corsino
259 posts
msg #48987
- Ignore corsino |
1/6/2007 1:33:13 AM
RumpledOne
I don't know why, but adding RSI(2) below 1 to my copy of the filter did improve the W/L ratio to 72/25. I don't know why others get different results. Maybe their entry filter or other testing settings are different.
|
corsino
259 posts
msg #48988
- Ignore corsino |
1/6/2007 2:05:44 AM
My Settings/Results
Entry Filter : 11 DAYS DOWN/RSI(2)<1
Start Date : 09/06/2006
End Date: 01/05/2007
Stop Loss :N/A
Minimun Holding Days : N/A
Maximum holding days: 5
Results: 33 stocks
Win % : 72%
Lose % : 25%
W/L Ratio : 2.88:1
Reward/Risk: 2.99
ROI(%): 143.99%
|
wallman
299 posts
msg #48991
- Ignore wallman |
1/6/2007 9:34:32 AM
"That's strange...adding RSI(2)<1 doesn't change the W/L ratio of 70/28. Or the number of stocks.But maybe it might for other 4-month periods."
Strange?
Think about it,how many stocks down that many days in a row do you think won't be RSI(2)<1
|
corsino
259 posts
msg #48997
- Ignore corsino |
1/6/2007 2:31:44 PM
Wallman
Actually, more than one would think.
The filter without RSI(2)below 1 resulted in 65 stocks.
The filter with RSI(2) below 1 resulted in only 33 stocks,although it raised the W/L ratio by 2%.
RumpledOne's must have had a syntax error, so the computer did not recognize his added line.When I added it to my version with a "below 1" wording,the computer recognized it.
So it would be a tradeoff whether a 2% increase in the W/L ratio would be woth cutting the number of stocks in half.
Of course, this is merely an exercise which may not have anything to do with actual trading.
|
corsino
259 posts
msg #49002
- Ignore corsino |
1/6/2007 6:03:12 PM
On further review, it appears that the backtesting feature, though interesting, is almost worthless.
I used the 11-Down filter starting with a 10-day offset, then 9-day offset, etc...
My count, buying only the first time a stock appeared, was 5 wins and 4 losses.
SF's count was 9 trades completed, with 8 wins, for 88.89% win percentage!
I don't understand......
|
wallman
299 posts
msg #49011
- Ignore wallman |
1/7/2007 12:10:28 AM
Corsino,
Hmmmm,it's hard to believe so many stocks down 11 days in a row would not have an RSI(2)<1.
Please don't get me wrong,I believe you, but did you check the charts on them to see if the filter is actually giving you correct matches?
Regards,Muddy
|
corsino
259 posts
msg #49012
- Ignore corsino |
1/7/2007 2:35:00 AM
Wallman
I think you are right. I reran the filter twice (for the latest 4 months), once with the Rsi(2)below 1 requirement, and then without the requirement and the results were identical : 65 stocks with a W/L ratio of 70/28.
However, I'm not convinced that their W/L ratios are correct, based on a 10- day backtest where I manually added the wins and loses, and the result was nowhere as good as their result.But I don't know how they arrive at their numbers.
|
corsino
259 posts
msg #49013
- Ignore corsino |
1/7/2007 3:18:34 AM
FWIW...
The RSI(2)<1 filter by itself, backtested over the 4-month period 09/05/2006 to 01/05/2007 with the same 5 day hold, returned 1406 stocks with a W/L ratio of 56/43.This is not to imply that it's not as good as the 11-day down filter, since the filters may be used in a different manner.I should add that most of the stocks returned by the 11-day down filter have RSI(2) values considerably below 1. So it's probable that a filter with the requirement of RSI(2)well below 1 would have similar results as the 11-down day filter.
|
TheRumpledOne
6,411 posts
msg #48983
- Ignore TheRumpledOne
modified |
1/7/2007 10:52:37 AM
Thoughts...
Added 1/7/2007
|
wallman
299 posts
msg #49020
- Ignore wallman |
1/7/2007 12:34:58 PM
Corsino,
"I should add that most of the stocks returned by the 11-day down filter have RSI(2) values considerably below 1. So it's probable that a filter with the requirement of RSI(2)well below 1 would have similar results as the 11-down day filter".
Ok,i believe that.
One thing you might try if you wish is backtest a filter to show fast%k<.01 to find very oversold stocks if that's what you are looking for.
Just add RSI(2)column but don't add it in the filter.
Also add column Williams%R(10).
This should give you all the fast%k 0's and williams -100.
You can see from the RSI(2) column if the stock is below 1 but I wouldn't automatically throw out those with over 1,hence only the RSI(2) column and not adding it into the filter.
There is actually someone who has a $39.95 pay for pick website that employs the fast%k 0's combined with Williams -100's almost exclusively,no doubt taken from Janeiro and myself. So I guess someone else thinks it works.
I've never formally backtested so I'm not sure.
|