Kevin_in_GA
4,599 posts
msg #105042
- Ignore Kevin_in_GA
modified |
2/18/2012 4:40:37 PM
Yes. It is built into the backtest filter in SS.
EDIT: But not into the SF filter. Ahh. Will try to revise filter to include liquidity as well as making sure that the open price in historical gaps was always above 1.
SECOND EDIT: Doing this pushes the filter performance too high for basic subscribers (like me). I'll think about this a bit to see if there is a coding workaround.
|
xzajic
14 posts
msg #105046
- Ignore xzajic
modified |
2/19/2012 1:23:20 AM
Why I'm asking: My backtest results ae similar to yours, but my watchlist is pretty different ;-)
|
duke56468
683 posts
msg #105053
- Ignore duke56468 |
2/19/2012 5:13:51 PM
mahkoh
173 posts
msg #105006
- Ignore mahkoh
modified 2/16/2012 5:49:53 PM
I found the best way to find trades on this filter is through Google docs. You'll have your picks within a couple of minutes.
I think it was mmaurice mentioning it. Use a form like this:
https://docs.google.com/spreadsheet/ccc?key=0AgVFJDk1gPekdFdxaTREX1QwV0hOX0NLcGt3eVF1WkE#gid=0
run the filter during the weekend, download csv file and copy "symbol" column and "%gapfill" column into columns A and B.
column C will show open price, column D previous close. E and F calculate % gap which you will find in column G. Sort sheet by column G from A to Z for gap down. Create a second document for the gap up and sort that sheet by column G from Z to A .
***********************************************************************************************************************************
mahkoh...Thanks for the spreadsheet, on monday morning does column G show the 2% or more gappers?
|
mahkoh
1,065 posts
msg #105063
- Ignore mahkoh |
2/20/2012 8:38:38 AM
Column G shows all the gaps. Sorting the sheet by column G (hover your mouse over it and click the triangle that shows up) puts the gaps in a descending or ascending order, depending on whether you want to see gaps up or gaps down.
|
xzajic
14 posts
msg #105066
- Ignore xzajic |
2/20/2012 11:45:23 AM
Kevin,
your current list of candidates on http://www.statisticalinvesting.com does not match your strategy presented here (or I'm missing something).
For example - your first LONG candidate is VRTU with 6 "gapdowns" and 6 "gapfills". Ok, backtests:
2011-05-16 - gap and filled (ok)
2011-08-08 - gap and filled (ok)
2011-09-06 - gap and filled (ok)
2011-09-19 - gap and filled (ok)
2011-10-31 - gap and filled (ok)
2011-11-21 - gap and filled (ok)
BUT: look at the AVG(VOLUME, 30)
2011-05-16 - 84970
2011-08-08 - 66895
2011-09-06 - 210904
2011-09-19 - 236275
2011-10-31 - 71800
2011-11-21 - 67509
4 from 6 trades were made with volume < 100000. So the question is: Is it really good "candidate"?
|
Kevin_in_GA
4,599 posts
msg #105068
- Ignore Kevin_in_GA |
2/20/2012 1:09:38 PM
Just letting folks know that the sharedlists are now updated.
I looked at adding in both a minimum liquidity (either $1,000,000 per day or $500,000 per day) as well as the requirement that all historical buys be at or above $1. Truth be told, the average trade return and win percentage were unaffected, but the absolute number of trades dropped by more than half. Thus total perfromance was reduced due to fewer trading opportunities.
I netted out at using a minimum liquidity of $500,000 per day for the historical stats but did not really care about the prior close. I also removed the requirement that the open be above the low (below the high) of the previous week, since it did not add much to the performance but makes real-life trading more complex.
The shared lists reflect the stratasearch list and not the SF list or a union of the two. The reason - each contains about 165 stocks already. That should be plenty.
|
Kevin_in_GA
4,599 posts
msg #105069
- Ignore Kevin_in_GA |
2/20/2012 1:31:09 PM
xzajic
6 posts
msg #105066
- Ignore xzajic 2/20/2012 11:45:23 AM
Kevin,
your current list of candidates on http://www.statisticalinvesting.com does not match your strategy presented here (or I'm missing something).
For example - your first LONG candidate is VRTU with 6 "gapdowns" and 6 "gapfills". Ok, backtests:
BUT: look at the AVG(VOLUME, 30)
2011-05-16 - 84970 x weekly open price of 19.58 = $1,663,172.6
2011-08-08 - 66895 x weekly open price of 16.14 = $1,079,685.3
2011-09-06 - 210904 x weekly open price of 13.66 = $2,880,948.6
2011-09-19 - 236275 x weekly open price of 14.97 = $3,537,036,8
2011-10-31 - 71800 x weekly open price of 16.03 = $1,150,954.0
2011-11-21 - 67509 x weekly open price of 14.04 = $947,826.4
4 from 6 trades were made with volume < 100000. So the question is: Is it really good "candidate"?
don't look at volume as much as average dollars traded per day. If the bet you are placing is less than, say 2% of this number, then the likely impact on the price is small. In each case even a relatively large bet of $20,000 would move the needle less than 2% in every trade but the most recent (where it still is only 2.11%).
Your point here is valid, and one that I have been addressing in the most recent list and modifications to the filters. If you are concerned about this specific level of liquidity, then just adjust your threshold to $1,000,000 per day or more. That should do it,but will give you fewer trades.
|
csupati
25 posts
msg #105078
- Ignore csupati |
2/21/2012 1:22:47 AM
Kevin, first of all I want to join those appreciating your efforts here. Thanks a lot.
I am wondering what is the real meaning of making a watchlist during the weekend and run the same filter against it on Monday morning. What would be the difference in the results by adding some rows like:
gapdown80 > 4
pct_gapfill > 0.75
close > 5
volume > 200000
sort by column 7 descending
...and run the filter against the global market. You lose 1 week from the statistics, that's it. You will try to enter the first 5 in any case.
I may be wrong of course.
Appreciate your answer.
|
xzajic
14 posts
msg #105080
- Ignore xzajic |
2/21/2012 6:53:39 AM
Kevin, it simply cannot work ;-)
Imagine you have perfect candidate with daily threshold cca 1 000 000 USD and 10 followers with fixed size 20000 USD trades.
10 trades x 20000 USD is 100000 USD, and it is 1/10 of average daily threshold on this share.
2-3 traders with LIMIT-OP-OPEN orders can win, but others ... 5 minutes after market open ... loosers ;-)
|
csupati
25 posts
msg #105081
- Ignore csupati |
2/21/2012 7:31:47 AM
Kevin,
OK, disregard my previous question, I found it on the 1st page. I was absent-minded. Thanks.
|