| Index | Recent Threads | Who's Online | User List | Search |
|
|
 |
Portfolio123 » List all forums » Forum: Simulations and Portfolios » Thread: Very Odd Ranker Performance Change |
|
Total posts in this thread: 4 |
[Request a Feature]
[Post new Thread] [Add To My Favorites] [Watch this Thread] |
| Author |
|
|
sglinski
Advanced Member UNITED STATES Joined: Feb 16, 2006 Posts: 139 Status: Offline |
For this ranker: http://www.portfolio123.com/app/ranking-system/90729 ---------------------------------------- 
Canslim_2a_ranker_Reuters.jpg
(44396 bytes)
(Download Count: 128)
----------------------------------------
Canslim_2a_ranker_Compustat.jpg
(42052 bytes)
(Download Count: 128)
|
||
|
|
DennyHalwes
Advanced Member UNITED STATES Joined: Apr 28, 2004 Posts: 1532 Status: Offline |
Sglinski, I copied your ranking system and re-ran your performance, except I changed it to 100 buckets. Here is my copy: http://www.portfolio123.com/app/ranking-system/158064 I got the chart below, and the values of the bars are copied here: -2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -0.8 -1.1 1.2 1.5 5.7 7.4 6.5 -3.2 17.8 -0.2 1.0 0.2 4.7 5.5 7.8 1.0 2.2 0.0 -0.3 -0.1 1.5 4.2 6.7 9.4 2.6 -1.1 0.0 -0.0 0.1 2.8 5.4 8.7 8.6 -0.6 0.3 0.0 0.6 0.2 2.9 5.0 9.1 5.8 -0.3 -9.5 0.2 0.7 0.7 3.8 4.3 6.3 5.9 11.1 0.0 0.2 1.7 0.1 3.0 2.6 7.9 12.4 -5.9 0.0 1.1 2.6 0.8 0.8 0.6 3.2 -6.0 19.1 0.0 0.0 0.0 0.0 0.0 -0.7 -2.4 -0.8 -21.6 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 As you can see, the 7 highest bars are all zero. Since you ran the performance with only 10 buckets, the zero buckets in my run were lumped in with the lower ranked buckets yielding a positive top bucket in your run. To help you understand this I copied all of your ranking system rules into the Screener here: http://www.portfolio123.com/app/screen/summary/49556 If you select the Totals button, you will see that by the time the filters get down to your Float < 1000 rule there are zero stocks that pass the screen. So, the rules are too restrictive. You only have 1 factor; Sortino(15,5,0)/Sortino(6,2,0) that is not Boolean in your ranking system. I think that the ranking system is ranking all the stocks by that factor and then filtering them by the Boolean filters. That is why the performance has zeros in the top buckets. In other words, the highest ranked stocks by the Sortino(15,5,0)/Sortino(6,2,0) rule do not pass the Boolean filters. So although they are highest ranked, they have zero performance. Only lower ranked stocks also pass the Boolean filters. Denny  ----------------------------------------
CanSlim_3.png
(8500 bytes)
(Download Count: 90)
---------------------------------------- "The significant problems we face cannot be solved at the same level of thinking that we were at when we created them". Albert Einstein |
||
|
|
sglinski
Advanced Member UNITED STATES Joined: Feb 16, 2006 Posts: 139 Status: Offline |
Denny, thanks for looking into this. After experimenting with your screener, http://www.portfolio123.com/app/screen/summary/49556 , I find that it is necessary to modify rule 7 (float<1000) to (Mktcap<100000), turn off rule 10 (Sortino) and loosen rule 8 (to Inst%Own<=60 versus <=30), to produce a list of stocks. The bad news is : Apparently for all x, float<x fails. Also, the Sortino ratio rule fails for ratio<1, and ratio >=1. which leads me to believe they may both be P123 bugs. The good news is: After making those above changes, the screener generates a decent equity curve (below). Steve ----------------------------------------
Canslim_3ScreenerDebug.jpg
(74543 bytes)
(Download Count: 78)
|
||
|
|
marco
|
Didn't see this post until now. I think I know what's going on: I think it has to do with the very discrete ranking you are doing. A node is either 0-1, so the number of stocks in a bucket varies wildly. Possibly due to the smaller universe of All Fundamentals with 10 buckets your top bucket sometimes has very few stocks, if any. You can check that using the 'Performance Graph' setting instead of 'Annualized Returns'. When the line goes flat there are no stocks. Here's some tests I run. With 9 buckets things go back to 'normal'. It's still not possible to see how many stocks per bucket though. To do that run a screen (image on the right) with only one rule and no slippage. You can see that number of stocks per period varies a lot With 10 buckets see image bottom left. The top bucket goes flat several times, meaning 0 stocks. I was able to reproduce with screener using Rank >90 which shows you at most 3 stocks pewr period, many 0's etc. Hope this helps ----------------------------------------
rankperf.gif
(83140 bytes)
(Download Count: 44)
---------------------------------------- Portfolio123 Staff |
||
|
| [Show Thread Printable Version] [Post new Thread] |
Re: Very Odd Ranker Performance Change
