stacker512 wrote:So the thread
https://lemonfool.co.uk/viewtopic.php?f=15&t=28540 says:
For each time that the portfolio received a dividend, the date, the cash value of the dividend and the value of the portfolio just before the dividend appeared.
How are you supposed to know the portfolio value just before the dividend arrived if the portfolio is made of varying price investments that change every minute / second, and you have no idea what the precise time of the dividend received will be?
Seems one would always have slightly incorrect values to use for the unitisation.
Indeed -but even if you somehow managed to get yourself totally accurate values to use as inputs to the unitisation calculations, the results of those calculations would be a bit out of date by the time you were able to use them for any practical decision-making... Basically,
all real-world data is subject to inaccuracies - complete accuracy is only possible for mental constructs such as mathematical number systems and other mathematical theories, and even for them, the practical problem of accuracy is replaced by the practical problem of determining how well they match the real world (*).
The practical answer to that inaccuracy problem is not to ask "Is this totally accurate?" but "Is this sufficiently accurate for my purposes?". And in practice, the individual percentage inaccuracies in a portfolio unitisation calculation produced by share price fluctuations shortly before an individual dividend payment are generally pretty small, and their effects across the large number of dividend payments a HYP produces will be roughly equally in both directions and so average out to even smaller percentage inaccuracies, at least if you're doing the unitisation calculation over a long enough time period (at least a few years) to match the long-term nature of HYPs.
In effect, that means that in "just before the dividend appeared", "just before" can be interpreted quite loosely. Just how loosely one can do it and still get sufficiently accurate results for one's own purposes is a matter for each HYPer to decide for themselves, but I'm personally entirely happy that valuing the portfolio at closing prices on the last trading day before the dividend payment date is sufficiently accurate for my purposes. And I'm pretty happy that the same can be said of the technique (mentioned both in the post you link to and by some in this thread) of treating all dividends paid over a week, a month or possibly even a quarter as a single payment made halfway through the period will still produce pretty accurate results.
On the issue (raised in this thread) of whether to record dividend payments on payment dates or ex-dividend dates, there is indeed an argument for recording them on ex-dividend dates, as that's when a shareholder becomes almost (**) independently entitled to the payments. But there's also an argument against, which is that between the ex-dividend date and the payment date, the shareholder doesn't have access to the dividend - in effect, the dividend goes into a no-access, non-interest-paying bank account on the ex-dividend date, that only matures and pays out on the payment date, and that has a very small chance of not paying out at all. Such notional bank accounts have small negative expected rates of return, and being partially invested in them is essentially a compulsory part of running a HYP (or indeed any other type of dividend-paying share investment strategy). Doing unitisation calculations (or other types of performance measurement calculations such as Internal Rates of Return) using ex-dividend dates rather than payment dates is therefore effectively ignoring a compulsory, negative-return part of the strategy. It's only a small part of the strategy, since the percentage of a HYP's value which is tied up in ex-dividend dividends is probably only something of the rough order of 0.5%, but that does mean that such calculations can be expected to slightly overstate the strategy's real performance if they use ex-dividend dates. (Which is not to say "don't do it that way" - just that
if you want to worry about minor inaccuracies, add that one to your list of things to worry about!)
(*) Surprisingly well in many cases as long as you pick the right mathematical theory, and why that's the case is a bit of a philosophical mystery - but not a subject for this board!
(**) "Almost" because it is possible for companies to cancel dividend payments after they've gone ex-dividend. It's a pretty rare event - I'd only seen it happen one or two times this century before the burst of them around March/April last year - but it does happen.
Gengulphus