StockRanks | 3D Plots Of Stock QVM Histories - A Weekend Diversion

17th Oct '20 by Nick Ray

9 comments 866 reads

This post exists to provide a link to some HTML that I have created which use "WebGL" to create three-dimensional plots showing how the QVM ranks for a group of stocks evolve over time. Below is a screenshot to give you a flavour, but follow the link to play with the 3D version.

Mostly this is pretty rather than useful, and looking at conventional 2D plots of Q, V, M separately is easier to take in.

However what it can do is to show some patterns which you just don't see in the 2D plots. I will point out a few that I noticed later on.

How to use the link

Just click on the link and the site should load into your browser.

There are some brief help notes on the home page which are worth reading but I know you won't!

So I'll just point out:

This works best on a large screen - desktop or laptop. You can run it on a phone but it is hard to see very much.
Some of the pages load very slowly because the whole history has to be loaded and then compiled into WebGL before anything can happen. But once it's loaded it is very nippy indeed. Open each stock list in a separate tab if you want to flit back and forward between them.
You don't have to use the automations. If you are using a mouse you can grab and drag the cube around with the mouse and zoom in/out with the scroll-wheel. You can also drag the time slider to any date you like by hand.
The stocks you select will leave a trail behind which is about 1.5 years long and the current date is shown next to the slider.
The trail is turned off if you select "show all stocks". In that case you get all the stocks jiggling about like atoms of a gas.
When there is no data (my database is full of holes, especially for very large stocks) the sphere for that stock will display in grey until the animation reaches a date where there is data again.

A few things to look at

Individual Stocks

FTSE100: AAL clearly gets bigger as time goes by. The radius of the sphere for each stock is proportional to the…

Unlock with Email

Unlock with Google

Already have an account?
Login here

Phuntt

17th Oct '20

Wow, what an awesome new toy! Amazing work. Going to have to read your small print above carefully to understand how to rotate on an iPad Pro but I’ll work on it

Nick Ray

Either click on 'rotate' or grab the slider just to the right of that button and move it.

Macro

Love it!

Calling Mr Edward Page Croft to have a look-see!

Patxi

18th Oct '20

This is great work Nick, really interesting. I expect you are already familiar with the Relative Rotation Graph concept of Julius de Kempenaer, https://www.relativerotationgr...

RRGs are a proprietary way to show the relative strength change of stocks within sectors, or sectors compared to other sectors, over time. I first stumbled across them several years ago and was intrigued about how they worked. I attempted to generate my own version but the formulae are not public and it was clunky in Excel. I have since been using R for a few things but I clearly have more learning to do as I couldn't come up with anything like you have here.

I think Macro is right, Stockopedia should take an interest in this demo, it literally adds a whole new dimension to their already good data.

Hi Nick,

Thinking about how to turn your neat animation into money ...

Instead of having a Stop-Loss based upon Price could you do something based upon a 'Surface'.

Essentially define a Surface in QVM-Time space above which a position is Held, and below which a position _might_ be Sold (according to confirmation with Price).

Alternatively, define a Surface in QVM-Time space above which a position becomes a candidate to be Bought (according to confirmation with Price).

Or more outlandishly, a football-pools 'Perm', say 10 out of 15. The idea being that once a position is Bought from the Top 10 in QVM-Time space it is then held until it falls out of the Top 15?

As you can probably guess I am looking at this as a portfolio & candidate management tool.

Mike888

How did you manage to get the data Nick, I've been wanting to use stocko data for a while to create more sophisticated screening but have never figured out to grab it to the level of granulation you obviously have.

jwebster

19th Oct '20

Where is the historical data please? It would be interesting to run a regression model on values from 12 - 24 - 36 months ago against performance since.

Just to answer the questions:

I didn't know about RRG. Very interesting. These things are called "state spaces" - very popular with physicists and not so popular with undergraduates. It's made me wonder whether there are any other groups of state variables which could give an insight?

The idea of using a plane across a higher-dimensional area was also used by Altman to create the Z score. He used a five dimensional space and found the parameters for the plane empirically (i.e. fancy trial and error.)

You could do the same for QVM space: you just need to choose a linear combination of Q V and M. Stocko already do this: the Stock Rank is basically the planes defined by Q+V+M. More refined analysis might change the coefficients, for example: 0.5*Q+0.1*V+0.8*M. I think many people use zero as the coefficient for V for example and just track QM.

If you want surfaces which are not a plane you could use Bayesian inference and create a 3D contingency table calculating P("good performance" | Q | V | M). [I have done that in the past but my experience is that because Q and M are so strongly correlated and V is much weaker as a predictor you end up with a plane again.]

As to where did the data come from? I got it the hard way! I take a quick snapshot roughly every two weeks.

"As to where did the data come from? I got it the hard way!"

Crikey that is quite an impressive level of dedication !

3D plots of stock QVM histories - a weekend diversion

How to use the link

A few things to look at

Individual Stocks

Unlock the rest of this article with a 14 day trial

About the Author

Nick Ray

9 comments

Unlock the metrics for all these stocks and more...

Unlock the metrics for all these stocks and more...