Stock Picks | QVM - Factoring In Odds For Success?

16th Aug '15 by iwright7

11 comments 13673 reads

Whilst Googling Stockrank style "Factor" investing I came across a June 2015 paper published by MSCI called, THE MSCI DIVERSIFIED MULTI-FACTOR INDEXES - Maximizing Factor Exposure While Controlling Volatility. The diagram below shows the net return of different Momentum etc Factors back tested over the last 40 years.

The risk/return characteristics of MSCI "Factors" 1975-2014

It seems MSCI offer various ETFs based on their Multi-Factor Index. MSCI factors are not the using the same QVM algorithms as Stockranks and the markets and re-balancing periods are very different. Nevertheless the MSCI factors are showing a useful return premium for each of the factors investigated, all of which have beaten the market over an extended period.

The principle MCSI Factor calculations are based on:

Value: P/B + PE + EV/CFO (except Financials)
Momentum: historic alpha, 6m return, 12m return
Quality: ROE + Debt/Equity + Low 5y Earnings Variability
Low Size: log of market cap

So it's clear that some ETFs have jumped on the Factor scoring bandwagon. We don't have access to the MCSI database, but we do have the Stockrank QVM Factor scores (and combinations thereof), which I believe are a brilliant extra addition to the private investors armory. In addition the Stockrank score is designed to be partially forward looking so we may have an algorithm edge that some other Factor scoring methods do not.

I do wonder if combining the high scoring Stockrank QVM Factor approach with additional weighting for additional "market anomalies", such as a significant recent Earning Surprise and/or Earnings Upgrade would further enhance returns? The same comment could apply to recent significant QVM upgrades ,or (as in my particular case) an additional Quality (>80) screening aspect?

For those interested in Factor type methodology the MSCI paper is available as a PDF download - but you may have to register as a company. Ian https://www.msci.com/documents/10199/a49f25c5-982e...

Unlock with Email

Unlock with Google

Already have an account?
Login here

Edward Croft

16th Aug '15

The Momentum Rank does already include Earnings Momentum as a 50% weight as a complement to Price Momentum. Much of the work of Lakonishok et al (who were key academics in backing up momentum as a factor) illustrated that price momentum was highly correlated with earnings momentum, and blending both has historically improved risk adjusted momentum returns.

While I personally believe there aren't enough systematic investors out there, we don't want to encourage over-crowding in a single ranking methodology (i.e. StockRanks). What we're working towards is the ability for anyone to build their own ranking methods at Stockopedia. So if someone wanted to weight their own method more heavily towards earnings surprises, then that's something we should be able to facilitate through the platform.

I'd rather we had a diverse family of competing community StockRanks rather than just our own set. So we're working on it (but it will take time) !

PS - the problem I have with many 'factor' funds is they tend to be designed to make money for the issuer rather than the end investor. There is a tendency for the issuer to 'design for scale', which means a bias towards market-cap weighting so that the fees generated are economic. Investors in these products need to make sure they understand what they are actually buying.

A classic example is the Dimensional UK Value Fund which has dramatically underperformed the index and the category by almost 4% annualised over 10 years. It's running £400m or so. The Dimensional website heavily markets the involvement of Fama & French in their efforts who were the original academics who discovered the value effect.

This fund is 'permitted to deviate from market cap weights' which means it tilts ever so slightly (scientifically) away from market cap weights... and invests very broadly. Sadly this hasn't helped returns.

55d0f1cdda2d0Fund_Performance_Total_Retu

I find it quite amusing that so many of the mega fund management groups have to make such a song and dance about 'tilting towards small cap value'. It seems what they really want to do is run a small portfolio! I remain to be convinced that any fund provider can do a better job than private investors managing their own money.

iwright7

17th Aug '15

Ed,

Interesting to hear about the future plan to be able customize the Stockrank factor selection criteria. What timescale are you looking at with this please - I am happy to be a Beta tester as soon as it is available.

You mentioned that Stocko subscribers should not all buy the same high Stockrank scoring stocks, but surely if this happened the tendency would be for high QVM stocks to rise and improved future returns would become a self fulfilling prophecy? In effect the computer is ironing out human bias to beat the market?

Nevertheless I like the human fallibility and computer pick combination and if this selection then goes wrong only I am to blame. Ian

Ramridge

"The Momentum Rank does already include Earnings Momentum as a 50% weight as a complement to Price Momentum."
.... except when there is no broker coverage, in which case the Momentum Rank is based only on price momentum factors.

cig

How common is the lack of broker coverage? Shouldn't you get at least the house broker passing through management estimates, except a few oddballs where the management has stopped talking to their broker?

30th Sep '15

I have been thinking further about the out performance of Momentum (M) in the MSCI study. In their 40 years of back testing Momentum factors beat Quality and Value by quite some margin (see top graph)

So I decided to cross check the historical MSCI Momentum outperformance with the more recent individual Stockrank factors since its inception. I excluded microcaps (>£50M) as too volatile and then went to the other end of the tab box selection (>£350M Market Cap) looking at the individual Q, V and M to see which if any are king. The % gain numbers in the table below are rounded to reduce numerical noise.

Stocko Factor	% Inception Gain >£50M Top Decile	%Inception Gain > £350M Top Decile
Q	40%	35%
V	30%	25%
M	60%	45%
QVM (Stockrank)	65%	45%

http://www.stockopedia.com/stockranks/performance/?marketId=gb&field=_MomentumScore_?=50&period=1&buckets=deciles

Sure enough the top Momentum decile beats the other 2 factors in both market caps sizes and smaller companies beat larger companies -An Elephants don’t gallop effect? The top Value decile has not done so well and Quality is intermediate.

Stockrank scores are calculated from a composite of Q, V, and M weighting and within a basket of shares with say a Stockrank >90 are companies with particularly high individual Q and/or V and/or M scores. This mix Q, V and/or M weighted stocks may be a very good thing for factor diversification.

But if we accept the historical MCSI data and the most recent Stockrank numbers then in a more mechanical selection system, maybe a high Momentum score should be receive higher Stockrank weighting, or be selectively screened for to maximise returns. Ian

Momentum tends to have rare but brutal crashes, which a conservative QVM weighting will mitigate. Stockrank since inception is a bit too short to assess the full lifecycle performance of the strategy, as we've been in a mostly benign bull market in the past few years, without such an episode.

Historical backtests look good because algorithms stay invested during crashes. It's harder for human investors...

WarrantStar

It's worth remembering that the very good screens on here ALREADY include the capability to screen for any combination of Stockranks. This can be done by Stockranks alone, or in combination with lots of other factors & ratios. Eg
Value Stockrank > 60
Momentum Stockrank > 70
PE ratio < 10
Yield > 2

If you don't like the way that one of the Stockranks is calculated on here then all you have to is create your own screen. eg You don't like the current Momentum Stockrank? Just create your own screen for the momentum factors and levels that you feel are appropriate.
I find that these screens are really good and well worth the effort of learning how to use them.

3rd Oct '17

For those of you interested in Factor investing there is a new discussion paper by AQR that argues that factor investing should be more of a craft than a science. The paper supports multi-factor selection containing an element of creativity. Ian

file:///C:/Users/Ian/Downloads/SSRN-id3034472.pdf

rpannell

I would love to read this paper Ian. Any chance of the web link rather than an internal files link

Thanks

pka

Hi Ian,

The link in your post doesn't work, but is this the paper you're referring to?

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3034472

pka - Yes that is the link. ian

QVM - Factoring in Odds for Success?

Unlock the rest of this article with a 14 day trial

About the Author

Iwright7

11 comments

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

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