I have written before about reinvesting in solid income bearing, slightly unfashionable stocks. I had an interest in seeing what happened over time to some renewable companies (long before I was in a position to liquidate holdings and invest).
Some time ago I noticed a peculiarity in some solid dividend paying stocks. Whilst it is known and regular that on going ‘exdiv’ (ie the point at which you can sell the share and still receive the payment) and the next working day the price drops. Nominally it is by the dividend payment. The theory is the price goes up to reflect the whole value of the business (which included cash) and then drops as a result of cash being payed out (or essentially moved to the outgoings tray.
Frequently the share price does not drop by the entire amount, and in the case of some higher and less regular payments actually drops more.
So I started thinking … what if ? .. could I exploit that in any way ?
—
This whole exercise gathered a life of its own as my mind kept asking ‘whatif’, so I have split up what would have been a monster article of around 7,000 words into four parts:

 Part 1 – My initial look at just my four tracked Renewable companies and extend that to 18 holdings but only over the initial short period of less than 4 years
 Part 2 – I extend the list to 18 holdings and look at a 10 year period.
 Part 3 – The big one: I look at 18 holdings picked randomly from a selection of around 70 over a 15 year period (covering two market corrections). I look at a Monte Carlo simulation
 Part 4 – Finally, just in one place a summary of my observations
Split up so you could move to what may interest the most – been quite an exercise ! But if nothing else, take a look at the summary in Part 4.
—
My Initial Holdings
So I started by looking at running the data of four stocks I have long term tracked and playing with dates to reinvest.
Although not a huge surprise, the difference over the 3.4 years of data I have is less than a piece of paper… it is negligible and noise. I was more surprised that picking different dates (up to 20 days before and 20 days after exdiv) made a negligible difference. I don’t know why, but I did expect a bigger variance. The effect of the relative small variances in stock prices in this 40 day (8 week) window for each stock made a negligible difference to the final portfolio value.
But I did start to think that I should look wider. So I filtered some holdings on my data package, broadly looking for:

 Shares that had gone up and down largely in a range
 Growth limited to no more than around +15% or 10% over that period (2nd Oct 2017 to 12 Feb 2021)
 Pay a decent dividend (3% or more)
 Make 4 payments a year.
I ended up with around a dozen holdings. I widened a bit and looked at 2 payments a year and added a few more. In the end, I had 18 holdings and aside from the 4 I was actively tracking (for other reasons) the rest are purely random. I have heard of a couple but really know absolutely nothing about them.
…. This is an academic exercise remember !https://www.firemusings.org/dividendreinvestmentdoesitmatterwhatstrategyyouhavepart4/
—
So what is the Idea/Hypothesis ?
I have three simple variables I would like to consider:


 Maybe it really does not matter when you reinvest – just so long as you reinvest in something
 Maybe only reinvesting on gains or dips is best
 Limiting reinvestment to meaningful amounts (and not necessarily the actual dividend which may be small), i.e., taking account of trading costs

Some Basics
There are 18 holdings I selected. Details in the appendix at the bottom, but really they are not as relevant as the picture I am trying to explore.
As the shares all have different pricing the only way to see the relative picture is to rebase. So a graph of the rebased prices relative to the first day is:
What does this show us ?

 Most holdings dance around the centre line (ie hold price within a small range) which is not a surprise given the criteria
 There are one or two poor performers and one or two very good performers. This is not unrepresentative of any portfolio of 15 or more holdings
 The clear impact of the March 2020 correction is visible
Overall, this is not (or should not) be a surprise in a portfolio: a small number of loser’s and some solid one and one or two really good ones.
Graphing out growth in terms of the min, max and end point, the holdings behave pretty much as many stocks do: they dance about but with a positive slant.
Again, not a surprise, they are modestly preforming higher yield stocks – growth over a period dances between limits but a generally upward overall momentum.
Dividends can be seen to be pretty regular and the cumulative amounts are as follows:
Obviously the big mining company Rio Tinto (RIO) has rather higher yields in this period.
—
So how can I analyse this and what parameters make sense ?
Well there are I feel 3 parameters here:


 I can just reinvest every dividend directly back into the stock it came from
 I can reinvest on either a SP rise or fall (over the previous period)
 I can only reinvest above some minimum amount (to reduce transactions and costs)

In addition I can include or not include costs (although makes more sense to include them so there is no distortion). I could look to:

 Invest all direct dividends
 Invest if the SP changes by a given percentage over the previous quarter
 Given percentage and the amount must be over a certain sum – this means an element of rebalancing and picking the higher moving stocks
Note – only a bit of the tables are show just for method – who really wants to see a vast data table – it is the lesson here that matters…
So Direct, this is what the investment looks like. Each line is a separate quarter and the amount is that reinvested
Now if I preselect (in this case) any holding growing (or dropping if I run that) by at least 2% in the quarter
And finally if I limit reinvestment to a minimum (in this case) of at least £300. The way I did this is sum all the dividends, filter those meeting the criteria and then apportion the sum based on how much movement, ie a bigger rise (in this example) gets a bigger reinvestment
By running this for each variable and tabulating I get a big boring table:
And finally from this I can run a few basic statistics Min, Max, Median, Standard Deviation and Number of Transactions.
Now there are two really interesting things come out of the data:

 The costs are linear. Yes each transaction is a fixed cost (I set at £10 + 0.5% tax). I am a bit surprised by this as I expected a curve. Intuitively more transactions, more costs, less compounding, but at least in this timeframe and scope it is pretty linear. (I wonder if the same would be true over a longer time ? Would compounding be noticeable ?)

 There is a very clear advantage to ‘buying the dips’. The central ‘point’ (at 0%) is the sum for direct reinvestment. The vertical line variability comes from picking the minimum threshold the reinvest
 There is an approx 5% net gain to buying the dips
 In each group the highest outcome comes from not having threshold investment but simply reinvesting if the dip (or gain) is met. For example, if I was looking at dips of at least 2%, then any holding that met that criteria the dividend was reinvested, not just if it was (say) above £300.
 There is a very clear advantage to ‘buying the dips’. The central ‘point’ (at 0%) is the sum for direct reinvestment. The vertical line variability comes from picking the minimum threshold the reinvest
In investing there is always the old hands whispering on forums ‘buy the dips’ … well…. first time I really looked and although it is dividend reinvestment, I expect that still holds generally. This is the first time I have really looked and can see where that maxim comes from
A final note on timing. As I stated earlier, the timing of the reinvestment made fractions of a percentage difference. Overall, the number of shares total at the end was no more than a handful – just noise, so to make things both simple and more likely how I would operate, I notionally collected the dividends and then did all reinvestment on the last trading day of the quarter.
In a practical sense, would anyone really just do all this on a daily/weekly basis ? I doubt it,and unless some form of cheap autoinvest was used why would you. Just shown above that selection based on the stocks that dip gives a better outcome so again, why would you do an automatic ‘fire & forget’ strategy ?
Now a sit down and review every 3 months is perfectly both believable and practical.
—
So I have been surprised but I am left with a few questions still ….


 Am I being a bit biased here ?
 Are my assumptions really just on the contrived side ?
 Does the fact that I picked very ‘likeperforming’ stocks really mean these results are less than fully valid ?

Onto Part 2 …..
—
Note: None of the above is investment advice. It is my observation and comment. Always do your own research and make your own decisions
—
Appendix
 The Stocks I chose were as follows: FSFL NESF TRIG UKW BBOX SEQI JLEN ACI CSH HICL DIGS MUT NG RIO SLA PENN ULVR UU
 Period of time is from 2nd Oct 2017 to 12th Feb 2021