tjh290633 wrote:OLTB wrote:Wow - thanks TJH, I'm not too sure I fully understand what unitising a set of numbers does, but from what I can see, it means that the historical dividend income has increased by 736.53% over the 29 year history (an average annual increase of 25.40% - if you can average a unitised increase??) compared to an RPI rebase of 156.78% (5.40% average over the same period). As your chart shows, there have been some pretty wild swings from one year to the next, but holding firm has resulted in a superb outperformance against RPI.
If correct - and I'm hopeful that someone will correct me if I'm not as I'm here to learn - that is a compelling argument.
You need to do the maths differently. The divis per unit have multiplied by a factor of 8.3653 over the 29 years. Rather than an average you need to find the rate at which they have compounded, which is the 29th root of 8.3653 minus one, ie 7.60%.
Meanwhile RPI has multiplied by 2.5678, leading us by the same method to a compounding rate of 3.31%.
So the divis per unit, in real terms, have compounded at (1.0760)/(1.0331)-1 which is 4.16%.
Bear in mind too that the actual income will depend on how the number of units has progressed, so will have increased by a larger factor as money was added and not all income taken.
Unitisation, as TJH explained, is straightforward. Few people, having taken the plunge, stop doing it. It is particularly useful at the building stage imho, and can be started at any time.
Its purpose is to remove the distorting effects of cash being added and withdrawn from the portfolio. Without such a method, an investor will look at the divi income increasing but gradual cash movements will mask how underlying dividends are actually progressing.
PS I too assemble key data in monthly buckets( I find that it makes graphs neater over long periods), including unitisation, doing it automatically by recording all portfolio activity ONCE as it happens in tables rather than a plain spreadsheet, and using pivots extensively to link to other tables and compute subsequent measures, thus creating a database-like tool. It is not complicated and only needs to be set up once and is very versatile, for example the full history of any holding can be extracted, including performances measures, without storing it separately, which would often lead to typing in the same data twice, a big no no for me. I cannot recommend this method strongly enough.