nibor100
1,099 posts
msg #145885
- Ignore nibor100 |
1/6/2019 11:56:16 AM
@ksk8,
1. Here are the my last results of my SF historical checking filter and they cover the 252 days preceding 12/28/2015:
a. 75 Trades, 53 winnners and 22 losers for a 71% win rate,
b. My SF historical checking filter returns 4 stocks whose trades are past the 999 day limit for looking at specific trade data(not sure why those 4 were allowed to appear)
Probably should subtract 4 from the total trades above, etc.
2. Since Mahkoh and I have backtested using 3 different data sets and are not finding winning percentages any higher than 75% or so:
a. Makes me wonder what stock data universe your 14 year backtest in Portfolio 123 was conducted upon.
and/or
b. Are you certain your corrected SF filter criteria exactly match the backtest stock selecting criteria you used in P123?
Maybe you used > = to .10 and < = to $10 instead of the >.10 and <$10 being used in your SF filter or something else is different because 95% is way different than 75% when you consider that the latest years data in SF should be fairly representative of the possible trades taken.
Thanks,
Ed S.
Ed S.
|
KSK8
561 posts
msg #145913
- Ignore KSK8 |
1/8/2019 10:27:56 AM
Nibor,
Even though I don't believe some of those results may be accurate, I do suppose you defeated the mechanical prospect of it. But if only it could backtest and count the other factors I mentioned. Nevertheless, it still outperforms and I still feel it holds a degree of semi-invincibility-when utilizing the other components I mentioned especially when it is properly analyzed fundamentally.
I'll be making a thread soon that tracks the performance of it...After all, the proof is in the pudding!
Once again, I am no expert at utilizing S/R in SF code so I still ask others to contribute a heavy resistance component.
|
nibor100
1,099 posts
msg #145914
- Ignore nibor100 |
1/8/2019 10:46:25 AM
This is a test,
I've uploaded a spreadsheet of backtest results for the period of Jan thru end of May 2018 to Flickr with link below:
https://flic.kr/p/2e1Ub47
Someone let me know if they can access it. If you can in most browsers if you hit CTRL + it will keep making the image larger.
Thanks,
Ed S.
|
johnpaulca
12,036 posts
msg #145916
- Ignore johnpaulca |
1/8/2019 10:58:13 AM
no problem seeing it.
|
KSK8
561 posts
msg #145917
- Ignore KSK8
modified |
1/8/2019 11:01:53 AM
INNT- reverse split
CPHI- didn't even happen?
HTBX- another reverse split
MARA- reverse split (even though it won)
DCAR- reverse split (even though it won)
NCTY- reverse split (even though it won)
---
Factoring in my additional rules I stated earlier in this thread;
No losses exceeded 20%, hmmmmmm.....
|
nibor100
1,099 posts
msg #145920
- Ignore nibor100 |
1/8/2019 11:34:52 AM
@ksk8,
I'm unclear about your reverse split rule.
a. is it only not taken if the reverse split is the day before? week before?
or
b. Once a stock has had a reverse split you never take a trade in it again?
Thanks,
Ed S.
|
nibor100
1,099 posts
msg #145921
- Ignore nibor100 |
1/8/2019 11:43:34 AM
@ johnpaulca,
Thanks for letting me know you could access the spreadsheet image file.
1. Apparently with a free Yahoo email account we get a free Flicker account for storing up to 1,000 images which after uploaded to Flickr will have a URL that will allow anyone to access the image.
2. The file I tested with was from an Edgerater Pro backtest of the KSK8 corrected shorting filter pasted into Excel and then copied with the Windows Snipping Tool and uploaded to my Public page in Flickr.
Please note that my data universe in Edgerater Pro has stocks that are adjusted for both splits and Dividends. Probably not a big effect on the prices for most of the stocks selected by his filter.
I'll have to work on what size spreadsheet pages to upload for immediate easy readability.
Ed S.
|
KSK8
561 posts
msg #145922
- Ignore KSK8 |
1/8/2019 11:54:00 AM
Reverse stock splits that were announced within 8-9 months. Keep in mind, dilution on the stock never starts immediately. Regardless of how high it shoots up, it'll come down eventually but the crowd hype initiated from it is brutal on short sellers, hence why I avoid it....
|
nibor100
1,099 posts
msg #145923
- Ignore nibor100 |
1/8/2019 12:57:30 PM
@ksk8,
Did you look at the Maximum Favorable Excursion column on that spreadsheet image, it looks like every one day short, in that backtest, would have had a greater profit if exited earlier than the close, including all of the losing trades to include the those involving reverse split stocks.
Not sure if there is logical and repeatable way to take advantage of that phenomenom....
Also, in early posts where you told us to not take a trade if RSI(2) was above 98 the day before the signal date on your original filter, it seemed to me that most of those signals if shorted on 2nd day after, once RSI(2) exceeded 99 would result in a profit. I believe Mahkoh noticed some better results in his backtests related to the 99 level.
Ed S.
|
KSK8
561 posts
msg #145924
- Ignore KSK8
modified |
1/8/2019 1:18:47 PM
On that note, perhaps someone could craft up a statistics filter that is based upon the RSI(2) and its threshold "98" or "99". I was a big fan of Run Forest Run but didn't quite know how to go about modifying it to be tailored for this type of filter.
So maybe exploiting this phenomenon could begin by giving it a statistical edge?
|