Welcome Guest, please sign in to participate in a discussion. Search | Active Topics |

Profile: Darkrider
About
User Name: Darkrider
Groups: Gold User, Member, Platinum User, TeleChart
Rank: Registered User
Real Name:
Location
Occupation:
Interests:
Gender: Unsure
Statistics
Joined: Sunday, November 21, 2004
Last Visit: Thursday, July 9, 2009 10:00:52 AM
Number of Posts: 13
[0.00% of all post / 0.00 posts per day]
Avatar
Last 10 Posts
Topic: Using Linear Regression Sorts to Help Spot Divergences
Posted: Friday, September 30, 2005 3:44:05 PM
Thanks. So if I'm comparing a 10-day LR of price and want a comparable LR of TSV, I should merely use TSV rather than TSV10 correct? By doing it my original way (10day LR of price vs 10day LR of TSV10), I'm using 10 days of price data and 20 days of TSV data, isn't that correct?

Topic: Using Linear Regression Sorts to Help Spot Divergences
Posted: Friday, September 30, 2005 10:22:23 AM
Allow me to rephrase the question then...does a TSV1 provide the same data regarding money flows as a TSV10? Perhaps if you can provide a link to the TSV calc, I can judge for myself if you'd prefer. Thx
Topic: Using Linear Regression Sorts to Help Spot Divergences
Posted: Friday, September 30, 2005 8:43:32 AM
It seems to work well in sorts and turns up some interesting charts in EasyScan. My next question regards your thoughts regarding "optimal" periodicity (as if there is such a thing). Obviously with data, more is better, however I'm looking to get a sense for the short-term flow of funds that could signal institutional movement. Is there a minimal # of data points below which you'd conclude that the TSV LS is meaningless? I'm asking b/c you've seen a lot more of the raw TSV calcs than I ever will and I'm attempting to determine if 10 days is sufficient to get a valid TSV calc. Perhaps to re-ask the question, what's the mininum # of days required to calculate a TSV that is valid (I'm assuming it's probably more than 2 and less than 100). TIA
Topic: Using Linear Regression Sorts to Help Spot Divergences
Posted: Thursday, September 29, 2005 7:10:58 AM
Craig,

Here are the formulas I'm using. I've decided to calculate 10-day slopes for both price and TSV and have set up PCF for price slope as follows:

((((C + (C1 * 2) + (C2 * 3) + (C3 * 4) + (C4 * 5) + (C5 * 6) + (C6 * 7) + (C7 * 8) + (C8 * 9) + (C9 * 10)) * 10) - (55 * (C + C1 + C2 + C3 + C4 + C5 + C6 + C7 + C8 + C9))) / ((((1 * 1) + (2 * 2) + (3 * 3) + (4 * 4) + (5 * 5) + (6 * 6) + (7 * 7) + (8 * 8) + (9 * 9) + (10 * 10)) * 10) - (55 * 55)))

I then took this formula and added a subtraction of the TSV slope as follows:

((((TSV10 + (TSV10.1 * 2) + (TSV10.2 * 3) + (TSV10.3 * 4) + (TSV10.4 * 5) + (TSV10.5 * 6) + (TSV10.6 * 7) + (TSV10.7 * 8) + (TSV10.8 * 9) + (TSV10.9 * 10)) * 10) - (55 * (TSV10 + TSV10.1 + TSV10.2 + TSV10.3 + TSV10.4 + TSV10.5 + TSV10.6 + TSV10.7 + TSV10.8 + TSV10.9))) / ((((1 * 1) + (2 * 2) + (3 * 3) + (4 * 4) + (5 * 5) + (6 * 6) + (7 * 7) + (8 * 8) + (9 * 9) + (10 * 10)) * 10) - (55 * 55)))

When I combine the 2, I should get a solution of the slope of price less the slope of TSV:

((((C + (C1 * 2) + (C2 * 3) + (C3 * 4) + (C4 * 5) + (C5 * 6) + (C6 * 7) + (C7 * 8) + (C8 * 9) + (C9 * 10)) * 10) - (55 * (C + C1 + C2 + C3 + C4 + C5 + C6 + C7 + C8 + C9))) / ((((1 * 1) + (2 * 2) + (3 * 3) + (4 * 4) + (5 * 5) + (6 * 6) + (7 * 7) + (8 * 8) + (9 * 9) + (10 * 10)) * 10) - (55 * 55))) - ((((TSV10 + (TSV10.1 * 2) + (TSV10.2 * 3) + (TSV10.3 * 4) + (TSV10.4 * 5) + (TSV10.5 * 6) + (TSV10.6 * 7) + (TSV10.7 * 8) + (TSV10.8 * 9) + (TSV10.9 * 10)) * 10) - (55 * (TSV10 + TSV10.1 + TSV10.2 + TSV10.3 + TSV10.4 + TSV10.5 + TSV10.6 + TSV10.7 + TSV10.8 + TSV10.9))) / ((((1 * 1) + (2 * 2) + (3 * 3) + (4 * 4) + (5 * 5) + (6 * 6) + (7 * 7) + (8 * 8) + (9 * 9) + (10 * 10)) * 10) - (55 * 55)))

The problem I'm having is that I want to find companies where the 2 metrics' slopes are diverging, so I need to look at the change from yesterday to today. I've calculated that PCF as well, but started to realize that unless I can get them into an Easyscan, I'm wasting my time. What I'm attempting to do is identify on at least a daily basis (I'd love to do real time, perhaps 2x/day), companies where the relationship between price and TSV is improving (long) or eroding (short). I can add other variables to get the list smaller if I end up w. too many companies, but I'm really interesting in monitoring closely those names were TSV is doing something opposite of price.

Any suggestions you can provide regarding creating this type of model into an Easyscan would be greatly appreciated.

Topic: Using Linear Regression Sorts to Help Spot Divergences
Posted: Wednesday, September 28, 2005 12:05:45 AM
Is there a way to send you a private message? I've played w. the video instruction and set up a couple of PCFs but would like to discuss w. you off-forum if possible. TIA
Topic: Using Linear Regression Sorts to Help Spot Divergences
Posted: Monday, September 26, 2005 4:34:10 PM
More days are better on any regression, however given that I'm looking for divergences, I'd suspect that another over 50 days would probably be less useful. As far as TSV periodicity, I'm not sure what would be appropriate. I'm still new to the TSV calculations so if there's a min/max range that is necessary, I'd work within that. My gut says use 21-day TSV, but I'm very much open to suggestions.
Topic: TSV divergences
Posted: Monday, September 26, 2005 4:28:00 PM
I'm looking to set up a PCF that calculates the slope of price vs the slope of TSV. Then I want to take these PCFs and use them in an Easyscan such as Slope P > Slope TSV [use to identify potential long]. Would be looking at no more than 21 trading days I suspect.

TIA
Topic: TSV divergences
Posted: Monday, September 26, 2005 4:11:35 PM
great video! is there a way to put the regression into a PCF?
Topic: Using Linear Regression Sorts to Help Spot Divergences
Posted: Monday, September 26, 2005 3:59:29 PM
Is there a way to build price/TSV regressions as PCFs?
Topic: TSV divergences
Posted: Friday, September 23, 2005 3:09:27 PM
Is there a way to plot divergences b/t price and TSV? As an example, price trend up but TSV trend down (widening absolute spread) or TSV trend rising while price is flat/down (again widening absolute spread).

TIA