parameters adjusted. I will post a pic of it for clarity.

Give me a few minutes and I will also attach the appropriate file.

Mr. Jon Grasty called me personally this morning to give me a detailed explanation for what was going on with the above strat. His explanation further added to Jeff Drake's post.

I must say that Mr. Grasty could not have been more helpful. I expect this kind of service only for club members. I am not. So for him to take time out of his busy day and call me and give me this kind of personal help is quite beyond anything I would have ever expected.

He said he worked on my strat for an ENTIRE day! I told him I would never have intruded on his time like that. I just thought Nirvana would take a quick five minute look and their expert eyes would quickly find out what was going on. And that alone would have been tremendous service. But this went far beyond any expectations.

So thank you to Nirvana and particularly Mr. Jon Grasty. Club Member-type service for a non-club member. Amazing. Thank you, sir.



[Edited by c on 4/26/2011 1:54 PM]

^ Top
Jim Dean

Idol
2000500100100100252525
Posts: 2884

Joined: 9/21/2006
Location: L'ville, GA

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/26/2011 1:50 PM
Post #23852 - In reply to #23849

Hey Charles ... I mis-spoke. It's just "unusual" to *me*. And, I did mean to say "on the same line" ... I doubt that putting a filter block between them would matter.

Everyone gets into "patterns" about how they do stuff. It's just not a pattern that I've used before ... I gravitate towards using child strings. Many ways to skin the cat. In theory, they should all work the same. ;~}

And, you got it in one. Jon is a champ!

[Edited by Jim Dean on 4/26/2011 2:11 PM]

^ Top
c

Legend
10010010025
Posts: 335

Joined: 6/28/2007
Location: texas

User Profile
 
Subject : RE: Zero losses for 2008 yields 45% drawdown
Posted : 4/26/2011 2:07 PM
Post #23854 - In reply to #23798

No question Jon is great guy. Very impressive service.

I'm attaching the promised OTS file.
Attached file : Hammer Cust Tgt DWC.ots (828KB - 166 downloads)
Attached file : HammerCandleOTForum.txt (0KB - 169 downloads)

Jump to page : 1
Now viewing page 1 [25 messages per page]
back reply

Legend    Action      Notification  
Administrator
Forum Moderator
Registered User
Unregistered User
E-Mail this thread to a friend
Toggle e-mail notification


Nirvana Systems
For any problems or issues please contact our Webmaster at webmaster@nirvsys.com.