OmniVest User Forums
General Discussion Topics
Evaluating Trading Strategies
15 replies, 1976 viewings
|
OmniVest User Forums
General Discussion Topics
Evaluating Trading Strategies
|
Last Activity 4/15/2019 7:43 PM 15 replies, 1976 viewings |
|
|
|
| Printer friendly version |
Jeff Mishlove Member  Posts: 31 Joined: 10/11/2012 Location: Las Vegas, NV  User Profile |
The spreadsheet shows the strategies organized alphabetically. In addition to the raw data from OT, I have added some columns of my own making. Columns B, C, and D, indicate the strategies selected by Mark Holstius for the three different market conditions. Columns E, F, and G show the strategies selected by Nirvana for the Universal Method. Columns T, U, and V are based on three formulas from the forward testing data that I have chosen to help rank the strategies. This should be clear from the column descriptors. Column W is highlighted in red for those strategies that also had unprofitable short trades. It is highlighted in green when the short trades were profitable. And, for all of the strategies with short trades, Column X is highlighted in blue indicating the statistics for the long-only trades of those strategies. I used Columns P, T, U, and V for ranking the strategies. Using the Excel sort function, I was able to highlight in yellow the top strategies for each of these columns. For each of the 27 strategies that were highlighted in yellow in three of the four ranking columns, I then highlighted the name of that strategy in yellow. I see that the strategies selected by Mark Holstius and by Nirvana do not overlap entirely with the 27 strategies selected by my rankings. So, there remains much to consider. I still wish to go through all of these strategies to make some sort of preliminary judgments as to which will work best under various market conditions. Jeff  Attached file : Summary Data Spreadsheet.xls (84KB - 544 downloads) | ||
| Mark Holstius Elite   Posts: 744 Joined: 10/11/2012 Location: Sleepy Hollow, IL User Profile |
Thanks for taking the time to do that, and especially for sharing it with all of us. I just downloaded your spreadsheet and plan to spend time with it over the weekend. I'm sure ATM will help us improve a great number of things this coming year. Mark | ||
| Mark Holstius Elite Posts: 744 Joined: 10/11/2012 Location: Sleepy Hollow, IL User Profile |
Would you mind sharing the date ranges you used for the BT and FT? Mark | ||
| Jim Dean Elite  Posts: 1059 Joined: 10/11/2012 Location: L'ville, GA User Profile |
Also, please share the allocation rule and the focus list info. | ||
| Jeff Mishlove Member Posts: 31 Joined: 10/11/2012 Location: Las Vegas, NV User Profile |
To my knowledge, this test did not use any of the ATM allocations. The report was generated, according to Ed's instructions, by running the to do list and then viewing the performance summary from the File > Print Reports menu. So, it was simply an opportunity to view the strategy performance over a lot of data. I ran the process eight different times, with about 20 different strategies checked off each time (because running all of the strategies at once caused overflow problems with my system). Jeff | ||
| gbarber Veteran Posts: 282 Joined: 12/30/2012 Location: Pearland, TX User Profile |
What ATM method did you use? | ||
| Jeff Mishlove Member Posts: 31 Joined: 10/11/2012 Location: Las Vegas, NV User Profile |
Jeff | ||
| gbarber Veteran Posts: 282 Joined: 12/30/2012 Location: Pearland, TX User Profile |
| ||
| Tom Helget Member Posts: 42 Joined: 10/11/2012 Location: Baldwinsville, NY User Profile |
SUPER JOB! Most appreciated. Tom Helget | ||
| Tim Lambie Member Posts: 6 Joined: 10/11/2012 User Profile |
This is awesome, thank you. I tried this as well before seeing your note and with a pretty beefy setup (5820k, 16GB RAM and SSD drive) my system seemed to give up around 10 strategies. I am curious to know if you ran the whole list without issues or, if you parsed/batched them in some way to get it done and roughly how long did the run take? My other tests have run fine until this elephant. Cheers, Tim | ||
| Jeff Mishlove Member Posts: 31 Joined: 10/11/2012 Location: Las Vegas, NV User Profile |
I actually tried to run all the strategies at one time -- and that overloaded my system. After that, I broke it down into 8 batches of about 20 strategies each. The whole process seemed to take about half a day. Jeff | ||
| SteveJ Veteran Posts: 105 Joined: 10/11/2012 Location: UK User Profile |
I thought I'd take advantage of Jeff's work (thanks Jeff) and run a number of his top strategies through the selection process for a bear market state. I included in the run strategies unlikely to be useful in the final result (ie long trending strats in a bear market state) - I wanted to see what happened! After about 6 hours the run ended but, unexpectedly, all the strategies I selected for the run are in the list of strategies returned. I'm probably doing something wrong or have misunderstood. Can someone help by explaining to me how to use the simulation statistics to establish which are the best / worst strategies form the run? Steve | ||
| Tom Helget Member Posts: 42 Joined: 10/11/2012 Location: Baldwinsville, NY User Profile |
You need to check the Optimize check box for the Bear Market state only after also checking off above in their respective boxes the check boxes for the Strategies you want to consider for the Bear Market State. Be careful here because after you have Run the Analysis only the Strategies that are the winners will subsequently be checked. You can also verify this in the Simulation Statistics column at the right of the Portfolio Simulation chart by scrolling down to the Bear Market state data. Indeed, sometimes all the Strategies you selected will be "returned" in the optimization. Note that if you now decide to evaluate, say, the Raging Bull Market state, you should un-check the Optimize check box for the Bear Market state and check the Optimize check box for the Raging Bull Market state so that in your next run only the Raging Bull Market situation is evaluated to save time (and, yes, I made that little error myself a couple of times - more or less re-doing what had already been done). I hope that helps! Tom Helget [Edited by Tom Helget on 2/26/2018 4:05 PM] | ||
| SteveJ Veteran Posts: 105 Joined: 10/11/2012 Location: UK User Profile |
Steve | ||
| Tim Lambie Member Posts: 6 Joined: 10/11/2012 User Profile |
Thanks for the details. That correlates with my experience. It appears we need some automation for the automation ;-) as big strategy and optimization runs seem to gum up the works. I don't believe this is purely about processor speed and available memory. We should all try to post where we find the capacity limits in this area to inform each other and Nirvana, avoiding some time wasting replication and perhaps getting Nirvana to address the scale/architecture issues we find in testing. Thanks, Tim | ||
| Tom Helget Member Posts: 42 Joined: 10/11/2012 Location: Baldwinsville, NY User Profile |
I thoroughly agree with you. One of my pet peeves here is that once the Run Analysis button is "pushed" you really don't know what you are letting yourself in for. Depending on what I am optimizing the run may only take an hour or two but it could easily go overnight or even several days. It certainly would be nice if some time "estimate" of the run could be given and even "paused" (like with Strategy Wizard). Wishin' and a Hopin' Tom Helget |
|
|
|
|
Administrator
Forum Moderator |
Registered User
Unregistered User |
 E-Mail this thread to a friend(un)/Freeze thread | |
Toggle e-mail notification |
