| Index | Recent Threads | Who's Online | User List | Search |
|
|
 |
Portfolio123 » List all forums » Forum: Feature Suggestions » Thread: Feature Request: Daily historical fundamental data |
|
Total posts in this thread: 26
|
[Request a Feature] [Add To My Favorites] [Watch this Thread] |
| Author |
|
|
hyper
Advanced Member
|
Most active traders trade on a day-to-day basis, not on a weekly basis. This is my understanding from all the books and web sites that I have read and other financial markets trading message forums that I have frequented over the past 5 years. I look for new opportunities daily and update trailing stops on my open positions daily. It seems that Portfolio123 was built from a fundamental investor's perspective, and it seems that the people who manage it still remain fixated with such a frame of mind. There is a clear mindset mismatch between Portfolio123 and some of its most active and prolific members. I found it very surprising that such members as Denny and Olikea only found out recently that daily fundamental historical data is not stored and therefore simulations are tested using weekly data, even if "Daily" rebalancing is selected. This means that Simulations in which the rebalancing is set to "Daily" are flawed. Here's the post in June 2007 in which Denny found this out: http://www.portfolio123.com/mvnforum/viewthread?thread=2574&offset=0#11117 Here's the post in July 2007 in which Olikea found out: http://www.portfolio123.com/mvnforum/viewthread?thread=2660&offset=10#11618 Things can change quickly and largely in the financial markets. When something does change we should react in a timely manner, which means on the same day or next day, not on the following Monday morning. I have personally seen factor rankings change drastically during the week on many occasions. If such data was not available until for the following Monday, my reaction to such changes could have been quite late and costly - for both entering and exiting positions. In other fields, lack of precision can lead to collapsing bridges and buildings, computers crashing and losing data, electrical faults that lead to fire breakouts in airplanes, and ultimately loss of lives. In our case in the financial markets, lack of precision can lead to loss of money. I don't know about other people, but when I'm dealing with my money I want to be spot-on precise. So, coming from an active trader's perspective, such a statement as "the benefits are questionable" as marco said in this post: http://www.portfolio123.com/mvnforum/viewthread?thread=2574&offset=0#11153 is absurd. Please make daily historical fundamental data available. View the feature request here. ---------------------------------------- [Edit 1 times, last edit by hyper at Nov 7, 2007 11:13:21 AM] |
||
|
|
DennyHalwes
Advanced Member UNITED STATES Joined: Apr 28, 2004 Posts: 1532 Status: Offline |
Phillip, I agree with your desire for daily Rank values, and I voted for it. However, we need to keep in mind the problem with providing this data. Until last December P123 only took a snap shot of all the data on Saturdays. So we had 5+ years of weekly only data. In December Marco announced here that P123 started storing daily snapshots of Price, Volume, and Rank values. I thought at the time that P123 would be keeping this daily data and it would be available from that date forward for Sims. However, we found out in June that P123 was only keeping 1 week back of this daily data, and only saving the weekend data. That was the surprise and disappointment that we voiced in the post you referenced above. If P123 had kept the data, then we would now have almost 1 year of stored Rank data. The real issue though, assuming P123 agrees to supply daily Rank data for the full 6 years of data, is how does P123 go back and fill in the daily data? They currently only have daily info on price and volume values for the 6 years of historic data. The Rank values can’t be calculated from these values. In order to add this data correctly P123 would have to add all the daily fundamental data on the correct date that it was released by Reuters, not the date it was available in the market. How do they determine on what date during the week in the past fundamental data would have been available? This data might be available from Reuters, but I would think that if it is, it would be expensive. We currently have the purest accuracy in a historic data set that I have found anywhere. How do we add data after the fact without compromising the accuracy of this data? P123 would have to be VERY careful in attempting this. It would also increase the size of the stored by at least a factor of 4. I for one would prefer that P123 concentrate on adding the additional years of data before they start any effort to add daily Rank data. Denny  ---------------------------------------- "The significant problems we face cannot be solved at the same level of thinking that we were at when we created them". Albert Einstein |
||
|
|
marco
|
I'm not sure if daily sims even make sense at this point, and maybe they should be turned off. The idea of the daily sim was to process sell rules that trigger on price action alone (like a profit taking or stop-loss), and be able to replace the stock mid-week. What people are actually trying are active trading systems, with ranking systems having price factors, or worse, short term TA functions. These systems produce very volatile ranks that make the simulation meaningless. Others are running daily rebalancing ports and getting different results when recreating the system by simulating. This is always a bad thing that make people doubt the accuracy of P123. There are four options: A. Create the perfect tool that has daily updates stored and can rank intra-week. B. Create a middle of the road compromise C. Get rid of daily sims D. Get rid of daily fundamental updates ISSUES WITH EACH A. This is very very hard, if not impossible. The resources demands go up dramatically and the data is simply not available, or very hard to re-create. It would also be one-of-a kind: Compustat, for example, has one of the most respected Point-In-Time databases and it's only monthly snapshots (and costs like $50K/year). B. Run daily sims by re-ranking daily, but only functions with price functions actually change value. In other words, only the TA part of a ranking system will change. This will still not produce the same results as a daily rebalanced live portfolio, but it should be closer. It will also run much, much slower C. Disable daily sims. Maybe only process sell rules daily, but do not buy a replacement stock since the rank is stale. D. Ranking will only be on a weekly basis even for live ports. This will make the ports/sim equivalent Thank you for your comments. ---------------------------------------- Portfolio123 Staff ---------------------------------------- [Edit 1 times, last edit by marco at Nov 8, 2007 12:51:13 PM] |
||
|
|
Stittsville123
Advanced Member
|
Marco - There are several people using daily re-balancing and I think there would be a strong negative backlash if you eliminated daily sims. You created an issue by introducing daily ranking updates but without the ability to simulate. So the fourth option is to go back to what we had previously and drop the daily fundamentals (and daily ranking) but keep daily simulations based upon weekly rankings. In this option the buy/sell rules would be daily but ranking systems not. Steve |
||
|
|
marco
|
Steve, I've added the fourth option. Thanks ---------------------------------------- Portfolio123 Staff |
||
|
|
Stittsville123
Advanced Member
|
Marco - Here is another option, although more convoluted and probably a lot more work to implement: I think what originally drove the desire for daily re-balancing was more immediate access to fundamental data. i.e. if a company's poor quarterly results as reported by Reuters was posted on Monday you don't want to wait a week before finding out you should sell the stock. My idea is that you could process daily rankings but they would only be used to alert the user to significant changes in stock ranking within a portfolio. The user could see the alerts and decide whether or not to take action i.e. sell the stock immediately. This daily ranking would not be used for the manual rebalancing or simulations. Steve |
||
|
|
DennyHalwes
Advanced Member UNITED STATES Joined: Apr 28, 2004 Posts: 1532 Status: Offline |
Marco, I think that you left out one very important option: Leave It The Way It Is. This may be your B option: Create a middle of the road compromise. The current implementation is already one form of a middle of the road compromise. What we have now, even with it’s shortcomings in Sims, works GREAT in Ports, and I would hate to lose this capability. Denny ---------------------------------------- "The significant problems we face cannot be solved at the same level of thinking that we were at when we created them". Albert Einstein |
||
|
|
hmorris48
Advanced Member UNITED STATES Joined: Feb 24, 2007 Posts: 76 Status: Offline |
I would vote for: B. Create a middle of the road compromise (as Denny has suggested - don't change it) or D. Get rid of daily fundamental updates Dropping daily sims would be a very bad thing - my only active portfolio is a daily sim. -herb ---------------------------------------- The holy grail of trading system design is a perfectly smooth equity curve. Beyond Technical Analysis |
||
|
|
mrm
Advanced Member
|
Denny, I must missing something If we are trying to match simulations should be choice D better? |
||
|
|
Stittsville123
Advanced Member
|
Emil - Denny's suggestion "leave it as it is" allows you to have a quick peak at your ports every morning to see if there is a reason to sell a stock. One reason might be updated company financial data that causes the ranking to drop. Another reason is if you have a price-based buy/sell rule that is triggered. You can do this easily by manually rebalancing and looking at the sell recommendations. You don't have to accept the recommendations, you can discard them if so desired. But it gives you advance notice without waiting for the weekend to get the auto rebalancing. You can't match simulation results with your manual port manipulation but the port should be a better than situation because you can react faster to new financial data. Steve |
||
|
|
[Show Thread Printable Version] |
Re: Feature Request: Daily historical fundamental data
