| Index | Recent Threads | Who's Online | User List | Search |
|
|
 |
Portfolio123 » List all forums » Forum: General Comments » Thread: Survivorship Bias |
|
Total posts in this thread: 9 |
[Request a Feature]
[Post new Thread] [Add To My Favorites] [Watch this Thread] |
| Author |
|
|
lynneeb
Member
|
Thank you very much for such a terrific web site. I'm really looking forward to improving my trades. But I do have a question about the database and lack of survivorship bias. If the universe of stocks is historically correct, then I expected to be able to find Enron's stock during 2001. At that time it traded on the NYSE with the symbol ENE; its market cap was about $26 billion. I created a screen with these parameters: UNIVERSE(NYSE) MktCap > 24000 The Enron stock did not appear when I ran a historical search on 9/1/2001. Could there really be survivorship bias in the database after all? |
||
|
|
marco
|
Great question. Our universes at the moment only have the current constituents and therefore suffer from historical bias. We do have the data to build them historically correct (except the BuyAndHold and FolioFN universes). We'll make them historically correct soon. ---------------------------------------- Portfolio123 Staff |
||
|
|
gviersen
Advanced Member
|
Just to clarify for myself and lynneeb. It is just the "Universe" populations that have survivorship bias, correct? i.e., Enron would be included in any port or sim if the Universe criteria were not used (and, say, a criteria of Mktcap > 25000 were used instead)? I just want to make sure that is true and that lynneeb does not get the impression that the whole site contains survivorship bias, if he/she is a new user. |
||
|
|
marco
|
That is correct. If you run a historical screen Enron will be there. Only the universes have survivorship bias, and that is why they are not used in any of our models. We're going to correct this soon to backtest universe specific models, for example models that only pick from the SP500. ---------------------------------------- Portfolio123 Staff ---------------------------------------- [Edit 1 times, last edit by marco at Nov 15, 2004 5:39:19 PM] |
||
|
|
lynneeb
Member
|
Thanks very much for the clarification! When I remove the test for Universe(NYSE), Enron does show up. as well as several other companies, such as Chevron and Texaco. |
||
|
|
splanky
Member
|
Lynn: O.T. I'm new and trying to learn P123. I don't know how to contact you directly on this question so will have to use "General Comments" and hope it reaches you. It's with regard to a simulation I did of "Sim Copy1 of 2Sim Copy 5 of Penny Stock..". I must be doing something wrong because for the period 03/31/2003 to 01/08/2005 I get the amazing Annualized retuirn of 1,161.57%! What am I doing wrong -- or right?  |
||
|
|
lynneeb
Member
|
Hi, splanky! Wow! Those are incredible results from your simulation - and worth further investigation. I took a look at your simulation "Sim Copy of Sim Copy1 of 2Sim Copy 5 of Penny Stoc", which I assume is similar to the one you are asking about. Here's what I found: First, I looked at the Transactions section for this simulation and selected the "Realized P/L" report. This lists stocks by return from best to worst. I'm looking to see if the high performance was due to just a few stocks or was more broadly based. In this case, the outsize performance was due to one particular stock, DDIC. The simulation shows a Buy of DDIC on Dec 08, 2003 at $0.05 per share. The stock was sold on Jan 05, 2004 at $16.75. That's a gain of over 30,000%. But was this possible? A look at the historical prices and SEC filings on the Yahoo! Finance web site show that DDIC had filed for bankruptcy in early December. By Dec 18 the share price was 11.50. But where these the same shares? I don't think so. In other words, if you were actually able to buy the shares at $0.05 (just before bankruptcy) it isn't likely your share was worth $11.50 by Dec 18. At best I would expect you got a reverse split (multiple old shares for one new share). Does anyone have any other information? Next, I reran the simulation with DDIC on the resticted Buy list. This resulted in a similar situation with LEAP during August and September of last year. This company had also gone into bankruptcy and reemerged with shares valued at a higher price. Last, I reran the simulation with both DDIC and LEAP on the restricted Buy list and got more believable returns in the Realized P/L report. The end results: Total Return: 158.75% Active Return: 118.90% Annualized Return: 70.89% Annual Turnover: 985.86% Max Draw Down: -55.89% But, there are many transactions here that show stocks being bought and sold at $0.00! The most obvious cure is to put a minimum price in the buy rules. At $1.00 minimum price, the annualized return comes out to a still-lovely 55.78%, and the max draw down eases to -24.88%. In sum, your simulation for trading these penny stocks has brought out some interesting survivorship issues: companies that survive bankruptcy but appear to have done a reverse split or some other revaluation to dramatically increase the per-share price. I'm curious to read what others in the community think about this! |
||
|
|
splanky
Member
|
Lynn: Thanks for your comments--much appreciated! To solve some of the problem stocks in the original sim, I changed to stocks >.29 and got a more reasonable annual of 201.48% ! Question: Although I enter a commission fee of $5 (BrownCo) the fees per transaction are considerably more....are fees related to the value of each transaction, or what? |
||
|
|
marco
|
splanky: when you run a sim there's a default 0.5% slippage for each transaction. ---------------------------------------- Portfolio123 Staff |
||
|
| [Show Thread Printable Version] [Post new Thread] |
