The NAPS & SNAPS Portfolio Update - 25% Annualised Returns To Date And 20 New Selections

5th Jul '17 by Edward Croft

38 comments 31492 reads

When do you find the time to focus on your investing? Do you get up at 7am to read the news and plan your trades? Do you scan the bulletin boards hourly for new ideas? Do you watch the market in real time like a hawk setting buy triggers, adjusting stop losses and analysing market depth?

I know this is what the cool gang do… but for me, this all sounds too exhausting. It’s summer, it’s too hot, and at times like these I’m glad to have an investment process that practically runs on auto-pilot.

If you read this piece I’m going to explain how I’ve been putting together a 20 stock, diversified portfolio that's been growing at 25% annualised with less than an hour’s work each January ( and some occasional work each July ).

Of course I’m not totally lazy. It’s just that there’s a thousand other things to do that are more fun than studying company accounts, more fun than worrying and more fun than endless, mindless trading.

Don’t get me wrong… I love stocks, but experience has taught me that I can love them far too much. All the trouble I’ve ever had in the stock market has come from over-attachment to stocks, over-confidence in their outcome and a complete unwillingness to let a favoured hand go.

So some years ago, after another extremely painful experience I had a zen buddhist a-ha moment. I realised that in order to more consistently win, I needed to completely let go of my attachment to stocks… I went systematic.

What is the NAPS Method?

If you already know your NAPS from your SNAPS then this section is not essential reading. But for those who are new to all this, I recommend studying the following material and links.

Alternatively if you prefer not to read, you can join me for an hour-long webinar where I’ll explain the full NAPS Portfolio Process in person on Friday 21st July at 1pm. If you sign up today you’ll be guaranteed a seat.

The NAPS Method essentially describes a systematic process for creating and maintaining a personal portfolio of shares.

This process is based on a few fundamental assumptions:

We’re highly prone to a whole heap of behavioural flaws in the mind which tend to damage our investment returns. Using a rule-based, or checklist-driven approach counters this self-defeating behaviour most…

Unlock with Email

Unlock with Google

Already have an account?
Login here

stock

volatility

1/volatility

weight

MGNS

31.2

0.0321

0.047

WIZZ

35.3

0.0283

0.0416

HVN

35.3

0.0283

0.0416

GAW

28.3

0.0353

0.0518

BMY

22.3

0.0448

0.0658

PSN

35.3

0.0283

0.0416

MGR

0.0244

0.0358

MCL

42.2

0.0237

0.0348

LSL

33.5

0.0299

0.0438

CEY

48.1

0.0208

0.0305

IGR

41.1

0.0243

0.0357

SYNT

27.8

0.036

0.0528

MCLS

0.0303

0.0445

MCB

30.5

0.0328

0.0481

FIF

26.4

0.0379

0.0556

VTC

22.2

0.045

0.0661

CCC

25.1

0.0398

0.0585

SSE

21.1

0.0474

0.0695

GSK

19.9

0.0503

0.0737

UDG

23.9

0.0418

0.0614

stock

volatility

1/volatility

weights

MNDI

26.4

0.0378

0.0594

CAML

34.9

0.0286

0.045

EXI

47.6

0.021

0.033

PFC

38.5

0.026

0.0408

INDV

55.9

0.0179

0.0281

CTH

16.2

0.0616

0.0968

DRX

40.2

0.0249

0.0391

FCRM

48.8

0.0205

0.0322

WIZZ

36.3

0.0275

0.0433

MGNS

32.1

0.0311

0.0489

CCC

24.6

0.0407

0.0639

IQE

49.3

0.0203

0.0319

MCLS

38.4

0.0261

0.0409

FIF

21.4

0.0467

0.0734

QRT

24.4

0.0409

0.0643

GAW

0.0385

0.0604

III

0.0185

0.0291

REC

38.4

0.0261

0.041

MANX

20.1

0.0497

0.0781

TEP

31.2

0.032

0.0503

SHRIVGANESH

9th Jul '17

Hi Ed
This is Very Nice Presentation
I am one of those who over trades and overactive watching all the
news,bb, and read all the news papers every morning and company results
as they are released at 7 am.
After losing 60 % in Tesco shares by holding from 30 Jan 2013 to 4Jan2016
thinking that over trading is bad, good company, sit tight!!

Now i want to put a stop loss but i am not able to decide at what percentage you
cut the loss!!
As you say in this article that you could have cut loss then the Snaps would
have been much more impressive
At what percentage you would have put a Stop Loss ?
Can Stockopedia system help to put Trailing Stop Loss Parameter which would
be more beneficial for your subscribers as we are approaching down slope soon
Thanking You and your team.

Nick Ray

11th Jul '17

While the StockRanks have performed admirably across all sectors , I’m convinced we’re leaving returns on the table using a rather inflexible approach. So in this rebalancing I’m going to be experimenting with the rules.

I think you are leaving some returns on the table, but I'm not sure that the tweaks you are doing will make a big difference. At the moment you are getting a return which is basically similar to "sampling" all stocks with a Stock Rank > 90 or thereabouts. ("sampling" means effectively picking 20 out of the available matches at random.)

You can get a noticeable improvement by also filtering for Altman Z Score (2) > 4. (Also works for NAPS2016). Why? I'm not completely sure tbh but I think it might be because you are using an arithmetic "blend" to combine Quality and Value and it tends to choose some stocks which are high Value but not for a good reason. It's possible a rethink of how you blend Quality and Value would be worthwhile, or there may be other filters which help to exclude poor prospects too. (In other words the Altman filter is not a solution so much as a pointer to what might be going on.)

_________________________________________________________________________________

For the SNAPS 2017 you have chosen to include the Risk Ratings and likened it to the diversification you introduced based on sector and size. But I think this is a bit of a stretch of the meaning of the term "diversification".

Diversification is used to reduce "risk" (i.e. volatility) of a portfolio by choosing stocks with low covariance, so that when they combine in a portfolio the overall volatility reduces.

Sectors are a factor where this is definitely true, and Size is a factor where it is reasonably true. However the Risk Rating (Volatility) is a factor where it is clearly not true. Spreading your risk across high and low volatility stocks is not a technique to reduce volatility.

Volatility might be better judged in terms of the return of the stock and I think you would have more success with filters which limit the maximum volatility you will accept based on the previous year's return, e.g.: % Price Chg 1y > 2.5 * 1 yr Volatility % (choose your own multiplier to taste!)

Of course buying high risk-factor stocks is, by definition, risky. And risk must be managed. It’s a cliche - but diversification is the only free lunch in investing.

Talking of free lunches, I think regularly rebalancing an equal-weight portfolio is probably in the free lunch category, since it forces you to buy low and sell high almost painlessly! (In fact there is a branch of financial maths now which shows that equal weight portfolios do out-perform market-weight portfolios provided that a market meets certain conditions.)

cig

Equal weighting is no panacea: first, it gives you a major small cap bias (compared to cap-weighted benchmarks) which is likely to drown the volatility-harvesting effect. And then, that effect itself only works if it's done at a frequency which does not clash with momentum (rebalancing is a mean-reversal strategy, the opposite of momentum).

Maybe it's not a free lunch - more like a free hors d'oeuvres!

But for most of us, what I like about the equal-weighted portfolio is that it dissuades you from allowing a stock to grow to dominate your portfolio as you start to fall in love with it and then find it difficult to decide when to sell. If you rebalance regularly (maybe only every six or twelve months) you automatically take some profits without exiting from good positions. I think it is a good discipline for the small investor and very easy to implement.

There is some theory behind the claim that equal-weighting beats market-weighting, but as ever the maths only apply over "long enough" time periods.

Edward Croft

Spreading your risk across high and low volatility stocks is not a technique to reduce volatility.

I disagree with this. From memory, without this constraint the portfolio would have been composed of 14 Adventurous/Speculative shares and 6 Balanced/Conservative. I wanted a higher weight to lower-volatility shares, so i forced an equal split 10:10.

As we've seen in our research into the RiskRatings, lower volatility shares tend to see lower drawdowns (on average). So this constraint should help lower the downside volatility.

Of course, we really should be analysing the the impact of these constraints on the overall portfolio volatility - which relies on correlation analysis. I haven't done this as of yet for the NAPS. We'll be bringing in correlation analysis to the portfolio tools in future. I'm hoping to build an equal-risk portfolio weighting algorithm, rather than just equal-weight.

Brilliant Leader

"I'm hoping to build an equal-risk portfolio weighting algorithm, rather than just equal-weight."

That would be terrific Ed, I look forward to it.

12th Jul '17

Yes, it helps for risk management, though it may be a little bit overkill for that. You can get almost all the risk management benefits at a fraction of the hassle and cost with a weight cap rather than rebalancing to equal weight. My own rule is approximately to enter positions with the mean weight of the portfolio and trim anything that gets above 2x the mean weight back to mean. That keeps exposures in check with a satisfyingly low turnover.

14th Jul '17

If you fancy experimenting with equal-risk weightings, you can estimate a risk-parity weighting fairly crudely from the data that Stockopedia currently provides.

Basically you want to weight based on 1/(1y Volatility %). So for the new SNAPS 2017 portfolio you would get weights like this (where weight=0.05 is the equal weight point):

stock	volatility	1/volatility	weight
MGNS	31.2	0.0321	0.047
WIZZ	35.3	0.0283	0.0416
HVN	35.3	0.0283	0.0416
GAW	28.3	0.0353	0.0518
BMY	22.3	0.0448	0.0658
PSN	35.3	0.0283	0.0416
MGR	41	0.0244	0.0358
MCL	42.2	0.0237	0.0348
LSL	33.5	0.0299	0.0438
CEY	48.1	0.0208	0.0305
IGR	41.1	0.0243	0.0357
SYNT	27.8	0.036	0.0528
MCLS	33	0.0303	0.0445
MCB	30.5	0.0328	0.0481
FIF	26.4	0.0379	0.0556
VTC	22.2	0.045	0.0661
CCC	25.1	0.0398	0.0585
SSE	21.1	0.0474	0.0695
GSK	19.9	0.0503	0.0737
UDG	23.9	0.0418	0.0614

Does it work? Well, I tried it out on the NAPS 2017 stocks using my own calculation of volatility since Stockopedia did not publish volatilities back in January and the difference between equal-weight and the pseudo-risk-parity looks like this (equal weight in green and risk parity in blue):

596899528f5adequalweightriskparitynaps.p

The effect is quite small. Admittedly this ignores covariance between the stocks, but I doubt the result would be hugely different. I suspect that risk-parity weighting is more effective when you use asset classes with significantly different volatilities (such as stocks and bonds.)

For the record these are the volatilities that I used for the NAPS2017 plot:

stock	volatility	1/volatility	weights
MNDI	26.4	0.0378	0.0594
CAML	34.9	0.0286	0.045
EXI	47.6	0.021	0.033
PFC	38.5	0.026	0.0408
INDV	55.9	0.0179	0.0281
CTH	16.2	0.0616	0.0968
DRX	40.2	0.0249	0.0391
FCRM	48.8	0.0205	0.0322
WIZZ	36.3	0.0275	0.0433
MGNS	32.1	0.0311	0.0489
CCC	24.6	0.0407	0.0639
IQE	49.3	0.0203	0.0319
MCLS	38.4	0.0261	0.0409
FIF	21.4	0.0467	0.0734
QRT	24.4	0.0409	0.0643
GAW	26	0.0385	0.0604
III	54	0.0185	0.0291
REC	38.4	0.0261	0.041
MANX	20.1	0.0497	0.0781
TEP	31.2	0.032	0.0503

Cjenkins8

16th Jul '17

Great to see a worked example. Nice work. However, I don't think you can draw any conclusions from one sample like this. I believe it would have to repeated hundreds or thousands of times to come to any meaningful conclusion.

Conceptually, I think it makes sense to vary position size by some measure of volatility. Volatility as a proportion of price (for example, 20 day average vol / last close) can vary 3 fold easily between stocks so your position size would vary by 3 times too (compare IQE to SSE, for example). A 20% fall in one of these if their position size had been derived by equal weighting would be comparatively more unpleasant than if the position had been derived from volatility measurements.

But, you have to be very careful with low volatility stocks suddenly increasing in volatility. I don't have a good example of a UK stock to hand right now, but if you access to European stocks, look at CNWK which is very volatile now compared to last year. Also, a lot of smaller stocks, don't move for long periods and suddenly jump eg ALMER. This can play havoc with your volatility assessments. From memory stockopedia have their own proprietary tools which look at volatilty over years rather than just the last 20 days or so.

It is said that the only free lunch in investing is diversification. Another is volatility-weighting your position size. The new stockopedia risk ratings should prove very useful way to diversify your volatility within the portfolio, when used like Ed proposes!

17th Jul '17

You are quite right that you cannot conclude anything from a single worked example. There's quite a large body of work theoretically and empirically about the use of volatility in weighting portfolios for those interested in searching out more evidence for and against.

You definitely need to compute volatility over reasonably long periods, certainly at least a year, to apply these techniques to NAPS-like portfolios. Stocks which "don't move for long periods and suddenly jump" are quintessentially highly volatile.

There is definitely interest in new rule-based ways to construct portfolios these days. MPT seemed to settle the argument by "proving" that in an efficient market you could not out-perform the market-cap-weighted portfolio over the long term, which gave rise to passive tracker funds, as well as debates about how to find anomalies/inefficiencies in the market which allowed you to beat the market portfolio.

But more recently a very arcane branch of financial maths called "stochastic portfolio theory" has shown how to construct portfolios which (in theory) out-perform the market-cap-weighted portfolio, including the equal-weight portfolio and the risk-parity portfolio. A curious feature of these new methods is that they do not rely on trying to estimate the future mean returns of the stocks, but only their volatility and covariance (which tends to be more stable over time.)

Blissgull

SHRIVGANESH,

Where to place a stop loss?

You might find this paper useful.

http://lup.lub.lu.se/luur/download?func=downloadFile&recordOId=1474565&fileOId=2435595

In that study (of Swedish stocks) the researchers found that a stop loss level of 15%-20% led to the best returns.

The optimal level does seem to vary by market and by investing strategy.

18th Jul '17

Hi Blissgull
Thanks for your reply
This paper consists of 65 pages
I will go through it
Your summary gives good idea .
My Next Question once you cut loss @25%
If that stock starts to move back what you do
Do You Buy back when it goes up 10% or 15%.
I am sorry i am asking another question
'Cas HSBC puchased at 640p in Apr 2015 and sold
504p in Jan 2016,Now It is around 730p + !!!
It is very very difficult to get back in!!! Difficult Game !!!

19th Jul '17

Shrivganesh,
On the issue of where to place stops you might also like to look at this article.
http://www.stockdisciplines.com/best-stop-loss

Interestingly they seem to have identified a couple of important levels. By increasing the stop loss from 8% to 9% from a recent high (not the buying level) they found that the number of times you were stopped out prior to a rebound was reduced by 50%, with another big reduction between 14% and 15%. So those levels of 9% and 15% from a recent high may be good places for stops. Of course, that may not apply to the UK market.

On the matter of whether to buy back into a share you are stopped out of, personally, I adopt a very bad approach.

It is very irritating to sell a share only to see it turn round and go much higher straight away, so I avoid the problem by burying my head in the sand. If a stop is hit, I sell up and remove the share from my watch lists and monitors and try to avoid looking at it for the next few months. That way I don't know or care what its price does. Probably a very bad and perhaps immature thing to do, but from a psychological point of view I find it useful.

I think in one of Robbie Burns books he writes that he uses a "three strikes and out" approach. If he fails to make money on a share three times he will not touch it again. The idea behind that is to try to avoid "revenge trading" where an investor is determined to make money on a particular share to make up for past losses.

22nd Jul '17

As a retail investor with no background in finance, computing or statistics and a very busy day job, I tend to take a lot of what I read as said. So having read a fair number of investment books, I have tended to take as gospel the value of position sizing your share allocations according to volatility.
So while swimming in the Ionian sea, bizarrely, my mind turned your recent comment “…You are quite right that you cannot conclude anything from a single worked example. There's quite a large body of work theoretically and empirically about the use of volatility in weighting portfolios for those interested in searching out more evidence for and against…” and I thought why don’t I look at it myself.
So, admittedly after a fairly boozy lunch, I forewent the siesta and wrote a small script to test portfolio position sizing. And I have to say I was shocked. Because, after randomly constructing 1000 portfolios both equally weighted and volatility weighted, I found the mean returns were much better in the equally weighted portfolios.
But then I remembered Stockopedias pdf on Risk Rating where Ed showed that his low portfolio stocks did better from 2013, but from 2016, the highly speculative and speculative shares outperformed the conservative, balanced group of shares.
So I went back and looked at my data in 2016, 2015 and 2014 and the results very much tallied with Stockopedia’s!
Broadly, what I did was select UK shares with 4 years of data. I constructed random portfolios of 40 stocks with total exposure £1 million. So £25000 each in the equally weighted portfolios but varying positions in the volatility weighted. Volatility was assessed by looking at exponential moving average (EMA) standard deviation of previous years closes divided (normalised) by EMA (256 days) closing prices. I started with 1000 random portfolios but, being on holiday and risking the wrath of my wife, I decided to reduce this to 200 to speed things up (the computer was working overtime too).
These are the results:

It’s difficult to tell what’s going on from the line charts, but one gets an ideas of the spread of the returns in the 200 random portfolios. This confirms that just looking at the returns on one portfolio is meaningless. By grouping the returns into 20 buckets, one gets an idea of how postion-weighting influences returns.

We see much better grouping (less dispersion) in the volatility weighted group, but we are missing out on the long tails (positive skew) in 2016. It looks like volatility weighting worked in 2014 but was counterproductive in 2016.
What do the statistics show us? These are the mean, median etc 1 year returns of portfolio starting the year with £1 million exposure.

597317287a552Results_of_Equal_vs_Volatil

I think this shows that:

Volatility weighting is no panacea - much to my disappointment.
Don’t belive everthing you read. Certainly not this – as I could have messed up the code somewhere considering I’ve only got one years programming experience. But it’s helpful to test out ideas yourself.
Volatility weighting helps reduce the dispersion of your returns.
These results back of Ed Cohen’s and Stockopedia’s findings that risk-rating is important. Weighting for volatility helped significantly in 2014, didn’t make any difference in 2015 and was distinctly counter-productive in 2016. As Ed remarked,volatility risk needs to be used according to the current market environment – risk-on vs risk-off.

Now we just have to answer the question – how the hell do we safely determine the market environment?

Perhaps I will have to wait for next year’s holiday…

Thanks for taking the time to do this. But I concur, the missing link for me is how to correctly determine, early doors, whether the market is risk on or risk off.

If you're always short volatility over an investor's lifetime, you get exposed to all the episodes where it helps. The nice thing is it doesn't hurt that much when it doesn't help, so you should come out on top in the end, even without market timing.

Also it helps with attitude: if you prepare for downturns by limiting your volatility exposure, you're likely to be in a good state of mind to weather the storm instead of selling everything at the bottom and buying back when it's toppy again (which is the real killer for high vol investors).

The other thing about 'being short volatility' is that "most investors are interested in wealth compounding" and this is better captured by geometric means (compound annual returns) ... an asset's geometric return is always lower than its arithmetic return and the difference is larger for high-volatility assets." Ilmanen (Expected Returns). Over the long term, lowering the volatility of the portfolio is a way to improve overall return.

The study I presented showed that the range, spread, dispersion of returns is much narrower when position size for high volatility stocks is reduced.

Obviously volatility can also be reduced by picking lower volatility stocks in the first place!

ACounsell

26th Jul '17

I posted following article quite recently in response to Stockopedia article in April specifically on the subject of Stop Losses. It demonstrates the frustration that I feel about this subject!

"Since investing in SmallCap stocks on a regular basis in 2014 I have tried to use manual stop loss set at c. 20-25% given the inherent volatility of small cap stocks. This has been frustratingly unsuccessful given that many stocks haven fallen precipitously before I could hit the sell button. Over a period of 2-3 years 11 stocks (admittedly a small sample) sold on hitting the stop loss which sometimes massively exceeded the target 25% due to dramatic falls during a single trading session (Revolution Bars (LON:RBG), Plus500 (LON:PLUS) to name but two) resulted in significant lost 'profit' due to 8 out of the 11 stocks 'bouncing' back to levels significantly higher than the sold price (which was often closer to 35-50% down from cost or the 52 week high, whichever was higher, rather than the 20-25% I would have liked to operate). Admittedly if an automated stop loss of 20% had been in place and it was possible to execute a sale at that price (not always the case with a dramatic fall) then the notional loss would have been reduced by c. 65% which I guess is a lot better but still not a great advert for a strict stop loss policy!"

An addition to this comment relates to Tracsis (LON:TRCS) which fell by c. 30% following a trading update earlier in the year. This should have prompted a sale but I allowed emotion to take over and really couldn't see that the update merited the share price fall so I held on. On this occasion I seem to be right as the share price has recovered from 30% + down to 'only' 20% down. Had I sold this I would have missed out on some recovery. Another illustration of the frustrations associated with a rigid stop loss policy - or perhaps I am just unlucky!!

R Khan

29th Jul '17

Hi
Thanks very much putting all the list of the stock you thing right for SNAP for jully 2017

justinian

27th Mar '18

I have a potential further enhancement of a portfolio such as NAPS but is potentially more widely applicable across what are put across as 'themed" investments such as along 'Guru' lines. I don't know the statistics of my suggestion and would invite comment.

My main concern is overtrading and trading costs, but also that at times, stocks will move out of its narrow definition of selection (such as a QVM score) only to mean revert back again. You allude to this above and the need to 're-balance' the portfolio at the end of the year to replace stocks that have fallen out of your strict criteria for selection, but with the acknowledgement that you 'don't have an answer' to the problems this poses.

Looking at the the performance stats of the various Stockranks shows that several combinations of Quality, Value and Momentum exhibit outperformance be it by vigintile, decile or quintile. So that could be QVM, VM, QM and even to an extent QV. Does one really have to adhere to holding by just QVM for one's selection? One could alternatively choose to use ANY of the Q,V,M combinations and statistically, one may assume a similar chance of outperformance for the relevant score.

So, for instance, if my stock had fallen below a QVM of 75 say, I wouldn't necessarily chuck it out mechanistically but look also at the other score combos and if the stock fulfilled any score above 75 say, use that as an excuse to retain.

Basically, I'm saying that constructing a single portfolio based on a single theme (be it 'Guru' or score based) actually potentially adds its own layer of risk of underperformance and therefore diversifying not just by stock size, sector etc. one can also choose to diversify across themes which could also help reduce trading costs, reduce risk and volatiity if you apply the criterion that any stock can stay in if it meets any one relevant score.

Is there anything I've missed?

The NAPS & SNAPS Portfolio Update - 25% annualised returns to date and 20 new selections

What is the NAPS Method?

Unlock the rest of this article with a 14 day trial

About the Author

Edward Croft

38 comments

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

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