Skip to content

May 2023: Doing Research with GPT

and some very loose notes on banking and crypto

Hi folks, this is indeed “A Month of Blips”, and here is the May 2023 edition.

Two topics today:

  1. Doing Research with GPT. I explain the importance of understanding context length and temperature.

  2. Latest thoughts on payments and crypto. I don’t explain a lot 🤣.

Doing Research with GPT

The full video is 25 mins, but you can also watch on higher speed and focus on the parts of interest:

Reflections on Banking and Crypto

Mean Reversion

  • When people say its the future, they are overly optimistic.

  • When people say its doomed, they are overly pessimistic.

A passive way to trade this phenomenon is to buy things in equal weight – e.g. 50% stocks, 50% cash – and then re-balance periodically. Automatically, you are buying low and selling high. This is like buying an equal weight stock portfolio (e.g. own the same value of each stock, e.g. RSP) instead of a market weighted portfolio (e.g. own an amount proportional to the value of each company, SPY – which is much more common).

Said differently, this is the Warren Buffett strategy. He’ll avoid tech stocks because they are overhyped (except Apple, apparently), and he’ll buy things that are boring like Coca Cola or newspapers.

That said, mean reversion can fail. Sometimes the winners continue to be winners. That has been the story of the stock market in the US over the last decade. It has also been the story with Warren Buffett’s relatively recent purchase of newspapers. He was betting they would do less badly than people thought, but they didn’t.

All of this is to say… crypto is doing very badly in the news in 2023. Government lawsuits have been brought against pretty much every exchange. Are people overly pessimistic? Is it a chance for mean reversion? I don’t know. The upside from here for crypto seems unclear. Crypto is a bet on change, not a bet on mean reversion. But, what is the change and why is it better?

What payment systems work really well?

  • UPI or PIX. These are new-ish systems in Brazil and India that allow for instant, almost free payments between bank accounts.

  • UK faster payments (and also SEPA instant in some EU countries, not Ireland).

These system rely on banks. Banks bear burden of fraud and dispute resolution and are indirectly compensated via interest earned on deposits (in high rate environments) + the permission to lend. Yes, there may be some revenue from transaction fees, but with UPI and PIX and SEPA instant, these fees are low.

Why doesn’t SEPA instant work throughout the EU? It seems to be a coordination problem whereby all banks aren’t up to speed.

Who gets the interest on deposits?

Banks may not get as much interest as they used to. Technology is making it easier to move money. There are more options for finding neobanks like Wise (who pay interest) or Revolut (who don’t, yet), or investment accounts [like Fidelity or Schwab, although options are very poor in Ireland].

Technology is bringing competition to the banks! Or is it?

The big banks are attracting more deposits in the US. There is more certainty the government won’t allow big banks to fail. Despite customers having more options for where to park their cash, JPMorgan is reporting rising profits.

eMoney is a little broken

It’s often unclear whether and how deposit insurance applies to money in neo-banks like Revolut, Wise, Venmo. These neobanks sometimes operate partly with eMoney licenses (whereby eMoney is backed by deposits in fully licensed banks) and not full banking licenses (which allow lending out more money than is on deposit). The US regulator just gave a warning on both a) stablecoins and b) money in neobank accounts! Ouch.

It appears a neobank can get deposit insurance, but involves the customer being registered back with the with fully licensed bank (e.g. JP Morgan) where the neobank holds their reserves. This must not be straightforward, otherwise I imagine deposit insurance on neobank deposits would be less murky.

A further drawback of eMoney is that you can’t easily move funds from Revolut eMoney to Monzo eMoney (or choose another competitor). The money is accounted for on different systems (Revolut’s and Monzo’s). If you allow for it to be all on the accounting system, then it would effectively be a Central Bank Digital Currency. This would partly undercut the business model of these neo-banks. So, are neo-banks getting squeezed between the upcoming central bank digital currencies and the traditional banks who cannot be let fail? I’m unsure.

As a side note, the only real difference there could be between regulated stablecoins (like USDC) and eMoney would be the openness, i.e. the ability to move USDC between accounts that are not controlled by Circle (the issuer of USDC). So, stablecoins could potentially undercut the models of neobanks. If only stablecoins had a regulatory framework… but they do, under the Market in Crypto Assets regulations in Europe. What then if the European Central Bank brings in their own digital currency? Wouldn’t that undercut everything?

Everything seems so choppy with eMoney and stablecoins. I don’t have a strong view on the future.

Government Debt is a problem without growth.

  • Spending more than you earn is a problem for individuals and governments.

  • There isn’t a critical debt to GDP ratio for governments. It’s whatever the system can bear.

  • High economic growth makes higher debt to GDP ratios more ok.

  • Right now, debt is getting higher in the US and EU, but growth isn’t. That should be a concern.

  • Raising taxes and retirement ages and reducing spending is very hard politically.

  • It seems important for countries with growing Debt to GDP ratios to think a lot about economic growth, not just raising taxes or cutting government spending (although maybe some of that is necessary too).

Privacy is for criminals.

This topic isn’t talked about much. Maybe some day it will make a return.

Three options for Crypto.

Crypto will either a) fail, b) be co-opted by the west’s traditional structure (maybe only in the EU?) or c) build a network of its own that gains legitimacy with governments internationally (not seeming likely). Lots of consequential court cases are happening in the US that will determine whether new US regulations allow for option b in the US. Economist Tyler Cowen has suggested crypto will evolve to be used by AI bots that need to pay for things. I don’t think that has to be the case. It may be that bots have to have a bank account tied to them if they want to spend, and that bank account has to be registered to a person or company.

If crypto didn’t have speculation, it would have less grassroots support than it does today. Given crypto does have a speculative upside, it has led the industry to be defined and dominated by speculation and fraud.


Yeah, people are still buying that stuff. And governments (less so the US and Europe).

Where crypto has an advantage (or not).

Maybe one area where crypto could have an advantage (setting aside some small chance of uptake as a form of open eMoney in the EU) is for international payments and competing with wires over SWIFT. However, the main currencies used in crypto are either a) dollar denominated or b) volatile (Bitcoin). If payments are US denominated, the US has de-facto control and can block payments (perhaps except with LUSD, although even there there are vulnerable points, such as price/oracles). So, even crypto is largely a form of SWIFT, maybe a slightly cheaper and instant, but much more risky – due to it’s lack of regulatory grounding – profile.

Where I’ve seen some crypto uptake for payments is in LatAm and Africa. Digital dollars (USDT, USDC) are a more portable form of US dollar bills. That’s a much smaller market than penetration of eMoney markets in the US and Europe, but it’s a market where better financial services are more in need. The same questions – as with dollar bills – applies about whether the US and Europe and India and China want to allow their dollars/euros/yuans/rupees to be used abroad. Do these countries want their currency to be used abroad? Do these countries want their currency to be not used abroad? Why shouldn’t LatAm/African/other countries use their own local currencies instead of someone else’s?

The “more of the same” bet here is on the continued strength of the dollar and the partial reserve banking system. Or is it? Partial reserve banking has evolved a lot over the last hundred years. There was a gold standard in the not too distant past. If we are mean reverting, then what is the mean? And does technology, particularly digitisation, take us away from the mean of the past, and if so, why?

I don’t have a clear synthesis on banking.

Growing debt seems unsustainable without economic growth. Beyond that, I don’t have a clear synthesis here.

I recently re-read a quote from Jeff Bezos about asking not what will change in the future but what won’t change. It’s better to build a business on what won’t change – like consumers wanting goods delivered more quickly and for a cheaper price.

I don’t have that kind of synthesis here on banking. It seems digitisation is having too much of an effect for banking and money not to change. Although, many people have said that before, while Buffett made the right bet on on American Express (credit cards).

Leave a Reply