OmniVest User Forums
OmniVest Success
What A Difference A Day Makes
25 replies, 2946 viewings
|
OmniVest User Forums
OmniVest Success
What A Difference A Day Makes
|
Last Activity 7/22/2015 1:06 PM 25 replies, 2946 viewings |
|
|
| Printer friendly version |
tgrafa Regular  Posts: 63 Joined: 10/11/2012 Location: Midland, Texas  User Profile |
Trey  Attached file : ETF Backtest.png (58KB - 286 downloads) Attached file : Forward_Test_'0'.png (82KB - 275 downloads) Attached file : Forward_Test_'1' (1).png (93KB - 283 downloads) Attached file : ETF Summary.png (48KB - 278 downloads) | ||
| Jim Dean Elite  Posts: 1059 Joined: 10/11/2012 Location: L'ville, GA User Profile |
| ||
| Steve2 Elite   Posts: 750 Joined: 10/11/2012 Location: Annapolis, MD User Profile |
Losing a day every time you close a position makes a big difference. Notice that your TPM dropped from 4.7 to 3.1. That's 35% fewer trades per month. With each trade using 50% of your account value your % invested really drops (although that is not evident from OV stats given the way that OV computes % invested). I'd be curious to know what happens if your backtest is done with settlement wait days set to 1. With the small number of trades you are executing per month, I think you really need to do a bunch of sims with randomly selected starting dates to assess volatility of the performance results. The one flaw in Mark's approach may be that with very few large trades per month, results may vary widely (I'm hoping that's not the case because you can put together some very interesting portfolios with his approach). Lastly, there is no need to set settlement wait days to 1 when trading an IB IRA Margin account. Set settlement wait days to 0 then in TP, enable the "Delay Opening Orders" option and set the time to say 9:35 am EDT. This will cause TP to submit closing orders before market open (assuming you have your trading window configured to do that) and opening orders 5 minutes after market open. This will allow closing orders to complete execution before opening orders are submitted and you shouldn't get any rejected orders. I've been trading an IRA Margin account at IB this way for almost 6 months with no rejected orders. Your slippage on the opening orders will be more volatile but I think it all evens out in the end. Steve [Edited by Steve2 on 6/21/2015 9:46 PM] | ||
| Steve2 Elite Posts: 750 Joined: 10/11/2012 Location: Annapolis, MD User Profile |
So my guess is what you are seeing is due to volatility in performance introduced when one starts to vary account settings. So, I think we all need to do a Monte Carlo analysis on our completed ETF portfolios to determine the most likely expected performance. Steve | ||
| tgrafa Regular Posts: 63 Joined: 10/11/2012 Location: Midland, Texas User Profile |
| ||
| tgrafa Regular Posts: 63 Joined: 10/11/2012 Location: Midland, Texas User Profile |
I have attached another graphic that answers your question regarding the back-test settlement days set to 1. Interesting that your BT and FT was actually better with settlement set to 1. I'm guessing that goes back to the small amount of trades and the possible volatile returns one might expect. I am going to take you up on your suggestion on checking returns for a wide range of starting dates - I'll let you know once I get a good number of runs. Attached file : Basktest 1 vs 0.png (103KB - 298 downloads) | ||
| Steve2 Elite Posts: 750 Joined: 10/11/2012 Location: Annapolis, MD User Profile |
What value for Buying Power are you using? Do you have the "Allow settings to reduce trade size" option enabled in account settings? If not, try enabling that and also enable "Minimum Trade Size" and set it to 5%. That might make things a little less volatile. Steve | ||
| tgrafa Regular Posts: 63 Joined: 10/11/2012 Location: Midland, Texas User Profile |
I am beginning to think there is an intermittent bug with OmniVest on how the program is allocating funds with the settings I am using: BP 100%; Allocation Max 50%; '0' settlement; Max strats per symbol '1'; Min Trade Size 6%. Later this evening on the forum, I will post some screen shots along with why I believe there is a bug. Hopefully I have found the reason that when things go bad they go very very bad!! Trey | ||
| Mark Holstius Elite Posts: 744 Joined: 10/11/2012 Location: Sleepy Hollow, IL User Profile |
What settings are you using in the Portfolios Max % and Allocations - just to make sure we're all "on the same page". I set them way up so that the Account Settings will control things (lower them as requested): Mark [Edited by Mark Holstius on 6/22/2015 7:29 PM] Attached file : Settings.jpg (37KB - 638 downloads) | ||
| tgrafa Regular Posts: 63 Joined: 10/11/2012 Location: Midland, Texas User Profile |
The first chart shows the runs from 1/1/2014 thru 2/25/2014. Note that on the 2/11/2014 run CAR drops to -5.2 and TPM drop to 1.2 - quite a change from the previous 37 runs. The 'bad' runs continue on the spreadsheet thru 2/21/2014 because not until 2/22/2014 did the 'bad' trades fall outside of the run date and therefore out of the history. The second chart attached shows the historical trades for this 2/11/2014 run. The first trade was for R4-L-ETFEEV which opened on 2/11/2014 and closed on 2/20/2014 - all is normal. Having 100% equity available on the morning of the 21st, trades fired for R11-L-ETFFXP and PB2-L-ETFGDX for 50% each. Note that the trade for FXP closed out on 2/25/2014 while the PB2 trade for GDX ties up 50% equity for 7 months - until 9/26/2014 (a discussion for another time but maybe PB2 and other long term trades are not be a good idea for a 50/50 ETF portfolio with out some type of conditions - 7 months with a 16.89% loss would not be fun in the real world). What caught my eye in the trades chart was that even though the FXP trade closed on 2/25/2014, freeing up 50% equity, that 50% is never reinvested through the end of the run with only one position at a time being held for the rest of the run through 6/18/2014 - thus the TPM figure of 1.2. The 50% seems to have been locked up by OmniVest and not available for investing purposes. On chart three you get more evidence of what has happened. Even though FXP sold on 2/25/2014 and with only 1 position at a time held for the balance of the year, the performance chart shows 100% invested most of the time with 50% minimum invested all the time - the 50% is somehow allocated to a non existent position. This is why the avg invested jumps to 91.5% for the run - far above the 75%-78% on the first 37 runs. I'm feel sure that this probably goes back to the original discussion about the '0' vs. '1' settlement day. The settlement day wasn't the problem - it just by happenstance threw me into the same likely error. I haven't had time to look into those trades and allocations but I suspect something similar since in that case the equity curve seems to drop and never recover through the whole run. Look over this and if I am misinterpreting something let me know. Otherwise this probably needs to be reported to Barry. Trey Attached file : ETF Start Dates.png (71KB - 281 downloads) Attached file : Trades 2_11_2014.png (167KB - 252 downloads) Attached file : ETF Run 2_11_2014.png (75KB - 259 downloads) | ||
| Steve Mayo Legend Posts: 414 Joined: 10/11/2012 Location: Austin, TX User Profile |
2) CAR is calculated from enddate and startdate equity, ignoring everything in-between. If you have a "funky" trade on either of those dates, or if they are just in the midst of drawdowns or draw-ups, it will throw the CAR way off. You really need to be using a better metric, say average daily return, to get a valid "experiment", but of course that means exporting all that data from OV. Alternatively, run one full-period export through the Monte Carlo tool on my site -- that will give you a better idea of how consistent (and thus how predictable) the return might be. | ||
| tgrafa Regular Posts: 63 Joined: 10/11/2012 Location: Midland, Texas User Profile |
Please let me know what you think. It is the 50% that is tied up that is my real concern. Thanks - Trey | ||
| tgrafa Regular Posts: 63 Joined: 10/11/2012 Location: Midland, Texas User Profile |
| ||
| Steve Mayo Legend Posts: 414 Joined: 10/11/2012 Location: Austin, TX User Profile |
Please post the spreadsheet and the equity curve (or export) you mentioned so we can see the "error" you are talking about. | ||
| Steve2 Elite Posts: 750 Joined: 10/11/2012 Location: Annapolis, MD User Profile |
Can you please post a snag of the account settings you are using. There is a bug in OV trade filtering that I noticed when testing my ETF account. I reported this to Barry last week but I don't think he's had a chance to look at it yet. I was testing with Strategy alloc set to 10%, portfolio alloc set to 500% and Max % Invested set to 50%. These settings cause OV to generate trades that use 49.xx% of account equity. I had no account settings enabled. With Buying Power set to 100%, OV generated 135 trades during the simulation. I then set Buying Power to 95% and OV only generated 3 trades for the same simulation date range. Given my configuration and the results for 100% buying power, I would have expected OV to generate at least 135/2 = 67 trades. When I set Buying Power to 90% only one trade was generated and when I set Buying Power to 85% no trades were generated. In both these cases, I would have expected at least 67 trades to be generated. I then enabled "Allow settings to reduce trade size" and got the correct behavior at each of my Buying Power Settings. So, I think there is something wrong with how OV is filtering trades. Not sure if this problem has always been there but just wasn't noticeable when trade size is small compared to Buying Power or whether it's somehow caused by the large allocations we're now doing. Hopefully, Nirvana will get to the bottom of this quickly. Do you have "Allow settings to reduce trade size" enabled? Steve | ||
| Steve Mayo Legend Posts: 414 Joined: 10/11/2012 Location: Austin, TX User Profile |
| ||
| tgrafa Regular Posts: 63 Joined: 10/11/2012 Location: Midland, Texas User Profile |
| ||
| tgrafa Regular Posts: 63 Joined: 10/11/2012 Location: Midland, Texas User Profile |
I do have the allow trade size to be reduced box checked. After hearing you are having similar issues I'm fairly confident that this is a bug that needs to be addressed by Nirvana. I don't know the protocol for reporting the error - do I need to email or call Barry? Do you know if they let you know when they get it corrected? Thanks - Trey Attached file : IRA ETF Trade Settings.png (93KB - 279 downloads) | ||
| Steve2 Elite Posts: 750 Joined: 10/11/2012 Location: Annapolis, MD User Profile |
If you want to report this, I would suggest posting to the Technical Support Forum or email Barry. Steve M, You are correct that Buying Power should not scale trade size, it just leverages how much money there is to trade with (i.e., should cause OV to be able to make a smaller or larger number of concurrent trades). The bug is that the OV code DOES appear to be using Buying Power to determine acceptable trade sizes rather than using the account value. Here is what I have observed. Assume that your allocations are set so trade size is about 50% of account value. You can't do this exactly due to share rounding but most generated trades come close. OV will always round shares down so sometimes the allocation is slightly smaller than 50%. Assume that your account equity is $100K and "Allow trade settings to reduce trade size" is OFF. If Buying Power is set at 100% then trade filtering appears to work correctly and up to two concurrent trades are taken at a time. If Buying Power is set at 150% then trade filtering appears to work correctly and up to three concurrent trades are taken at a time. But if Buying Power is set to say 95% then OV incorrectly filters out any trade whose allocation is larger than 47.5% (50% of the Buying Power when it should be using 50% of the account value). The correct behavior with a trade size of 50% and Buying Power set to 95% is that OV should take only one trade (at 50% allocation) at a time. The net result of this bug, when using 50% trade allocations is that you see a dramatic reduction in the number of trades taken if Buying Power is set below 100% with typically all trades filtered out when Buying Power is set below 95%. Steve | ||
| Steve2 Elite Posts: 750 Joined: 10/11/2012 Location: Annapolis, MD User Profile |
With a 50% trade allocation and 95% Buying Power, I had assumed that up to two trades would be taken at a time with the first trade at 50% allocation and the second trade at 45% allocation. But that's not how it works, instead every trade is reduced to 50% of Buying Power or 47.5%. Steve | ||
| Mark Holstius Elite Posts: 744 Joined: 10/11/2012 Location: Sleepy Hollow, IL User Profile |
Interesting thread and points, and I'll try to clarify a few things I've found in testing and development. Maybe this table will help with what happens to the trade sizes and numbers in OV with that 10,000% hack in the Portfolios settings I mentioned above and different Buying Power and Max %/ Symbol Settings: Mark [Edited by Mark Holstius on 6/23/2015 10:52 AM] Attached file : Trade sizes.jpg (142KB - 536 downloads) | ||
| Steve2 Elite Posts: 750 Joined: 10/11/2012 Location: Annapolis, MD User Profile |
Try it with Buying Power set to 95%. [Update: It's probably not worth the effort to do that. I'd just wait for N to take a look at the filtering issue when Buying Power is set below 100%. It's important because we can't actually trade IRA's at 100% buying power without risking rejected trades.] Steve [Edited by Steve2 on 6/23/2015 11:19 AM] | ||
| Mark Holstius Elite Posts: 744 Joined: 10/11/2012 Location: Sleepy Hollow, IL User Profile |
I really don't understand why OV isn't entering another trade during that long period. I do notice that the FXP Strategy on 2/21/14 has a (2) appended, but I suspect that's just because it was a duplicate during your Custom Strategy building process. That said, so far every time I've run into "strange things" with these new ways of using OV I've dug around and found myself (or my lack of understanding of how OV 'actually' works) at fault. First point - I've only used RTMs and VBX1 (with the exception of 2 VERY short term TLB systems and those only on 2 ETFs). I didn't make that point at the Bash, but your trades are a good illustration of why I avoid long term systems. There are a couple ways to "troubleshoot" trade problems. What follows is a way to try to get OV to hopefully expose ALL the trades it will take to see which ones are causing the problem: Try using: 1) Buying Power 100% 2) "Allow Trade Settings" checked 3) Max Exp %/Sym to only 5% 4) Min Trade Size UNCHECKED Then run it at the Account level and check "Display All" for the Historical Positions. At that point I usually copy all the trades and then paste them into excel to make it easier to sort them by date entered, size, etc. Give it a whirl and see if we can find the problem... maybe that PB2 "steals equity" somehow. Maybe edit your Portfolios to contain only RTMs...??? Hint: If you want to do that, just click on the Portfolio and when it comes up with it and the list of the Strategies in it, then you can uncheck all the non-RTMs and save the Portfolio with a NEW Name (that keeps the old one intact) Mark | ||
| Mark Holstius Elite Posts: 744 Joined: 10/11/2012 Location: Sleepy Hollow, IL User Profile |
Here's the synopsis of what I just found in the OV trades by doing what you said when bringing it down to 90% (to make the math easier). It appears to takes the first trades at the requested percent and then adjust the last one: Hmmmm Maybe that knowledge can be utilized: "It's not a bug, it's a feature..." Good if you want to do 40%, 40%, and 10% positions. Maybe we should move this to it's own thread - I have to leave for a while so don't have time right now, Mark [Edited by Mark Holstius on 6/23/2015 11:39 AM] Attached file : Buying Power.jpg (187KB - 544 downloads) | ||
| Steve2 Elite Posts: 750 Joined: 10/11/2012 Location: Annapolis, MD User Profile |
General Discussion Topics > Trade Allocation & Trade Filtering to continue the discussion. |
|
|
|
|
Administrator
Forum Moderator |
Registered User
Unregistered User |
 E-Mail this thread to a friend(un)/Freeze thread | |
Toggle e-mail notification |
