How to know when to sell shares

Do you stick or twist? On the back of a big winner, there is a temptation to take your chips off the table. Investing isn't gambling, however, and as celebrated fund manager Peter Lynch once said: "Selling your winners and holding losers is like cutting the flowers and watering the weeds." To apply this thinking to today's markets, it doesn't make sense not to back growth in US tech simply because the S&P 500 IT and communications sectors are each up around 50 per cent in the past year.

Still, there comes a time when selling is the right call, and surveys of Investors’ Chronicle readers repeatedly show people agonise over when to make this decision. That's no surprise, as solving this conundrum is incredibly difficult. To begin with, investing is personal: objectives, timeframes and risk tolerance vary enormously. On top of this, many humans are afflicted by the cognitive biases and emotional impulses that, unchecked, lead to making bad portfolio decisions.

But there are checks that can be put in place. To borrow a classically inspired analogy from Greg Davies of consultancy Oxford Risk, there is something to be gleaned from the “ropes and beeswax” tactic employed by Greek hero Odysseus. Wanting to hear the sirens’ song but anticipating it would seduce him and lead his ship’s crew to crash on the rocks, Odysseus had them tie him to the ship’s mast and block their ears with beeswax. Thus, he was able to indulge his curiosity yet safely navigate the peril.

This is a parallel with rules-based investing: creating a framework for making choices is one of the best ways for investors to avoid succumbing to their worst instincts. And when it comes to selling, the starting point in mitigating poor choices comes before you even buy a share.

Your initial plan

Buying for the wrong reasons puts you on the back foot right from the off. Jumping on a bandwagon due to fear of missing out (FOMO) is in fact a manifestation of several psychological flaws. In doing so, investors may fall prey to cognitive dissonances such as framing risk and rewards too narrowly, or making mental short-cuts based on limited information. Emotionally, it suggests overconfidence and weak self-control, which leave the door wide open to other unhelpful traits that could make that impulse buy a problem holding.

So when starting out, you should have a clear investment goal and ask whether, in the grand scheme of things, owning a particular share is going to advance that goal without making your overall portfolio riskier than necessary.

On a previous occasion when the IC confronted the selling dilemma, we posited that there are only three good reasons to sell a listed security: the original assessment was wrong; there was an unforeseeable circumstance or misfortune; an irresistible opportunity arose that was a better use of capital.

Drawing up an investment policy statement ahead of buying stocks is a good way of managing your approach to reasons one and three. In practice, as we noted at the time, making and re-evaluating investment cases isn’t straightforward. With that in mind, if your objectives are achievable by investing in steady, quality compounders – a strategy whose success depends less on understanding the complexity of those businesses – why invite drama by going after deep value turnaround stories or speculative growth plays that demand a more studious approach?

Refer back to your investment thesis

As all this implies, it is not just about the 'when'. Another thrust of our 2015 missive was that the 'how' matters at least as much as the 'why'. This brings us back to rules-based decision-making. However, as Philip Ryland entertainingly wrote at the time, many rules can be asinine in themselves. Arbitrary rules of thumb, such as selling once a price drops 10 per cent, or damaging mantras such as “any sale that results in a gain is good", are too broad to be useful to us here.

To return to the 'how': referring back to an original assessment won’t guarantee that you correct mistakes before the market punishes you, but nor will it hurt your chances of doing so. Being prepared to question yourself and reset is a great way to counter loss aversion – the phenomenon whereby people hate losing more than they like winning, which paradoxically can lead to holding onto bad positions to avoid crystallising the loss and the associated psychological pain. Delays can be damaging, and you shouldn't be too proud to admit when you call something incorrectly. 

Professional analysts, flying in the face of caricatures, can show the way to exercise humility in this regard. Goldman Sachs' rationale for a recent downgrade of its view of luxury clothing designer Burberry (BRBY) is an example of how to back away from a prior stance. 

Having previously argued that new products would drive like-for-like sales growth and margin expansion, Goldman revised its view after Burberry's 12 January trading update and expressed scepticism over whether the £4bn revenue ambition could be achieved without greater investment. Large end markets such as China are having a tougher time than thought and, in the west, anecdotally Burberry is seen as a second-tier luxury brand, exposed to the pinch on middle-class households' finances. 

At the time of this u-turn, the shares were down 34 per cent from where Goldman first rated Burberry as a buy, so the decision could have come sooner. Importantly, however, the move means no clinging on waiting for a turnaround.

Based on a 15 per cent downward revision to underlying earnings and free cash flow forecasts and an expected increase in Burberry's cost of capital, Goldman drastically reduced its 12-month price target. This is selling based on cold, hard numbers that predict further headwinds. Making good on a disappointment like this can be best done by finding something better to invest in. 

To guard against the endowment effect (only focussing on what you own and therefore missing the big picture) it’s worthwhile applying an evaluation rubric to a wider universe than just your own portfolio. You could, for example, employ a stock screen to help find shares that might do a better job than those you currently own. In other words, stay abreast of opportunity costs.

Would I buy this today?

Keeping this strategy front of mind, investors can build and manage a portfolio of risk-appropriate stocks. The checklist for holding a share should involve the same questions asked when it was first bought, encompassing macro, industry, business, operational, financial, internal investment and execution risks.

Take Rolls-Royce (RR.), for example. The shares trebled in value last year on the basis of the post-Covid recovery in aerospace and enthusiasm for the chief executive’s efficiency and productivity drive. At the same time, as we noted in December, the company has considerable contract liabilities because it services its engines under long-term agreements, which require significant working capital. This doesn’t mean it can’t continue to grow earnings, but arguably circumstances don’t support a repeat of the shares’ spectacular 2023 performance.

This on its own isn’t reason enough to sell, however. The sensible approach is to resume a neutral outlook and reset your thinking: ask how you’d make the buy case to a man from Mars. In the case of Rolls-Royce, data from FactSet shows 68 per cent of analysts surveyed still have the shares on a buy, and the mean price target implies a further capital gain of almost 10 per cent. The estimated uptick in earnings per share between the unreported 2023 financial year (results will be announced on 22 February) and 2024 is above 30 per cent. If there aren’t grounds for disappointment against previous guidance, keeping the holding could well make sense. But as we’ll discuss, some portfolio maintenance might still be required in such cases to ensure the position is kept at an appropriate weight.

Momentum and valuation

On top of fundamental analysis, there are other factors that are quantifiable but hard to interpret, namely momentum and valuation. Obsessing over either is a sure way of succumbing to cognitive and emotional bias. Equally, mechanically making entries and exits based on momentum or value signals can be self-defeating.

When asking “do I still want to own this stock?”, you must put yourself in the here and now, not anchor beliefs to what occurred in the past. Valuations may increase, but that’s because the market is more confident that previous expectations will be met, or because there is an improvement in the sober fundamentals.

Animal spirits alone are dangerous, but if they are underpinned by something more substantial, being “fearful when others are greedy” could lead to missing a large chunk of upside. Far be it for us to second-guess musings by the world’s greatest investor, but one suspects Warren Buffett’s observation was prompted by investors in the grips of FOMO rather than those riding bonafide upgrade cycles.

With this in mind, our own momentum stock screens also monitor the delta in earnings growth and, crucially, the rate at which analysts have altered their expectations for companies’ profits in given accounting periods. Of course, there are things to be mindful of here, too – analysts are only human and can also get caught up in mania for certain stocks and themes, plus there is career risk in deviating from a powerful consensus.

The long-run performance of equities would suggest optimism has been rewarded, so swimming against a strong tide isn’t advisable as a rule. When, as we have seen with artificial intelligence (AI) chipmaker Nvidia (US:NVDA), the pace of upgrades is so rapid that the share price to forward earnings per share (EPS) estimates ratio (the price/earnings or PE ratio) doesn’t expand, it’s daft to bail out at a cut-off point that was set based on old information at the time of purchase.

There are other even more growth-centric metrics with which investors can reassure themselves they aren’t holding onto a company that’s stuttering at the end of a good run. Dividing the PE by the relevant period’s forecast rate of earnings growth gives the price/earnings growth (PEG) ratio, and by this measure Nvidia still looks reasonable.

These expectations are based on companies adopting AI to solve real-world problems, and the need to build out enormous cloud-based infrastructure to facilitate this. But stocks without the tailwinds of transformative technology can be judged on a variant of PEG that’s more suitable for mature business models.

For companies that aren’t in highly cyclical industries, using a so-called genuine value (GV) ratio, which divides the PE by a combination of recent and expected earnings growth rates (plus, where relevant, the dividend yield), which assesses valuation in the context of smoothed-out total returns. This isn’t appropriate for businesses with intermittent profit windfalls (such as housebuilders, energy or mining firms) but is a good way to judge defensive stocks and quality compounders in industries where growth is solid but not spectacular.

These metrics can all just as easily apply when considering a company whose share price has fallen materially – albeit in those cases, it's worth noting that earnings estimates sometimes lag on the downside (analysts can be slow to revise their numbers, at least in sum), which means shares that still look good value may prove nothing of the sort.

Yet because price targets and earnings forecasts are moveable feasts, if a company is growing well there is no reason to leave the table early, even when the PE multiple alters the dynamic. It’s also worth remembering that equity valuations must be judged in the context of other assets.

When interest rates are high, and investors can get an attractive yield from government bonds, the amount they are prepared to pay for riskier shares falls, meaning PEs must contract. We may note that a company looks cheap against its five-year average PE, but we should keep in mind that central banks have hiked rates aggressively over the second half of this period: the risk-free rate was very different back in 2019. But, by the same token, when the long-awaited cuts to base rates arrive, investors shouldn’t mistake expanding multiples as a sign they have become overbought.

Perspective in this matter can be gleaned from changes in the difference, or ‘premium’, shares offer versus less risky assets. One method is to look at the spread between earnings yield – forward EPS estimates divided by price (the inverse of the PE) – and the gross redemption yield on a benchmark government bond. All other things being equal, share prices will fall to maintain the spread when rates rise, so if this spread remains tighter than historical levels, it’s a policy-neutral indicator of expensiveness.

For companies outside financial industries, it’s possible to improve on earnings yield and thereby make allowances for different capital structures. Adding the market capitalisation of equity to a firm’s debt and subtracting cash provides a company's enterprise value (EV). The multiple of this figure to earnings before interest, tax, depreciation and amortisation (Ebitda) shows how a business’s cash generation compares with the value of claims on it by share- and bondholders. 

Dividing one by EV/Ebitda gives the more comprehensive enterprise earnings yield (EEY). The same spread analysis described above can then be applied to changes in this metric. But again, context is everything: for an indebted firm in a period of higher interest rates, one ought to expect the market to require a greater risk premium. That means a spread that is wider than historical levels is simply a sign to investigate further, rather than a buy signal.

Nor do tighter EEY to bond yield spreads scream 'sell'. Interest rates are higher than they have been for many years, but the next direction of travel is likely to be down, or at the very least no further up. With equities edging towards the time when this point arrives and the market re-rates, it's not surprising shares in the best companies twitch for the starting gun.

In summary, valuation alone isn’t reason to completely exit holdings, but it can inform decisions on rebalancing if a long-running winner has, due to past success, become heavily overweight in a portfolio. The decision to trim should be based on the forward expected rate of return from that stock, others you own and alternatives you don’t. Crucially, the overall risk/reward make-up of the portfolio, and how this matches objectives, must remain front and centre.

Diversification and position management

Postmodern portfolio theory can help guide the decision process. Methods such as the Black-Litterman model (developed by Goldman Sachs economists in 1990), which we have discussed previously ('How to profit from the energy switch', IC 24 Nov 2023), enable investors to incorporate their beliefs on the outlook for a stock. In this way, unlike basic modern portfolio theory (MPT), Black-Litterman doesn’t rely only on past data.

Having re-evaluated the investment theses for stocks in your portfolio, Black-Litterman can be used to factor in past volatility and correlations to demonstrate optimal portfolio weights based on the rates of return you expect from the current point in time. This can be sense-checked using the expected returns suggested by another metric, such as the capital asset pricing model (CAPM). Logically, it follows that you should prune holdings if they have grown to be significantly out of kilter with your desired weightings.

Our chart (below) shows the percentage weights you would assign six stocks based on their market capitalisation (as of 7 February 2024), versus what CAPM would suggest is a good risk-reward trade-off given its prediction of future returns. The weightings were calculated using the Black-Litterman framework; they suggest CAPM devotees would get far more utility from a more modest exposure to pharma behemoth AstraZeneca (AZN) and a higher allocation to Spirax-Sarco Engineering (SPX). The CAPM weightings represent a hypothetical, highly concentrated portfolio of just six stocks; in reality an investor's portfolio would be spread much more widely. The S&P 500, for instance, is nowadays deemed to be top-heavy because its top three shares account for 18 per cent of the index – albeit worrying about how to balance 500 different positions is not a common problem for private investors.

As few investors are perturbed by volatility to the upside, another method is to consider returns versus downside risk alone. Subtracting your return target from your actual return and dividing by downside volatility creates a figure known as the Sortino ratio. The higher this number, the better.

For those so inclined, it's possible to use MS Excel Solver to optimise portfolio weightings for a maximum Sortino ratio. When considering downside risk, past data would be required, but you could also input the predicted upside for each stock as outlined above. Again, the results may suggest the portfolio should look different than it does, letting you know it’s time to modify.

Interestingly, Sortino-optimised portfolios can often suggest concentrated holdings, which is a problem if you get your investment thesis wrong. The other enormous caveat that follows on from this is that they can leave you very vulnerable if the second 'why' to sell reason occurs – some terrible misfortune torpedoes the otherwise solid investment case.

Returning to the example of Nvidia, we can easily imagine what such an event might be, given 90 per cent of the world’s most advanced semiconductors are manufactured in Taiwan, the disputed island long coveted by China. An invasion would be a disaster, but does not meet the Black Swan definition of being unforeseeable, so supply chain disruption should form part of investors’ thinking.

For stocks like this, potential rewards justify what ought to be a quantifiable and, over time (if the US succeeds in developing its own semiconductor industry) potentially diminishing risk. Selling because of it would be akin to investing agoraphobia.

All the same, it is a reminder not to abandon the principle of diversification and the knowledge that hypothetical models should be used as a guide rather than followed to the exclusion of all else. But models at least give you strong reassurance not to sell when you shouldn’t.