Via Twitter, from @theredcorner66:

@RichardBeddard Do you have a write-up on when/how you exit individual positions?

No, so let’s put that right. My deteriorating confidence in French Connection, and Games Workshop’s escalating market valuation are prompting me to think about selling, and putting my hitherto un-articulated guidelines into writing.

In theory, selling should be the inverse of buying and like all value investors I buy when I think the market undervalues a company by a significant margin. That implies I can at least in broad terms, estimate the true value of the company, a judgement that depends on what type of company it is.

If a company has performed erratically, and I have been attracted to it because of its extremely low price, I might buy it if its market value is significantly less than the value of its assets, preferably in a liquidation scenario.

Otherwise I will look to profits, often averaged over many years, adjusted in various ways, and divided by the price to give an earnings yield. Other people divide the earnings into the price to give a PE ratio. Both figures tell the same story.

The earnings yield is an estimate of the typical annual return an investor can expect if profits in the future are similar to profits in the past. With the exception of liquidations, I’m mostly interested in companies that will make more profit in the future either because they are turnarounds or I expect them to continue thriving. So, if I apply it to the right companies, the earnings yield will often underestimate my return.

There’s no right or wrong earnings yield, if you are confident that a company will earn much higher profits in the future it’s rational to pay more for it, and the earnings yield will be lower, but since growth is inherently speculative our baseline should be a company that is just plodding along earning much the same as it did in previous years, perhaps a bit more because of inflation.

I’d buy if the earnings yield was 10% or more (which, coincidentally is also a PE of 10). Shares are risky, less so if you own 30, but 10% is three times the best interest rate on cash savings, so I think it’s worth the risk, even though you don’t get all that return paid into your bank account annually. You might get a dividend, but the rest will come in capital gains years in the future, if you don’t sell before the market adjusts to your perception of value.

So here’s my guideline for buying:

• Buy if you are confident in the company’s asset value or earning power, and that it’s cheap

And here it is reversed, and split into two guidelines for selling

1. Sell if you’re confident you were wrong about the company’s asset value or earning power
2. Sell if you’re confident it’s expensive

Expensive would be somewhat more than asset value or an earnings yield of somewhat less than 10%. Note that to buy a company both conditions have to be true, to sell only one needs to be.

There’s another factor that comes into selling. I loathe it. The more often you sell, the higher the cost and it’s too easy to sell when things don’t initially go your way. My best ever trade was Inchcape, the motor vehicle dealership, which I bought in 2000. It halved in price, then returned to my buy price and then went up more than ten times by 2007.

Value investors buy when shares are cheap, but momentum is against them and they often get cheaper. My defense is pretty pathetic. It is to resist selling if I’ve owned the shares for less than three years. Knowing I’m going to do that also provides an incentive to get the buy decision right. Here are the guidelines reformulated:

1. Try to give a company three years to prove itself
2. Sell if you’re confident you were wrong about the company’s asset value or earning power
3. Sell if you’re confident it’s expensive

There’s one final complication, I want my money to be working for me in the stock market. I think it should be possible to find thirty undervalued companies even in a bull market so I’d rather not sell if I don’t have another company lined up to replace it. It’s another method of resisting the urge to sell. It’s also an incentive to develop new ideas.

Now there are four guidelines:

1. Try to give a company three years to prove itself
2. Resist selling a share until you have found a replacement for it
3. Sell if you’re confident you were wrong about the company’s asset value or earning power
4. Sell if you’re confident it’s expensive

These guidelines are deliberately fuzzy. It’s OK to sell inside three years, or if there is no replacement, if the case for selling is truly compelling. ‘Expensive’ is only vaguely defined. The guidelines have to accommodate many situations, but tomorrow I will show how I apply them by considering three companies that have troubled me this week: French Connection, Games Workshop, and Metalrax.