| danclark |
|
Gold User, Member, TeleChart
|
| Registered User |
|
|
|
|
| Unsure |
|
| Thursday, March 31, 2005 |
| Friday, March 3, 2006 7:38:28 PM |
29
[0.01% of all post / 0.00 posts per day] |
|
|
Thanks!
|
Bruce,
Thanks! I think weighting is an excellent idea. To help me understand the weighting algorithm, is this Close and Volume example correct?
MG1 Close =
( StockA_MG1 Close * SQR(StockA_MG1 Market Cap)
+ StockB_MG1 Close * SQR(StockB_MG1 Market Cap)
+ StockC_MG1 Close * SQR(StockC_MG1 Market Cap)
+ ... to n stocks in the industry
) / n stocks in industry
Industry1 Volume =
( StockA_MG1 Volume * SQR(StockA_MG1 Market Cap)
+ StockB_MG1 Volume * SQR(StockB_MG1 Market Cap)
+ StockC_MG1 Volume * SQR(StockC_MG1 Market Cap)
+ ... to n stocks in the industry
) / n stocks in industry
Regards,
Dan.
|
Hi,
I find the MG... industries very useful. But, I can't find anything in the docs about out how the industry price and volume is calculated. I.e. how is the price and volume related to the underlying symbols?
Is it volume weighted? Market Cap weighted? An average?
Regards,
Dan.
|
netmgr,
I use TC2005 and AmiBroker software with the TC database providing data for both.
AmiBroker is somewhat like MetaStock with some heavy back-testing capabilities. Many AmiBroker users are ex-Metastock users.
I just started using AmiBroker, but I like it a lot in combination with TC2005. They are both good packages, but have different strengths. The combo is excellent, IMO.
One major benefit of AmiBroker is that it reads data directly from the TC2005 database. No Export/Import is required. Just run TC2005 each night and update your TC data. Then open AmiBroker and use the TC data.
And, it uses the native TC Industries and Sectors (which is a biggie for me). When you first install AmiBroker and set it to use TC2005 data, you click a button that says something like "Retrieve TC Industries and Sectors". It runs for about 10 seconds and from then on AmiBroker uses the TC categorizations.
I hope this helps.
Regards,
Dan.
|
Gents,
Thanks much for the great feedback. However, the winner is (drumroll)...
Bruce_L!
Bruce, that will do the trick.
Thanks and regards,
Dan.
|
Bruce, Doug,
Thanks for the feedback.
Regarding the scan itself, I expect that the scan will perform correctly. However, just because a symbols meets scan criteria does NOT mean that is a reasonable choice for trading.
If you have very "tight" criteria, you get fewer symbols, but you may exclude good quality symbols that are excluded because an ideosyncracy in the data . If your criteria are "loose", you have a better chance of including good quality symbols, but you will get more symbols that are "bad" overall. I.e., they meet criteria, but the chart looks horrible.
Regarding using a Personal Watchlist of "Good" symbols is that it would difficult to create and then difficult to maintain. Here's the problem...
For example, you are looking for symbols that meet two criteria: a) price has increased 10% in the last month and b) has a "good-looking" chart. The first criteria is easily scanable, while the second is much more difficult.
In my experience, when you run a scan, you will get three categories of results:
1) Symbols the meet all of your criteria nicely. They meet the criteria of the initial scan and then your visual chart review criteria. E.g., It's up 12% in the last month, trending up nicely, and bouncing off of it's trendline (or whatever).
2) Symbols that fail the first criteria. If a symbol's price has dropped 10% in the last month, it's pretty much irrelevant what the chart looks like.
3) Symbols that meet the initial scan, but fail your visual chart review criteria. E.g., Price has increased by 12%, BUT it has wild volatility.
The problem is that, on any given day, you might get 10-20 symbols in category 1, 7000 in category 2, and 80-100 in category 3. Category 1 symbols are fine. But category 2 and category 3 symbols are a big problem...
If I simply scan all symbols, I'll keep getting the "bad" 80-100 symbols over and over for several days. That's a waste of time.
I want to maintain a list of symbols that I do NOT want to trade. I do NOT want to waste time seeing a bunch of "bad" symbols included in my scan results when I have NO interest in trading them.
Regarding a separate "good" watchlist, it's a whole lot easier to maintain a list of say 200 or so "bad" symbols than to maintain a list of 7000 "good" symbols.
Let's say I scan for a 12 month high (Price > 3, Volume > 100,000). Right now that would return about 10 symbols. Some look good, but FORD and XPRT "Poor" to me. One because of overhead supply concerns and the other because of wild volatility.
But tomorrow, if I run the same scan, I'll may see the same symbols again. And I have to exclude them for the same reasons. That's a waste of time.
I hope I've explained this well.
Regards,
Dan.
|
Is it possible to filter a Watchlist or apply a SubList with a "Not In" filter? I.e., one of the criteria for inclusion in the list would be that the symbols were "NOT IN" another WatchList of "bad symbols".
Here's the problem...
When you run a scan or a Sort By process to find a list of symbols you will typically get a lot fo false hits that need to be filtered out by further review (like on a chart). They may fail second-level criteria because of bad price action, volume doesn't support price, too much or too little volatility, etc. But your scan will probably pick these bad hits day after day.
For example, on Day 1, I run a scan and get MSFT, CSCO, IBM and AAPL and 100 others as hits. After reviewing their charts, I see that only MSFT and AAPL meet my criteria. CSCO and IBM are false hits. (Let's assume that CSCO has too much volatility and IBM has a bad trend.)
So next day, I run the scan again and I'll probably get the same four stocks again. The probablity that the CSCO's volatility and IBM's bad trend have changed in one day is miniscule. But I have to recheck them, because they scan picked them up. This is a major waste of time.
What I'd like to do is to load the "Bad" symbols to a WatchList to filter my scan list so that the "Bad" symbols are NOT included in the scan.
Is there a way to do this in TC?
Thanks and regards,
Dan.
|
Craig,
My apologies... I should have complimented you too. The video clearly describes Jim's approach and greatly helped in understanding the systemic process. It is EXCELLENT!
Regards,
Dan.
|
Jim,
These are excellent concepts. Perhaps others should have already said it, but at least I will:
Thank You!
Regards,
Dan.
|
Doug,
Thanks for the feedback. One minor clarification...
The Sub-List needs to be applied at the Industry or Sub-Industry level to stay "stuck" for the components. I was applying it at the component level and then navigating back to the "Last List". (I.e., my bad.)
On the other question, is there a keyboard shortcut to implement this a little more quickly?
Also, is there any way of getting the components for ALL of the watchlist Industries? Or do you have to do it industry by industry?
(It's still a great feature, I just want to speed it up.)
Regards,
Dan.
|
|