How many stocks should you own in your portfolio?

One of the problems with owning too much of some small caps is that it can be easier to get in than to get out.It is a bad idea to hold so many shares that one is unable to make a swift exit at an advantageous price when a stock becomes overvalued.So,then,if you have enough money to invest,diversification becomes a necessity.

Of course Ed, one solution to this problem is to have a concentrated portfolio of high-conviction stocks (likely small/micro-caps where the individual investor is not competing so much with institutional investors), and then to add to this a position in one or a few "smart beta" ETFs or investment trusts, which would then act to hopefully add a little alpha over the benchmark index while reducing portfolio volatility much closer to the overall benchmark...

Ed,

This ignores the cost of owns and trading 25+ shares vs 8, it also avoids the idea of actually balancing that portfolio in real terms. To cut down on transaction costs you would need around £25,000 to start without making a complete joke of brokerage fees etc. If you choose to ignore these elements that impact real investor returns than fine, by they direct impact any alpha generated over the market and add nothing back to the portfolio.

As said you are far better to have a balanced portfolio of high conviction companies and a passive or active fund to generate a safety net. Also investors shouldn't look to diversify for the sake of it on the back of returns from random selection. Just looking at firms with Stock ranks < 90 I would excluded all the Russian firms etc and likely improve returns on a portfolio.

Evoh - The thrust of this article is just to investigate breadth for breadth's sake - not costs, correlation, or diversification angles like sector, factor or geography.  The last paragraph though links to a previous article I wrote that discusses costs of trading and rebalancing. See here - http://www.stockopedia.com/content/when-does-it-pa...

re. "you are far better to have a balanced portfolio of high conviction companies" - I feel that this is the advice generally given by the investment community that can lead to the construction of poor portfolios.  Warren Buffett's line "diversification is a hedge for ignorance" plays on this theme - a comment that I've previously  labelled as extremely dangerous.

I think there must be thousands of investors reading Stockopedia who skewed their portfolios massively towards oil and gas during the 'supercycle / peak oil' meme that popped in 2011.   Similarly many had their portfolios skewed towards dotcoms in 2000 or China in 2008.   High conviction plays that ultimately led to enormous levels of herding amongst investors and subsequent underperformance.

 I repeatedly link toKumar and Goetzmann's terrific research on the subject (of 60,000 accounts) that shows most private investors only own a handful of stocks in highly correlated securities - they dramatically underperform as a result.  The smart response is to do the opposite and behave differently to the chest beating male stereotype.

Not sure how you find the time to write so much Ed but I wanted to say that I very much enjoy/appreciate/concur withyour recent work on this thread and the Psychopath thread.

Regards

T

T - thanks for the kind compliment. I don't really have the time, but I do find that writing is extremely productive in terms of organising one's thoughts - as I'm sure you do. We are rebuilding the portfolio analysis tools - I like to dive quite deep into a subject when building out features... so this is part of the process. Hopefully we'll get an ebook out of it too.

Regarding the 'psy' thread - I actually read up a lot on the subject at the weekend - I found this - the Levenson Self-Report Psychopathy test - http://personality-testing.info/tests/LSRP.php - I imagine a lot of outside thinkers (and good investors) will score much more highly on this scale than they might imagine.

Ed, Thanks for the very interesting article. You wrote: "In the following pair of tests we've created random portfolios of £10m+ sized stocks with a StockRank > 90. The first portfolios have only 8 stocks in them, the second have 25 - the difference in variability of outcomes is startling." I think it would be very interesting if you repeated this experiment, but this time with the rule that each random 8-stock portfolio must not have more than 1 stock from each sector of the stock market, and each random 25-stock portfolio must not have more than, say, 2 stocks from each sector. (A simple algorithm to enforce this rule for the 8-stock portfolio would be to reject any randomly chosen stock which is in a sector that already has a stock in the portfolio, and select another stock at random until one is chosen which is in a sector that is not already in the portfolio.) I suspect that this would significantly reduce the variability of outcomes, because stocks in the same sector tend to have high correlations between their share price movements. If I am right about this, then diversifying one's portfolio between stock market sectors would enable one to construct low-variability portfolios with a smaller number of stocks than if one does not diversify in this way.

This is all very useful stuff but I would suggest that timescales play an important role. Diversification is very important as you apprioach retirement. However, diversification is much less important for say, a 30 year old aiming to retire at 65,. The difference in volatility in returns shown above when investing in 8 versus 25 stocks will reduce over a longer time frame. If you were invested in 8 stocks over a 20 years with an average holding period of 1 year then you would have made a total of 160 investments. This should be more than sufficient exposure to ride out volatility on overall returns. If you were instead invested in 25 stocks instead of 8, you would very likely end up with the same amount of capital. The only real advantage is that you would have had a smoother ride.

Excellent work Ed, I always thought the 8-20 stocks recommendation for reducing volatility was hogwash. As you say it's not risk measured as volatility that investors want to reduce, but the risk of performing badly.

Also I think the risk of something bad happening is critical, both in terms of actual losses (i.e. what if 2 out of your 10 holdings go bust, compared to 2 out of 25 or 30) and in terms of the behavioural response (i.e. 2 out of 10 holdings going bust would quite possibly lead the investor to sell, probably at a market low, and seek eternal safety and poor returns in cash or similar, while 2 failing out of 25 probably wouldn't get the same knee-jerk reaction).

As for the problem of small portfolios, I don't think it's even worth buying individual shares until you have north of £20K and therefore the ability to put £1K into 20 stocks. Until that point just get your head down and save, save, save, sticking it into a tracker or similar just to get used to the ups and downs.

Diversification is much less important for say, a 30 year old aiming to retire at 65.

I'm not sure if I agree with you actually due to the very different risk of ruin in each case. There's a far higher 'risk of ruin' in high concentration / higher volatility strategies - the average performance may be the same, but the distribution of returns is vastly different.

If an investor gets unlucky he might catch tails in the performance coin toss a few years consecutively. The probability of a disaster outcome for the concentrated investor is far higher than that for the diversified investor.

In spite of the above and the clear benefits of diversification, I've always said that younger investors with small portfolios should get started with a concentrated strategy early - the real cost of loss is normally minimal (they normally have less to invest and can recover from small capital losses over many working years) - but the upside through real self-education is massive.   As soon as capital grows sufficiently though, investors should take advantage of the clear benefits of diversification. 

So I disagree statistically but agree with you from a practical perspective !

I've spent the weekend trying to prove the point I made earlier in this thread only to prove myself wrong!

To recap, I suggested that it was less important to be diversified early in ones investing career because over a long run there will be time to recover from a early run of bad luck.

I backtested this hypothesis using data from Sharelock Holmes from March 2000 to March 2014. I selected random portfolios from stocks with low "Market" scores (equivalent to StockRanks > 90). Each portfolio is held  for a year and then replaced by a new random set  of stocks. I ran two tests, one for an 8-stock portfolio and another with 25 stocks. I then repeated this 30 times for each to get a distribution of returns, similar to Ed's tests earlier in this thread. The results are plotted below.

This clearly shows that a longer timeframe does not reduce the volatility of possible returns for an non-diversified portfolio. If you have average luck, then the returns for the 8 stock and 25 stock portfolios will be pretty much the same. However, crucially, it you are very unlucky, your returns for an 8 stock portfolio could be half that of an 25-stock portfolio.

The reason I was wrong was that I didn't take account of the effects of compounding returns. A run of bad luck early in an investment career, will have a profound effect on long term returns as you will have forever lost the opportunity for those returns to be compounded. So even if you have a good strategy for selecting stocks, you may up with returns well below par if you are unlucky and are not sufficiently well diversified

MB - yes your analysis is similar to that in the paper which I referenced. The important point is that an investor should choose the number of stocks in his/her portfolio with reference to their comfort level at missing / beating the average.

There's a vast range of portfolio performances when chosen randomly - you can either end up at the top of the range or the bottom depending on either skill or luck. The more stocks you own the more likely you are to end up getting the average return - which of course one hopes will be a market beating return if the criteria used are well constructed.

The chart below is my mockup of the chart in the paper. The horizontal line at 100% represents the average return, the high and low lines represent the range of outcomes for portfolios with different numbers of stocks in the portfolio.

Clearly one can see that owning more stocks leads to higher confidence that one achieves somewhere in the vicinity of the average returns.  The risk in concentrated portfolios is that you could wildly underperform.

Of course some stock pickers will think they can invest in concentrated portfolios with skill and put themselves consistently in the area above the 100% line, but my experience is that investors often aren't as good as they think they are - we are all geniuses in a bull market after all - diversifying more widely and trusting in compounding is a better bet.

Too much diversification can lead to difficulties in keeping track of how the companies are doing.Good companies can easily go bad,especially where there is a change of management.

It's interesting that John Lee has about 35 stocks in his portfolio.

I hold around 35 different stocks.  There is no particular magic about 35, although curiously many private investors seem to find this a comfortable number of holdings to have:  keeping track of significantly more holdings can be quite demanding.

With varying position sizes.

My stocks significantly vary in value, probably by a factor of up to 10.

And on losses...

One of my cardinal principles has been to focus on avoiding losses rather than chasing profits.  In golf it is the shot in the river or wood that destroys the round; similarly with investing it is the portfolio losses that drag down an overall performance.

I have found this article and the comments following it interesting and useful.

They have led me to conduct the following experiment with real money.

I have today set up a 20-stock portfolio with £5,000 in each stock, making a total portfolio value of £100k. My online stockbroker charges me a fixed commission of £10 for each purchase or sale, which represents just 0.2% on a purchase of £5,000. Assuming an average bid-offer spread of 0.1%, and Stamp Duty of 0.5%, a round-trip transaction is likely to cost about 1% of the initial purchase value (= 0.2% + 0.5% + 0.1% + 0.2%), which is not too bad.

I selected the stocks from a Stockopedia screen that I set up, which has the following very simple rules:

(1) Each stock must belong to either the FTSE 100 or FTSE 250 index, i.e. it must be in the UK FTSE 350 index which tends to contain the largest 350 companies by market capitalisation in the UK stock market. I've restricted the portfolio to the UK stock market because that is the one whose constituents I am most familiar with and feel most comfortable owning directly (rather than via a fund).

(2) Each stock in the screen must have a Stockopedia StockRank that is greater than or equal to 90, i.e. it should be in the top decile of companies when measured by StockRank.

The screen then ranks the stocks in order of decreasing Market Capitalisation.

This screen produced a list of 43 stocks today. I then selected stocks from this list in order of decreasing Market Capitalisation, such that there was no more than one company in the portfolio in a particular 'Industry Group' (as defined by Stockopedia).

My reasons for selecting stocks from the list in order of decreasing Market Capitalisation and restricting the selection to the FTSE 350 are:

(a) This strategy tends to reduce transaction costs, because big companies tend to have smaller bid-offer spreads than small ones.

(b) This strategy tends to reduce risk, because big companies tend to be less risky than smaller ones.

(c) This strategy tends to make the stocks selected for the portfolio on different days more stable, because relative Market Capitalisations are likely to change less rapidly than StockRanks.

My reason for selecting no more than one company for the portfolio in a particular 'Industry Group', as defined by Stockopedia, was to provide a crude but easy-to-implement method of diversifying the portfolio. I think there are about 40 'Industry Groups' in the companies in the FTSE 350, so selecting 20 stocks from those was not over-restrictive. I would expect companies in the same industry group to have share price movements that are likely to be more correlated with each other than they are with companies in other industry groups.

I have not yet decided on my selling rules. One option is to review the portfolio every few months and replace any companies whose StockRanks have fallen significantly below 90, which means they are either no longer such good value or quality or their share price momentum has deteriorated. For example, if a company in the portfolio has a StockRank of 70, I could sell it and reinvest the proceeds in a company that is then in my screen, so the entering company would have a StockRank of at least 90, and the difference in StockRanks between the entering and leaving company would be at least 20. I should ensure that the new stock selected is the first in the list, in order of decreasing Market Capitalisation, in an Industry Group that is not already in the portfolio. I think a difference of at least 20 in the StockRanks of the companies sold and bought would probably be enough on average to justify the transaction costs of trading.

Alternatively, I could keep the portfolio unchanged for a year, and then rebalance the whole portfolio once a year, reinvesting any dividends that were received over the previous year. This policy would have lower transaction costs than trading every few months, but its performance possibly might not be as good.

It will be interesting to see how this portfolio performs in the future!

I would welcome any comments people may have on my selection strategy for this portfolio.

I hold between 20 and 30 stocks in the JIC Portfolio and am currently at the top end with 29. My largest 10 holdings currently comprise 46% of the Portfolio by value.

I have just had a look at Neil Woodford's portfolio and see that he currently holds 66 stocks. (I'm not proud! Happy to get new ideas to investigate from anywhere.) However, I think the important point is that the the top 10 holdings comprise very nearly 50% of the portfolio; the other 56 holdings make up the other 50%. The 26 smallest holdings comprise 10% of the fund by value. I can only guess that this a sort of "suck it and see" approach given that many of the names are at the more "risky" end of the spectrum. Some will presumably turn out to be duds but will do minimal damage to the portfolio whereas others may turn out to be ten baggers. Also I guess that as some of the more "speculative" names start to prove themselves he will add to them and build up the size of the holding.

In conclusion just looking at the number of stocks alone is too simple; the concentration in the largest holdings is what is likely to drive performance in the short to medium term and having a long tail might not be a bad idea as long as you are disciplined in what stocks you buy and are not just punting around.

https://www.investopedia.com/articles/stocks/11/illusion-of-diversification.asp