[QBO] Part 2: Is Momentum More Profitable Than Quality?

6th Aug '21 by greensiax

10 comments 2590 reads

In my previous post, Part 1: Some Analysis of 2020 Using Back Testing and Stockopedia Rankings in a Breakout Strategy, I introduced my Quality Breaking Out (QBO) strategy, and my back testing approach. I then analysed its performance with simulated trading in 2020.

The first area for further investigation that came out of that analysis was:

"Further Analysis Item 1: The make up of the Tradeable World clearly has a big impact on the strategy performance."

The QBO Tradeable World is formed of 4 aspects:

An initial Quality Rank based selection (the main criteria)
A MarCap threshold as a filter
Additional ‘quality’ filters, on profit and debt
Liquidity filters, on spread, average traded value, and EMS

In this post, I explore the first of these aspects: the main ranking based selection criteria, and specifically whether a ranking other than Quality Rank could yield better results.

Key findings:

Taken in isolation Quality Rank performs much better than Momentum Rank
When Quality and Momentum are combined, with the other Stockopedia ranks, into the Stock Rank™ the performance can be much better than any individual rank in isolation (depending on the overall market conditions)
Quality is the most consistent factor over both of the the tested time periods (and overall market conditions)

Background and Approach

In my initial attempt at defining the QBO strategy I simply started with the basic idea of "Quality Breaking Out" and then refined the strategy using 'guesses' (or 'hypothesis' if I wanted to imply there was more scientific rigour...). The guesses were based on reading and research, which I then tested using back testing. Changes which seemed to work were incorporated into the strategy, ones which didn't (and there have been very many of them...) were dropped.

This time I decided to take a more methodical 'data centric' approach.

In order to examine the data, I first created a data set containing ranking and performance data for every Instrument as it was on the first trading day of each month across all of the data I have available. I added to that the future price movements over the next 1 month, 3 months and 12 months. I also added an 'Index Adjusted' version of the future price movements to smooth out the effect of general market movements, specifically the Covid-19 downturn (since we know pretty much all price movements during Feb / Mar 2020 were down!). All the…

Unlock with Email

Unlock with Google

Already have an account?
Login here

DavidMcCabe

8th Aug '21

Hi Greensiax

Thanks for taking the trouble to post this, it is always interesting to see someone take a purely statistical approach to investing, which I believe helps to take out the emotional risk out, and is far less work.

I have a few comments:

A spread < 1% really limits the stocks you can trade in, and is likely to exclude those that are smaller companies. If you take a look at the stockpopedia stock ranks backtesting, you will see the effects of the rank are most pronounced for small stocks. Obviously you don't want to throw away the nominal performance in spread costs, so a lot will depend on if you are short term trading i.e holding for less than three months, or looking at 1 year plus.

Also, again looking at the stockpopedia stock ranks backtesting, you will often see that a particular rank will do well for a few months, then dip, then do well again. The same applies to all of the Guru screens. Hence any test that is for less than two years, is likely to not be helpful.

Finally, it can be quite dangerous to look at very specific factors, over the short term, as it can result in just picking, or excluding a small number of very high, or very low performing shares that coincidentally share the characteristic. This gives a false sense that this is the key to making the right investment.

Hope this is helpful and of interest.

David

greensiax

Hi David,

Thank you for reading and taking the time to share your thoughts. You make some good points.

I said in my last post (and perhaps should have re-stated in this one) that my aim with this exercise is not to create a mechanical strategy, rather to explore the impact of, and formalise, some of the decisions / judgements I make during my trading. And also to learn by experiment. In my real portfolio I’m afraid the emotion (and work...) is still very much there ;-)

As for spread and company size, completely agree and I’ll be looking at MarCap next, followed by spread (and liquidity). I have quite a long list of things I’d like to explore - several years worth at my current posting rate!

I’d love to test over longer time periods but I’m limited a bit by the amount of (Stockopedia) data I have available.

Thanks again for your helpful comments.

Warm regards

Greensiax

DWit199

10th Aug '21

Is quality inversely correlated with volatility? If so, I would expect longer holding periods to be beneficial to a portfolio of high quality rank shares. What is the average holding period of your strategy?

Hi DWit199,

I don’t know how Quality correlates to volatility, though I would guess it is inverse. I’ll add it to my list of things to try. Thank you for the suggestion.

Holding periods is also on my list of things to experiment with. Currently the strategy is buy and hold (with the exception of the 49 day sell if not scaled in rule). I started with stops and a breakdown sell signal, but testing through the Covid-19 downturn proved that buy and hold was more effective. Again I’m going to re-visit stops and sell signals.

Thanks for reading and your comments.

Warm regards,

Michael_P

13th Aug '21

Hi Greensiax,

Thanks for taking the time to share your extensive findings. I've read this and your earlier post which I missed the first time around. I've been through a similar process to you finding that my personal trading results didn't match up to what I could have got mechanically following the Stockopedia ranks. Also similar to you I've done quite a bit of backtesting and faced some of the same hurdles. Issues with data quality and scope spring to mind (not from current Stockopedia data I hasten to add). There is loads of good stuff in what you've written. Having said that I wonder what your thoughts are about having such a big gap between your development and test timescales? Could you try using all the data up to the start of your test phase for development? I know that would take the backtest through the covid dip, but personally, I think that the best backtests should incorporate extreme conditions as this can give more confidence in how the portfolio may perform under stress. I've just been playing around with your interactive results in Figure 5 as well, which is a very expressive and effective way of presenting a lot of data interactively. Great work!

*** The following comments might not be of interest to most readers as they are about the software implementation used. ***

I've also got a technical background and was wondering how you did the backtesting? Did you write your own backtester or you use an existing library? It looks some of your charts are seaborn or ggplot. I'd be especially interested if you used python. Is there a github for any of this? Or perhaps you have a blog on a more software-oriented site that contains more of the technical details? If so please share a link. This is the sort of conversation normally I'd try to take offline but I don't know if there is a way to do that on Stockopedia, so apologies to any readers who are not interested in the software implementation side of this.

Anyway, as I say, this was a fascinating read and I look forward to any future content you are able to share.

All the best
Michael

Hi Michael,

Thank you for your very kind comments *blush* ;-)

I have a list of topics I plan to cover in the future, including: MarCap (next); Liquidity; additional quality filters (on debt and profit); Stops; Sell Signals; Minimum holding period (and a 'get out quick' early exit period); Budget management (e.g. expedited sales for scaling in to winners when fully invested); Position sizing; and using SUK2 as a hedge.

As you may have observed it takes me quite a while to prepare a post so this list should keep me going for quite a while yet...

I'm deliberately choosing the dev / test periods based on what I am trying to explore. In this case I was very interested in the very different market conditions pre- and post- COVID, and was trying to find a base selection criteria which performed with some consistency across the two. I totally agree that a valid test should include extreme conditions and I do plan to do a 'mega test' over all the data I've got at some point, probably once I have a better defined strategy. Perhaps as the 'grand finale' to this series ;-)

Technically....

I do use Python a lot, especially for data acquisition, analysis and visualisation. The plots in the post are all either matplotlib / seaborn or plotly. The interactive ones are plotly.

The back testing is all home brew. I have backlog items to look at zipline and PyFolio for backtesting, but when I started with this what I wanted to do was to simulate my own trading patterns as closely as I could (without the input of **ME** of course) and for that I had to roll my own. Plus I wanted lots of instrumentation to see what was going on.

I don't have a public GitHub repo, but one of the many backlog items I have is to do a post on data visualisation in Python which I would make public. The main reason I don't make what I have public is that it is so dependent on the data - some of which I pay for under a 'personal use only' licence.

Your suggestion of a technical blog is interesting, but I am very time-constrained (young family, day job, ...) so probably not right now.

Thanks again for your comments and suggestions, and I'm glad you found my humble contribution to the Stockopedia knowledge base interesting.

Lucas

27th Aug '21

I suggest a great book “What works on Wall Street”. And yes, momentum outperforms quality.

2nd Sep '21

Hi Lucas,

I have read "What works...", and I seem to recall that the book used relative strength to mean momentum. It is a fascinating book (if not exactly light reading...).

I'm not in any way trying to suggest that its not the case that some 'momentum' factors may outperform some 'quality' factors (or 'value' or 'growth' for that matter) in some strategies over some time period(s).

What I am suggesting is that the Stockopedia Momentum Rank does not outperform the Stockopedia Quality Rank for ME in THIS strategy over the periods I tested. My apologies if that was not clear enough.

Juan Carlos Bethen

3rd Sep '21

Outstanding work Greesiax, thank you!!!.

I speak as a truly noob here so pardon me for my lack of knowledge in the matter ^^. I started investing this January and what i've found so far during this time is that a good Relative Strength, let's say 3m/6m, of more than 15% plus a low volatility, less than 60% and checking your stock every month to see if it is still good for the checklist has worked pretty well for me, but if you mix it with a good (+80) Quality Rank then is AMAZING how this stocks tend to give you pretty good returns in a short period of time (1/3 months) so i have to agree that Quality really matters when you're looking for a good investment and want to have good short term returns.

For that i like to mix Quality rank of more than 80 with the Lakonishok Strategy but investing only in low volatile stocks.

5th Sep '21

Thank you Juan Carlos,

You describe yourself as a 'noob' but you seem to be learning very quickly! You look to have developed a strategy that works for you so very well done! I wish I could say the same ;-)

I hadn't come across the Lakonishok Strategy before so thanks for highlighting it. I do keep thinking I should invest some time looking through all of the Guru screens on Stockopedia as there is a lot of great knowledge / research encapsulated in them.

[QBO] Part 2: Is Momentum More Profitable Than Quality?

Background and Approach

Unlock the rest of this article with a 14 day trial

About the Author

Greensiax

10 comments

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

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