Case studies

[northSaver]

Yes I used Python.

So, I've had a quick play with your data in Excel (haven't learnt Python yet). I used the All World data because there are less rows and more applicable in my case. This is what I've found:

1. First I verified that your stated results are correct with the data given. Nothing personal of course, I would always do this before continuing!

2. Then I noticed that USD DCA outperformance is -2.6%, which is a lower magnitude than the LS performance (2.6 < 3.3). But after rounding they're the same number: +- 3%

3. Then I wondered what the results would be when the USD.JPY rate is relatively high, like it is now. This reduces the dataset considerably but here they are:

USD.JPY > 120: USD LS wins = 63% (72/115). JPY LS wins = 51% (59/115).
USD.JPY > 130: USD LS wins = 61% (33/54). JPY LS wins = 48% (26/54).
USD.JPY > 140: USD LS wins = 47% (8/17). JPY LS wins = 29% (5/17).

4. Then I wondered what the results would be when the USD.JPY rate is at least 5 yen higher at the start of the 12-month DCA period than at the end. This roughly simulates a falling yen. Here they are:

USD LS wins = 66% (97/147). JPY LS wins = 46% (67/147).

The results aren't too surprising. If the yen is historically weak then DCA wins. And if the yen is falling then DCA wins. There's not much in it though. Personally it's edging me towards DCA my New NISA next year rather than Lump Sum in January if conditions are the same. I think I would need to see signs of a higher for longer exchange rate (e.g. break and stay above 150) for me to prefer lump sum. Just my opinion based on what I've got

[TBS]

..

Broadly I think part of what you are trying to say is true. In order for JPY-based DCA to win, it needs the combined effect of the market movement and the forex movement to be a drop. So if the subsequent forex movement after a LS investment is a drop (yen getting stronger), that's a good thing for the chance of DCA winning.*

However the reasoning you then use to argue that DCA is therefore a good option for Jan 2023 contains basic flaws. Working backwards:

4. Then I wondered what the results would be when the USD.JPY rate is at least 5 yen higher at the start of the 12-month DCA period than at the end. This roughly simulates a falling yen. Here they are:

USD LS wins = 66% (97/147). JPY LS wins = 46% (67/147).

This calculation leaks information from the future back to the point you have to decide between LSI and DCA. You'll need to know what the exchange rate will be in Jan 2024 when you make your decision in Jan 2023 for this calculation to be relevant.

It's like saying you analyzed tennis matches where the underdog was 2 sets up, and found they were more likely to then go on to win the match than the favorite. So therefore that's evidence that in future tennis matches, before even any points or sets have been played, that underdogs are more likely to win than favorites.

3. Then I wondered what the results would be when the USD.JPY rate is relatively high, like it is now. This reduces the dataset considerably but here they are:

USD.JPY > 120: USD LS wins = 63% (72/115). JPY LS wins = 51% (59/115).
USD.JPY > 130: USD LS wins = 61% (33/54). JPY LS wins = 48% (26/54).
USD.JPY > 140: USD LS wins = 47% (8/17). JPY LS wins = 29% (5/17).

Similar here, the analysis uses information about the exchange rate from the full period to decide what a "high" and "low" yen is. However the "strength" of the yen at each point should only be judged on the data before that point.

Moreover, you chose to ignore all the exchange rate & USA market data from before the period you chose, because...??? probably because it wouldn't support your argument

2. Then I noticed that USD DCA outperformance is -2.6%, which is a lower magnitude than the LS performance (2.6 < 3.3). But after rounding they're the same number: +- 3%

"So relatively LSI did 27% better than DCA in magnitude (3.3/2.6-1), but because they round to the same whole number that's evidence that their performance is actually the same"

Also, could you explain a little more how you calculated 3.3 & 2.6? Intuitively they feel wrong too me - magnitudes are too small if they are meant to be the average outperformance of LSI when it wins, and for DCA when it wins.

1. First I verified that your stated results are correct with the data given. Nothing personal of course, I would always do this before continuing!

Thanks, did you find any issues?


* it's not a sufficient thing to guarantee that DCA will win, as a market movement upwards could outpace the rate of yen strengthening and the LS investment will wins.

[northSaver]

Broadly I think part of what you are trying to say is true. In order for JPY-based DCA to win, it needs the combined effect of the market movement and the forex movement to be a drop. So if the subsequent forex movement after a LS investment is a drop (yen getting stronger), that's a good thing for the chance of DCA winning.*

However the reasoning you then use to argue that DCA is therefore a good option for Jan 2023 contains basic flaws. Working backwards:

Hmm, I don't really get what you're saying. I'm just calculating statistics based on historical data to perhaps offer some peace of mind when it's time to choose LSI vs DCA. You said on page 2, "The currency issue doesn't make a big material difference tbh". If you don't have an opinion about the strength of the yen or it's future direction, then the average you provided (68% win rate for LSI for MSCI ACWI) is a good guide. But it's just an average across all market conditions. If you might have an opinion on the yen then you need to dig deeper.

The period I used was the maximum available for ACWI: 1988 to 2022. It's all the data that you gave. It's 35 years, which is long enough I think. During that period the highest USD.JPY rate was about 150, which is why I chose the parameters I did to simulate a weak yen. I'm not interested in buying USA-only funds with my yen, so I didn't use that data. If I had, then I'd have chosen different parameters because before 1987 USD.JPY was above 150 all the time. I've said several times that if it breaks and holds 150 then we could be entering a new era for the yen, like the years before 1987. But I don't believe that to be the case at the moment.

I'm sorry if my analysis came across as incomplete or manipulative. That wasn't my intention! Perhaps I should have kept in private? I don't know. Like everything I post on this forum, please take it or leave it

[TBS]

If you might have an opinion on the yen...

Yes this is the crux of it. It's tempting to have an opinion on what the usd-yen will do in the short-term, but in practice most people will be wrong.

And...

to perhaps offer some peace of mind when it's time to choose LSI vs DCA.

Accepting that there is no way to predict what fx rates and the market will do in the short term, introducing a choice between LSI & DCA becomes entirely unecessary from a maximizing expected returns standpoint. It's introducing a choice about investing where there doesn't need to be one.

I'm sorry if my analysis came across as incomplete or manipulative. That wasn't my intention!

That's not the case. I just think you're making classic investing mistakes - regret aversion, mining historical data for associations and believing that they will help you make better investment decisions in the current climate. You won't be able to make a sound argument from a numbers and causality standpoint that you'll get a higher probability of return than just throwing the money straight in.

[bellule]

Hello,
I am planning to send a large amount (EUR100k +) from Europe to Japan. I will do it by bank transfer.
I have a question about FX. Would it be better:

• Option 1: To ask my bank in Europe to change the amount in JPY, then send it to Japan?
  Option 2: To ask my bank in Europe to send it in EUR and to receive it in EUR in my Shinsei bank account, then make the change from here?

Or maybe it does not matter??
Any FX/bank transfer expert around, please advise!

[sutebayashi]

For the transfer, there are two aspects.
1) sending money - just sending money can incur fees, which is understandable for a valuable service. But hopefully you can send for a low rate flat fee, no a fee that is proportional to the amount you send
2) the forex conversion. You can certainly get a very tight spread on a conversion by using a forex broker here in Japan, but on the other hand establishing such an account might be a chore one would rather avoid. And more than the conversion spread, the timing of the conversion is probably more important. E.g. you might be able to change at .02 from the market rate, but if the rate moves against you by 2 yen that would defeat the purpose, whereas you might pay a .5 conversion rate, but if the EUR/yen had moved in your favour then it would compensate for the poor spread.

I used to use a forex broker a decade back, but these days some banks are offering somewhat more reasonable rates, so there is a case for just converting at a good Japanese bank and trying pick a good time to covert.

Wise is another oft mentioned option that I have never used.

In short, I don’t know what your European bank offers, but would guess your number 2 option could be more in your favour.

[bellule]

Hi @sutebayashi,

Thank you very much for the detailed answer. As it is a one off transfer, I will not go through a forex broker, it seems too onerous to me. Get it for your advice on the second option, I will take this one.
Have a great day!

[RetireJapan]

You can probably check the rate your European bank will use vs the rate Shinsei uses. I would assume that you will get a better rate from Shinsei, but it might be worth investigating.

[adamu]

Another thing to consider is that when using a bank, the intermediary bank fees are not published, so it's basically impossible to know what the eventual fees will be in advance.

[tagawa]

Another data point, if it helps:
In the past I've used OFX (GBP->JPY) with good rates, although yes, it's a bit more onerous than simply bank->bank.

I also did a US->Japan transfer to my Shinsei account a couple of years ago that went over some threshold and triggered a security check. They phoned me up and wouldn't release the funds until I'd sent paperwork to show where the funds came from. It was fine in the end but a bit of a hassle, and made it hard to predict how long it would take.

I seem to remember the Shinsei rates being better than I expected so I'd probably choose that for a one-off transfer, but be prepared for a paperwork request.