Most investors use some sort of benchmark to track their performance against. I’ve been thinking a little about benchmarking and have come up with some improvements for how I benchmark my performance.

What is the point of benchmarking?

The basic idea behind benchmarking, of course, is to have something to compare your portfolio performance to in order to get an idea of how well you are doing. As a default, many investors, including me, will use a well known index such as the FTSE100 as a benchmark. This is a useful benchmark as it directly tells you whether you are ‘beating the market’ – or how you are doing relative to the average investor.

I feel I should make a big caveat here, which is that there are huge limits to what any sort of benchmarking can tell you. The problem is basically that the timescales in investing are too long to generate sufficient data for it to have much statistical reliability. By the time that you can be statistically confident that what you are doing is working rather than is just being lucky, or alternatively requires changing rather than is being unlucky, it’s probably too late. So benchmarking (until you’ve done it for ages) is really just sticking a thumb in the air – it might be useful but you should really take it with a massive pinch of salt.

I think indices like the FTSE100 are perfectly adequate as benchmarks. However, I also think you can get a bit more out of benchmarking, by using benchmarks that more closely represent alternative approaches to your current strategy. For example, I’ve come across several debates over whether it is better to use a market-cap weighted index (like most indices e.g. the FTSE100) as a benchmark or an equal-weighted index (which gives equal weight to each share regardless of market cap). Proponents of the latter suggest that an equal-weighted index would more closely resemble an alternative portfolio the investor might choose and that a market-cap weighted index is overly influenced by the largest businesses. I don’t have a strong view of whether one is better than another. I think both can be useful as benchmarks but they tell you slightly different things:

• Market cap weighted index: how am I doing relative to the ‘average’ investor in the market?
• Equal weighted index: how am I doing relative to a strategy of investing in a random selection of shares in the market? 

While a market cap weighted index is useful for telling you how you are performing against others, an equal weighted index can be more useful to you if it more closely resembles how you select your stocks i.e. if like most investors you do not weight your holdings according to market cap. It can therefore tell you a bit more directly about how well your strategy is working i.e. if you do better than picking stocks at random.

I don’t actually care very much about this particular distinction.  I highlight it solely to illustrate the point that benchmarks can be more useful if they more closely capture alternative strategies you might follow.  The equal-weighted benchmark has you competing against a monkey with a pin – interesting to know, but I think this is a pretty low bar and have set myself some more challenging benchmarks.

My benchmarks

I have chosen benchmarks to capture alternative approaches I might follow, but also to isolate different aspects of my investing strategy. This might hopefully give me some clues as to which aspects are working and which are not.

I’ve chosen the following:

• FTSE100: I use the FTSE100 as a benchmark to give me an idea of how I’m doing relative to the average investor and whether I’m ‘beating the market’.
• Stockranks: If I didn’t employ my own strategy I think I’d be most likely to do something based on Stockopedia’s Stockranks as there is empirical evidence supporting their outperformance of the market. I’ve therefore used a basket of the top decile Stockranks as a rather challenging benchmark. Stockopedia tracks their performance here. If I didn’t outperform this over time then I’d be better off shutting up shop and using the Stockranks instead.
• Buy and hold: This benchmark is based on buying all the shares in my portfolio and watchlist in equal weights and holding them . The idea is to separate the aspect of my strategy that is based on identifying the highest quality shares, from my actual decisions of when to buy and sell. By comparing this benchmark to the FTSE100 and Stockranks I can get an idea of how effective I am at identifying high quality shares (that should outperform in the long term). By comparing this benchmark to my actual portfolio I should get an idea of how much value my individual buy and sell decisions, based on momentum and valuation as well as quality are adding. This porfolio is simply ‘buy and hold’ – it will only change when I review and change the watchlist.
• Mechanical: This benchmark replaces my actual buy and sell decisions with entirely algorithmic decisions based on the scoring in the watchlist spreadsheet. As my actual buy and sell decisions are based on the watchlist, I would expect the performance of this benchmark to be fairly similar to my actual portfolio. The key differences are that: a) this benchmark will (initially) be equally weighted across 25 positions rather than weighting according to judgment; b) my approach to selling in this benchmark portfolio will be different – I will update the benchmark portfolio monthly, dropping the lowest scoring share from the portfolio and replacing with the highest scoring from the watchlist, rather than following my selling principles (you can see in my investing strategy) which require more judgment; c) I will always choose to buy the highest scoring share on the watchlist rather than also using my judgment. Broadly the idea is to get an idea of whether allowing some judgment when trading is positively contributing to performance or whether a very similar but even more mechanical approach might do better.
• Mechanical concentrated: This benchmark is the same as the previous but concentrated into only the 10 best scoring shares from the portfolio and watchlist. I want to get an idea of whether and to what extent a more concentrated portfolio might improve my performance. My suspicion (and hope) is that this one might do really well.

The first two benchmarks already exist and I already report my performance against them. I have just now set up dummy portfolios to model the last three and will be able to report performance against them in the future.