| Index | Recent Threads | Who's Online | User List | Search |
|
|
 |
Portfolio123 » List all forums » Forum: Screener » Thread: Stitts Challenge 2 |
|
Total posts in this thread: 7 |
[Request a Feature]
[Post new Thread] [Add To My Favorites] [Watch this Thread] |
| Author |
|
|
Stittsville123
Advanced Member
|
I am issuing this challenge to take people's minds off recent performance results. The challenge is simply this: Who can come up with the best "canary in the mineshaft" indicator for the Russell2000? I am kicking this off with my effort. This indicator is built around the P123 screener. I have built a screen which sorts by the highest yielding stocks on the PRussell2000. http://www.portfolio123.com/screen_summary.jsp?screenid=10257 Then I run a performance backtest of 5 highest yielding stocks with weekly frequency. I dump the performance results to a file (see attached spreadsheet). WIth the spreadsheet I use the weekly change in the 5 highest yielding stocks to determine whether to go long the Russell2000 or short the Russell2000 the following week. If previous week had better performance than -3% then go long the following week. Otherwise short the Russell2000. I didn't spend too long on this effort - I am hoping that the P123 community can come up with something better than this. Steve ---------------------------------------- 
CanaryIndicator.GIF
(16847 bytes)
(Download Count: 258)
----------------------------------------
CanaryIndicator.xls
(160768 bytes)
(Download Count: 15)
|
||
|
|
crakes
Advanced Member
|
I played with this a little bit. I changed the logic to go short at -3 and stay short until positive. The results were about the same/slightly better with a little less volatility. The spread between Rut and SwitchRut peaked in Sept 2002 at 50+ points, dipped down to about 30 points better and has turned back up to 48 points better.... I see that the top 5 had some big swoons... even while the market moved upward. I like the concept - finding leading stocks. I just don't know how to pick one set that would fit the bill. Carl ----------------------------------------
ContinuousCanaryIndicator(2).xls
(182272 bytes)
(Download Count: 18)
---------------------------------------- [Edit 1 times, last edit by crakes at Aug 9, 2007 9:56:33 PM] |
||
|
|
dpf
Advanced Member SWITZERLAND Joined: Nov 18, 2003 Posts: 34 Status: Offline |
I like you idea Steve. However there may be some practical problems: 1)P123 has some weird data for dividend yields. For example, currently NFI is number one in the screen with a yield of 359.6 ! 2)How do you calculate RUT return? Do you take friday to friday close? If so, there is a look ahead bias, as you cannot know what the top 5 yielders are and what they performed and simulatenously buy or short the RUT on that close. Monday open to next monday's open is probably more correct. Daniel |
||
|
|
Stittsville123
Advanced Member
|
I am using the screener for both the high yield stocks and the Russell2000. I know the high yield stocks are apparently Monday Close to Monday Close. I'm not sure about how the benchmark is processed in the screener. Maybe there is a lookahead problem there. |
||
|
|
pcrutchfield
Advanced Member
|
The 25 day highs and lows have been very effective for timing RUT. Go long when a new 25 day high is made, short (or cash) on a new 25 day low. Avg long 0.076%/day, short -0.024%. Also, standard deviation on long days is 0.88%, but goes up on short days to 1.32%. I view this as more of a "risk management" tool than market timing. Spreadsheet attached. Also, the 25 day new high/lows can be used in ports -- buy rule -- HighestBar(#High,25,0) < LowestBar(#Low,25,0) sell -- HighestBar(#High,25,0) > LowestBar(#Low,25,0) These have improved many of my sims. A stop loss could be used below the 25 day to improve real results, but of course that can't be backtested here. This signal works across all indexes, though only RUT and COMP show negative averages on the lows. Standard deviation is lower across all indexes on the 25 day highs. We are currently at a 25 day low across all indexes. Here is the exposure list for RUT -- 2/21/01 4/18/01 1/23/02 3/4/02 6/19/01 10/26/01 05/10/02 08/22/02 09/18/02 10/24/02 12/30/02 01/13/03 01/24/03 03/17/03 08/07/03 08/18/03 03/16/04 04/02/04 04/30/04 06/07/04 07/14/04 09/01/04 01/05/05 02/16/05 03/23/05 05/18/05 08/12/05 09/09/05 09/22/05 11/10/05 01/03/06 01/06/06 05/15/06 08/30/06 01/08/07 01/16/07 03/01/07 04/03/07 07/24/07 08/10/07 It would be helpful for me if I could use exposure lists in the buy rule to limit new entries during these periods. That is what I do in the "real world". ----------------------------------------
RUT-25dHL.xls
(795136 bytes)
(Download Count: 14)
|
||
|
|
Stittsville123
Advanced Member
|
Hi Paul - This looks very interesting but it is a lagging signal. I guess I am interested in a leading indicator that would flag when to get out of the stock market, like use of a canary in a mine. I don't care so much about re-entry of the market. I probably misled everyone by producing graphs of long / short the Russell2000. The concept I was trying to develop was use of high yield stocks to indicate a pending stock market correction. The 25 day highs and lows look like good confirming indicators. Steve |
||
|
|
DennyHalwes
Advanced Member UNITED STATES Joined: Apr 28, 2004 Posts: 1532 Status: Offline |
Steve, An interesting approach! I'll play with this idea some and see if I can find a different colored canary. Yield is calculated by using the last 12 month's dividends and the current price. This is correct and works fine if the price is stable. However, NFI paid $11.20 in dividends when its price was about $120. It is now trading at $6.77, thus, a meaningless Yield. This will occur with all stocks that pay dividends and then crater in price. To minimize the effect of these meaningless yields you might want to mod your Screener rule to something like; Yield > 3 & Yield < 20 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 ---------------------------------------- [Edit 1 times, last edit by DennyHalwes at Aug 13, 2007 4:42:02 PM] |
||
|
| [Show Thread Printable Version] [Post new Thread] |
Re: Stitts Challenge 2
