Price between Linear Regression Channels 50 and 100 Lower band

[smartinv]

Hello,
I am using Stock Hacker (Scan function in TOS) to find stocks from my watchlist when the price comes in between the lower bands of LinearRegression Channel 50 and 100 bands (the standard function from TOS). On the chart this works fine, but the scan for 5 min aggregation does not produce accurate results. Any idea how to fix the scanning issue?

Thanks,
Shai

Shared Link - https://tos.mx/!vnnZ172S

Code for the scan:
declare lower;
def lookback = 1;

plot NearBottomOfChannel = LinearRegCh50().LowerLR > close or LinearRegCh50().LowerLR > low within lookback bars;
NearBottomOfChannel.SetDefaultColor(Color.GREEN);
AddLabel(yes,"BofCh: " + NearBottomofChannel, color.DARK_GREEN);


When you plot this on the 5 min chart it works fine. In scan it won't pick all stocks from the watchlist that match this condition as i can see this visually in the graph e.g. doesn't pick up CRM or COIN

 

[BenTen]

What's the watchlist for the scanner set to? Are you using extended hours on your current chart?

 

[smartinv]

What's the watchlist for the scanner set to? Are you using extended hours on your current chart?

Watchlist is set of my stocks. I am not using extended hours.

 

[BenTen]

In scan it won't pick all stocks from the watchlist that match this condition as i can see this visually in the graph e.g. doesn't pick up CRM or COIN

Can you share a screenshot of this?

Did you run the scanner again? Still not working?

 

[smartinv]

Can you share a screenshot of this?

Did you run the scanner again? Still not working?

Yeah the scanner is not accurate. I also have set alerts to send me email when the scan results changes so I don't have to monitor this all day.

 

[useThinkScript]

Sadly no. you can not use linear regression line in scans on lower timeframes.
On the highest timeframes, it is possible to sometimes get consistent results depending on the volatility.


An issue for all statistics attempting to use LRC
Linear regression is a dynamic calculation, not the pretty fixed line that you see after the fact.

As new data comes in - whether it's a new tick on the current bar or a completely new bar - the linear regression is constantly recalculating itself.

This means: The current bar is constantly repainting. Every single tick changes as the bar's high, low, close changes; it changes the linear regression calculation.

Once the bar closes. That data is gone and not accessible as it is now recalculating the new bar.
Given the ToS Scan Hacker is 3-4 minutes delayed; it will be unlikely to match up with a 5-min chart or a watchlist.


@smartinv

 

[smartinv]

Sadly no. you can not use linear regression line in scans on lower timeframes.
On the highest timeframes, it is possible to sometimes get consistent results depending on the volatility.


An issue for all statistics attempting to use LRC
Linear regression is a dynamic calculation, not the pretty fixed line that you see after the fact.

As new data comes in - whether it's a new tick on the current bar or a completely new bar - the linear regression is constantly recalculating itself.

This means: The current bar is constantly repainting. Every single tick changes as the bar's high, low, close changes; it changes the linear regression calculation.

Once the bar closes. That data is gone and not accessible as it is now recalculating the new bar.
Given the ToS Scan Hacker is 3-4 minutes delayed; it will be unlikely to match up with a 5-min chart or a watchlist.


@smartinv

Ok will try for higher timeframes.

 

[merryDay]

@useThinkScript is correct, statistically, attempting to nail down a consistent LRC plot is like trying to catch the wind.

But that is not the main issue with LRC channels in the ToS app.
There are two reasons why LRC numbers in the scans, watchlists, and charts will never match up.

The biggest reason is that ToS LRC math uses InertiaAll() and HighestAll()
These functions are written to create math based on the number of bars that you have residing on your chart.
Since the Scanner and Watchlist have no way of identifying how many bars are on your chart.
They are not doing the same math as your chart is doing.

If that was not a showstopper; Schwab compounded the issue by deciding to depreciate the
inertiaAll and the highestAll functions; as it was determined that they require an unfair share of server processing time. As such, these functions are no long updated in real time.

read more:
https://usethinkscript.com/threads/...in-real-time-in-thinkorswim.8794/#post-115472

@smartinv @BenTen @useThinkScript