[COFFEE] History and histories: historical data, charts, long-term trends in investment

Ha! The analogy with baseball is even better than I first thought: a single inning lasts as long as it takes until three hitters of each team* strike out. Typically, a game of nine innings takes a couple of hours, but there have been single innings that have lasted over an hour while other single innings have lasted only minutes.


* One team hits in the top half (first half) of an inning, the other team hits in the bottom (second half) of an inning. Striking out means not hitting a pitch (the ball thrown by the pitcher of the other team) if the ball arrives in a certain zone, or hitting the ball but the ball is caught by the other team. It's actually a little more complicated, but that's the essence.
2 Likes

I’m not sure why they say less impulsive in the final innings. Typically as we reach the end, is when things go crazy and charts go parabolic.

Like openAI launch of CHATGPT pushed US markets to all time high, would deepseek launch change the fortunes for Chinese markets?

The model seems to be one of the best , built with 6 million dollars budget , in few months and without using latest NVDIA chips.

Wasn’t quite sure where to post this, in this topic or in some Chronicles thread. Anyway, 2023 snapshot of who holds U.S. (government) debt 


(Source)

1 Like

that’s a weird infographic, are the boxes proportional to the values?

1 Like

For me the main point of the infographic (even if the boxes aren’t exactly proportial to the values stated): I was amazed how much US debt was held domestically versus internationally and how little debt is held by, say China.

Incidentally, after a bit more doomscrolling on Twitter I came across this infographic of Foreign Holders of U.S. Debt in 2022:

(Source)

But since you asked whether the first infographic was really exact: I asked a couple of AIs “are the boxes proportional to the values in this infographic that I would like to upload”

Chat-GPT failed at analyzing this and gave me pseudocode on how to check this myself.

DeepSeek said that it can’t analyze images directly, but that if I provided the area sizes and repective numbers, it would be able to tell me if they were proportional. Duh.

Gemini says in short: no. In length see below.

(If you question the exact proportions of the second infographic, you’ll have to do the measuring yourself – or ask you favorite AI yourself :wink: )


Gemini’s long answer:

Let’s analyze the infographic “Who Holds U.S. Debt?” you’ve uploaded.

Are the boxes proportional to the values?

No, not in a straightforward, visually consistent way. Here’s why:

  • Area vs. Linear Dimension: The boxes seem to be attempting to represent the values by area, but it’s not done accurately. Ideally, if a value is twice as big, the box’s area should be twice as big. This doesn’t appear to be consistently followed.
  • Inconsistent Scaling: Look at the “Federal Reserve System” box ($5.2T, 15%) and compare it to the “Mutual Funds” box ($3.7T, 11%). The Federal Reserve’s value is significantly larger, but the box area doesn’t reflect that proportionally. Similarly, the “Intragovernmental Debt” ($7.0T, 20%) box looks far too small for its value compared to the others.
  • Focus on Percentage, Not Absolute Value: The infographic seems to prioritize showing the percentage of total debt held by each category rather than the absolute value ($Trillion). While the percentages are helpful for context, the visual representation of the boxes should still reflect the actual values they represent.

What the infographic does well:

  • Clear Categories and Labels: The categories of debt holders (Federal Reserve, Mutual Funds, etc.) are clearly labeled, making the information easy to understand.
  • Percentage Breakdown: Providing the percentage of total debt held by each category gives a good sense of the relative distribution.
  • Total Debt Figure: Including the total U.S. debt figure ($34.4T) provides context for the individual values.

Suggestions for Improvement:

  • Consistent Area Scaling: If the intent is to use area to represent the values, the boxes should be scaled accurately based on the absolute values ($Trillion).
  • Alternative Visual Representation: A horizontal bar chart might be a clearer way to represent these values, as bar length is easily compared.
  • Clarify Visual Method: If area is intentionally not proportional, the infographic should explicitly state how the boxes are scaled (e.g., “boxes are for visual categorization only and do not represent exact proportions”).

In conclusion, while the infographic provides useful information, the visual representation of the values is not proportionally accurate and could be misleading. It’s crucial for data visualizations to accurately reflect the underlying data to avoid misinterpretations.

but is the debt between various government agencies/institutions really relevant or is it a technicality due to how things flow?

I think the (first) infographic only says who holds U.S. Government debt (debt issued by the U.S. Treasury), nothing in terms of the question you ask above.

Looks like US will remain in debt forever :slight_smile:

2 Likes

Human intelligence says yes, I had a few minutes, so I checked a few box areas and they were close enough, for human visual comparison, i.e. their area / total area corresponded close enough to the % debt and the absolute values that the boxes were supposed to represent.

Gemini, I can’t confirm your examples of inconsistent scaling. Maybe you should squint more.
No, but seriously, I think you are faking your intelligence, by giving some good sounding explanation. I compared the examples you quote, and the areas are spot on.

width height area % area debt % of debt
Total 8.8 9.4 82.72 34.4
Fed Res System 4.4 2.9 12.76 15.4% 5.2 15.1%
Mutual Funds 3.15 2.9 9.135 11.0% 3.7 10.8%
Intragovernmental 7 2.4 16.8 20.3% 7 20.3%

Gemini, you say “The Federal Reserve’s value is significantly larger, but the box area doesn’t reflect that proportionally”, whereas actually this box area is a little larger than it should be proportionally!

Ja, Gemini, such a infographic definitely prioritises percentage held by each category, thanks for stating the obvious. The boxes do actually represent the actual values quite well also, so your implied critisicm is wrong.

3 Likes

A Oy is going to take over the world.

This year at the latest.

:wink:


Edit:

Gemini probably learnt from the eternal wisdom of the Interwebs that if someone asks a question whether an infographic represents the data stated accurately, it’s best to say no. Gemini’s answer by the way came immediately after I hit enter after my question, while both Chat-GPT and DeepSeek took a moment to give a non-answer.

Someone put a wait instruction into their code to “simulate hard thinking” and get people to trust its replies more, I’m sure. :sweat_smile:
Ah the human psychology. :slight_smile:

(Next they will remove or reduce the timeout by half and claim “we are now twice as fast!”.
One of the oldest tricks in the book. :wink: )

5 Likes

https://www.visualcapitalist.com/the-worst-performing-stocks-in-the-sp-500-2005-2024/

1 Like

https://www.morningstar.com/funds/15-funds-that-have-lost-most-value-shareholders-over-past-decade

With some news about our favorite fund manager:

ARK, home of the flagship ARK Innovation ETF ARKK, saw the biggest aggregate losses in dollar terms. After garnering huge asset flows in 2020 and 2021 (totaling an estimated $29.2 billion), its funds were decimated in the 2022 bear market, with losses ranging from 34.1% to 67.5% for the year. Many of its funds enjoyed strong rebounds in 2023 and 2024, but that wasn’t enough to offset their previous losses. As a result, ARK funds incurred approximately $13.4 billion in realized and unrealized capital losses over the 10-year period—about twice as much as the next fund family on the list. ARK Innovation alone accounts for about $7 billion of this total. Both ARK Innovation and the other funds in the group earned a positive total return over the 10-year period ended in 2024, but poorly timed flows into its funds proved costly because most shareholders bought in after performance had already peaked.

4 Likes

The daily deluge of “AMAGAD SHOULD I SELL NOW” posts on reddit had someone post this article. I think it’s a good article, actually, as it makes an effort to add some depth to the “time in the market beats timing the market” quote. Key points:

  • Missing the X best days in the market is catastrophic indeed, missing both the best and the worst days beats the index, but not missing either has been pretty GOOD ENOUGH historically.


I’d like to see this graph from as far back as possible, though, as I wonder if 2010-today could have skewed this by being an unprecedented bull run. Things don’t look that bad pre-2010, and it doesn’t seem like there were any days from 2010-onwards making it to be either among the best or worst, just compounding of steady gains.

  • Volatility clusters - the best days often come very close to the worst days (days to weeks), making trying to be right twice and guess local tops and bottoms is futile. As is resistance, all your base are belong to us. Lower your shields and prepare to be assimilated to the market.


What’s interesting about these numbers is that 75% of best and worst days happened before 1947, making me wonder if this kind of volatility is a thing of the past. 1987 I understand to be due to early trading algorithms getting out of whack. Note that 2000-2002 doesn’t even make it to the table, yet it’s remembered as the catastrophic dot com crash when it really was a prolonged grinding way down where hundreds of days of small losses added up to something big over many months.

  • Most best days happen in the worst years, leading to think that the “best days” argument is misleading, because staying invested catches the best days but the damage done by the worst days/months/years is not offset. A 50% loss needs a 100% gain to get back to zero.

  • Time in the market most likely beats timing in the market, but not because of best vs worst days, as these can’t be caught by more or less anyone.
3 Likes

I would say all this theory is just to send one message to investors

Be a trader OR be an investor

If the choice is made to be an investor then make a decision (diversification, asset allocation , regional allocation, etc) and stick to it for long term and only change it if something really changed in fundamentals. I know this is very hard because people always want to be a bit better than what an average person can expect.

In other words -: don’t say I am an investor but act like a trader

I am more or less certain than anyone who sticks to one of the following for 20 year period would not end up being unhappy with their decision in absolute terms. Of course they can always find a reason to be unhappy in relative terms (what if I invested in that best portfolio ;))

  • 100% world equity ETF
  • 60-40
  • Golden butterfly
  • All weather

„Absolute terms“ means investor had a good inflation adjusted returns that made meaningful impact on their financial situation

6 Likes

To try to counteract that, I like to think that the average person tends to buy more after the market/some fund/stock has experienced big gains and sell when stock prices are getting low.

In my understanding, getting the average market returns is beating the average investor, as they tend to not capture all of the gains (nor all of the losses but with a psychological bia leading to under performance).

7 Likes

This one can be downloaded for free, has the original Spanish at the left and modern Spanish at the right. Remembers me to the Don Quijote I tried to read in original 40 years ago


Versión en español moderno, introducción y notas
Ricardo A. Fornero
Universidad Nacional de Cuyo (Mendoza, Argentina)

1 Like

True.

Realised average is lower than expected average

1 Like

That‘s not only your thinking, the statistics on that are very clear.

Don‘t have it on hand right now, but have seen some statistics on that in the past. Retail investors sell a lot in drawdowns and buy a lot at peaks.

1 Like