| Index | Recent Threads | Who's Online | User List | Search |
|
|
 |
Portfolio123 » List all forums » Forum: Latest Release » Thread: IMPORTANT: changes to EPS%TTM may affect many ranking systems |
|
Total posts in this thread: 15
|
[Request a Feature] [Add To My Favorites] [Watch this Thread] |
| Author |
|
|
marco
|
We've implemented the changes to the way EPS%ChgTTM is calculated to reflect how Reuters is now handling this ratio. The new way does not assign an NA to EPS%ChgTTM with -ve EPS. It is calculated as follows: 100* (EPSTTM - EPSPTM)/ Abs(EPSPTM) In other words if a company has smaller -ve EPS vs the previous period a +ve EPS% growth is assigned. For more on this see this post: http://www.portfolio123.com/mvnforum/viewthread?thread=1055 Since many ranking systems use EPS%ChgTTM they will give considerably different results. To revert to the old EPS%ChgTTM a function needs to be used. The factor needs to be replaced by the following formula: eval(EPSPExclXorTTM>0 and EPSPExclXorPTM>0,EPS%ChgTTM,NA) Since the effects of the new EPS%ChgTTM have not been fully explored, the Portfolio123 Ranking Systems that use it have been modified to use the formula. The following P123 Ranking Systems have been updated: Balanced4 Foolish 8 Technamental Valuesheet IBD Smartselect IBD Smartselect (normal dis) At first glance the Model 'GARP Top Ranked' seems to perform better with the new EPS%ChgTTM. We will test the models to see if the new EPS%ChgTTM is a better factor. Let us know your findings as well !! ---------------------------------------- Portfolio123 Staff |
||
|
|
tonysf
Advanced Member
|
Using EPS%ChgTTM seems to yield better results than EPS ChgTTM in most simulations. |
||
|
|
BJS
Advanced Member
|
I have a couple of copied sims using VladsNewGrowth. The EPS change caused the results of these sims to go down (annual gain went down roughly 10-15%). Brian |
||
|
|
pafrisbie
Member
|
I changed the EPS%ChgTTM in my ranking system the way macro described, so my ranking systems would be the same. When I rerun old simulations and do not change any buy/sell rules, I get different results. My return may slightly improve or get worse, but my maximum drawdown got slightly worse. At first glance, Reuters may have changed other EPS numbers like EPS%ChgPYQ and/or EPS3YCGr%. When I tested with the new EPS%ChgTTM, all simulations and ranking systems that I used, the return lowered with a higher drawdown. |
||
|
|
eaves
Advanced Member
|
It is important to know which data items have in fact changed. Other examples are EPS%ChgPYQ and NI%ChgPYQ. Marco, can we get a comprehensive list of which items have changed? Thanks, Ben |
||
|
|
p123robert
Advanced Member CANADA Joined: Nov 7, 2005 Posts: 372 Status: Offline |
Yeah, let's salvage what's salvageable! May we ask Marco for a comprehensive list of items that have changed? I've got a couple of copied sims that use Balanced4. The problem is, this EPS change has caused the active return of these sims to drop from 89% in 2005 to 19% in 2005. And that's a huge drop in performance that we cannot and shouldn't ignore. |
|||
|
|
Stittsville123
Advanced Member
|
I raised two feature requests in the past. One was for a S/W revision history. Second was for proposed features in next release. I don't think these feature requests got much attention. In my opinion we need more communication before and after S/W & database structures are revised. Otherwise we are all working in the dark. Simulation results change from one day to the next, new errors crop up, etc. In many cases, the full impact of a change may not be understood and it would be a good idea to hear from the community prior to change being made. |
||
|
|
p123robert
Advanced Member CANADA Joined: Nov 7, 2005 Posts: 372 Status: Offline |
I agree, communication is important both before and after revisions are made. P123's ranking system (including all invisible software algorhythm and database) is the foundation upon which thousands of manhours of our labor is based. When the foundation moves, it has a big impact . I'm concerned about changes in the P123's ranking system. What should we do, if the (mostly hidden) algorhythm and/or database calculations in P123's ranking system are changed? Should we abandon yesterday's star performer port, and choose another, better-looking port? And do this every 6+ months? Some changes are inevitable, and some changes are not. Some changes come from Reuters, and some changes come from P123. As an investor, don't these changes make you feel a bit more concerned about the future? Even if we accept that some changes are inevitable, what's that we can **do** to make these changes work for us, not against us? My sims are based on "Balanced4" or "Fiscal Momentum". For some unknown reason, this change has had a negative effect on all my sims. Should I abandon yesterday's star performer port, and choose another, better-looking port? And do this every 6 months? ---------------------------------------- [Edit 1 times, last edit by p123robert at Nov 24, 2005 11:27:32 PM] |
|||
|
|
p123robert
Advanced Member CANADA Joined: Nov 7, 2005 Posts: 372 Status: Offline |
One observation: Annual returns are mixed... sometimes the effect is a small positive change, sometimes a large negative change. But the following is universal... this change has had a universally negative effect on all drawdowns in all of my sims. ---------------------------------------- [Edit 1 times, last edit by p123robert at Nov 26, 2005 10:07:52 PM] |
|||
|
|
marco
|
We're still trying to figure out all the ratios that have changed. Reuters is telling us that the change affects all period to period growth rates, therefore we believe ratios like EPS%ChgPYQ, NI%ChgPYQ, EPS3YCGr% are affected. We've been asking them to notify us when changes are made to their ratios, but sometimes this does not happen. While we believe that Reuters generally makes the best decisions on how to calculate ratios, for our purposes consistency is the key. We're exploring how to calculate the ratios ourselves from the financial statements but this is quite a ways from now. The best we can do for now is let you know what is happening so you can decide whether to use a ratio or not. The different results due to EPS%ChgTTM were expected because suddenly companies with -ve earnings may get high ranks. What we're interested to know is if big changes are still occuring using the workaround. You may not get exacly the same results using the work-around since you are depending on two different factors, but it should be close. I'd like to stress that calculating ratios is a very complex task and changes/errors happen. The simulations are, and will always be, an aproximation. When a large porfolio is simulated outliers cancel out. When small portfolios are used changes can be more dramatic. If small portfolios is what you are after, I suggest using large portfolios while designing the model, and going to a smaller # of positions at the end. THis makes sense since there really is no difference between two stocks with ranks of 99 and 98. Too much science applied to what is basically human emotions can lead you nowhere. Therefore success with quant methods is a balancing act. Even with all these "challenges" we're commited to this way of investing. After all, who wants to go back to a daily grind of decisions and indecisions? ---------------------------------------- Portfolio123 Staff |
||
|
|
[Show Thread Printable Version] |
Re: IMPORTANT: changes to EPS%TTM may affect many ranking systems
