| Index | Recent Threads | Who's Online | User List | Search |
|
|
 |
Portfolio123 » List all forums » Forum: Simulations and Portfolios » Thread: Out-of-sample -v- in-sample |
|
Total posts in this thread: 2 |
[Request a Feature]
[Post new Thread] [Add To My Favorites] [Watch this Thread] |
| Author |
|
|
sburrill
Member
|
Marco, Great tool! Can you provide or point me to the information on the simulation or backtests within Portfolio123. Specifically, are the backtests run out-of-sample/in-sample, is there survivorship bias in the data, what about corporate actions, etc. I want to understand completely what I am backtesting to ensure the stability and robustness of my model(s). Thanks in advance. Scott  |
||
|
|
marco
|
The models are typically created using sims of different periods, like start date and length, then the # positions is varied as well as other variables. If a model holds up against all this variations it gets promoted to a P123 Model and is then managed daily with no further changes. Most of the models were created about 1 year ago with 2+ years of data, so they've been "live" for about a year. Others like the ValueRank ones, were created recently using 3+ years of data. The data we have is actually quite good because we collected it every weekend for the past three years, so it has no look-ahead bias and has inactive companies. It does have problems though because we don't have corporate actions, and sometimes a merger or spinoff is missed. We are working very hard to integrate a new and improved data feed to correct all the work-arounds and still mantain a clean, historical database suitable for baktesting. ---------------------------------------- Portfolio123 Staff |
||
|
| [Show Thread Printable Version] [Post new Thread] |
