Personal Finance (23): The 200 Billion RMB Lesson

I recently came across a financial research paper that I felt was worth sharing — not because its conclusion is comforting, but because it is not. It quietly documents something most of us sense but rarely see proven: that in a market bubble, the losses of the many become the gains of the few. What follows is my attempt to make its findings accessible to more readers. 

How a Bubble Quietly Moved Money from the Many to the Few

Based on "Wealth Redistribution in Bubbles and Crashes" by Li An (Tsinghua PBC), Dong Lou (LSE), and Donghui Shi (Fudan), published in the Journal of Monetary Economics, vol. 126 (2022), pp. 134–153.

A Rollercoaster With a Hidden Toll Booth

Between July 2014 and June 2015, the Shanghai Composite Index climbed more than 150%. By the end of December 2015, it had crashed 40% from its peak. Most people remember this episode as a wild ride that ended roughly where it started.

But three researchers got access to something extraordinary: the Shanghai Stock Exchange’s regulatory bookkeeping data — the daily holdings and trades of all ~40 million accounts in the market. Not a survey. Not a sample. Everyone.

What they found should be required reading for every retail investor.

The Numbers That Should Keep You Up at Night

The researchers sorted all household accounts into four wealth groups by account value, with cutoffs at 500K, 3M, and 10M RMB. The bottom group holds 85% of all accounts; the top group is the wealthiest 0.5%. Remarkably, despite that huge difference in headcount, the two groups started with roughly the same total wealth in the market — which makes the comparison between them clean. Then the researchers watched what each group did, day by day, through the entire boom and bust.

The pattern was almost embarrassingly clean:

During the boom, the wealthiest group (accounts above 10M RMB — the top 0.5%) poured money in early and aggressively. By the market peak on June 12, 2015, their cumulative adjusted inflows reached 406 billion RMB. The bottom 85% were actually net sellers during the rise — their adjusted outflows reached 460 billion RMB by the peak.

After the peak, the roles reversed with brutal speed. In the ten weeks it took the index to fall from 5,166 to 2,927, the wealthiest group pulled 365 billion RMB out of high-beta positions. Who bought their shares? The bottom 85%, who piled in during the crash — adding 257 billion RMB of adjusted inflows after the peak.

The bill: over the full 18 months, the bottom 85% of households lost 252 billion RMB from their active trading, while the top 0.5% gained 252 billion RMB. Measured against what each side started with, the poor lost about 28% of their initial equity wealth, and the ultra-wealthy gained about 31% — roughly 30% either way.

Same market. Same 18 months. Same information officially available to everyone. Opposite outcomes.

Where Did the Losses Actually Come From?

The researchers decomposed the transfer into two channels, each responsible for roughly half:

1. Market timing (when to be in the market). The wealthy entered early in the bubble and exited shortly after the peak. The poor entered late and — devastatingly — kept buying through the crash.

2. Stock selection (which stocks to hold). During the boom, the wealthy tilted toward high-beta stocks — shares that amplify market moves — and rode the updraft harder. After the peak, they rotated out of high-beta names while smaller investors rotated into them, absorbing the steepest part of the fall.

And here is the detail that elevates this from anecdote to indictment: the researchers ran forecasting regressions and found that trades by the poor negatively predicted future stock returns, while trades by the ultra-wealthy positively predicted them. A one-standard-deviation increase in weekly buying by the top group predicted a 0.44% higher return the next week; the same buying by the bottom group predicted a 0.48% lower return — a gap of 0.93% per week that is both large and statistically overwhelming. The poor weren’t just unlucky. Their trades were systematically pointed the wrong way.

One more finding deserves emphasis: the skill gap is not constant — it widens with turbulence. That 0.93% weekly return gap between the top and bottom groups during the bubble-crash was nearly five times the 0.19% gap the same investors showed in the calm years of 2012–2014. The authors ran the whole study again on that calm period as a control, and the wealth transfer there was an order of magnitude smaller. Volatility is not neutral. It is the amplifier through which skill differences become wealth differences.

The Authors’ Verdict: It’s Skill, Not Luck

Could this be innocent? The researchers tested the polite explanations:

●        Rebalancing needs? Their model shows rebalancing-motivated trades explain less than 20% of even the market-timing half of the transfer.

●        Changing risk appetites? There’s no plausible reason risk aversion should gyrate in exactly the pattern required.

●        Trend-chasing that happened to work? The regressions show no clear trend-chasing by the wealthy at all.

What remains, in the authors’ words, is heterogeneity in investment skills — the ultra-wealthy have better access to information on both aggregate market movements and individual stocks. And crucially, this advantage is amplified in bubble-crash episodes, precisely when volatility and trading volume peak.

In plain language: bubbles are when the skilled harvest the unskilled.

What the Paper Doesn’t Say

Here the paper stops — almost. The authors document the transfer and identify the cause, but offer the ordinary investor no remedy. They do draw one broader implication. Bubbles and crashes are as old as markets themselves — from Dutch tulip mania to the South Sea Bubble to 1929 to the dot-com bust — and this quiet transfer from the unskilled to the skilled is a recurring feature of all of them, rich world and poor alike. The authors add one sobering note: because so many people enter the market for the very first time during these bubble episodes, the damage can be especially lasting where stock-market participation is still young and first-time investors are the majority. Their one policy caution follows from this: passive investing can help anyone, but active investing in bubble-prone markets "may result in the exact opposite."

That is an honest warning. But it is also a narrow one — a caution about how people participate, not a full answer to what an ordinary investor should actually do. The paper leaves that question open.

Where the Commentary Goes Wrong

Into that open question, popular commentary has rushed with a simple lesson. I first came across this study through exactly that kind of commentary, and the takeaway was always some version of the same thing: the little guy gets fleeced in bubbles, so the safest move is to stay out.

I understand the appeal of "just avoid it." But I don’t fully agree — and it’s worth explaining why, because the flaw in that advice is the whole reason I wanted to write this. Notice what "stay out" quietly concedes: it tells the little guy to sit at the edge of the pool. And that is exactly where I part ways, because sitting at the edge does not stop you from getting poorer relative to those who swim. If the disease is a skill gap amplified by volatility, the treatment is not abstinence. It is skill.

So What Can an Ordinary Investor Actually Do?

First, understand the game you’re in. The wealthy in this study didn’t beat the poor with secret stocks. They beat them with relative thinking. Retail investors typically ask a time-series question: "Is this stock going up?" Skilled investors ask a cross-sectional question: "Which stocks are positioned better than which, right now — and where do I stand relative to everyone else holding them?" The study’s own evidence shows the wealthy’s edge lived in the cross-section of stocks. You may never match their information access, but you can stop playing the single-stock guessing game entirely.

Second, refuse to supply the exit liquidity. The single most destructive behavior in the data was buying after the peak, from wealthy sellers, in high-beta stocks. If you cannot say precisely why a falling stock is cheap, "it’s much lower than before" is not analysis — it’s the exact reasoning that cost the bottom 85% some 252 billion RMB.

Third, make volatility your tripwire, not your temptation. The transfer concentrated in the most volatile stretch of the episode — the skill gap between top and bottom investors was nearly five times wider in the bubble than in calm years. When markets get wild and everyone around you is opening accounts, that is when the gap costs you the most. Calm markets forgive amateurs; volatile markets bill them.

Fourth, automate what you cannot discipline. The wealthy’s timing looks like skill, but you don’t need timing skill if you remove timing from your process: fixed periodic investment into broad, low-cost index funds; a written allocation you rebalance on a calendar, not on emotion; position sizes small enough that no single crash forces your hand. A rules-based process is the retail investor’s substitute for the information advantage you’ll never have.

Fifth, treat leverage as the amplifier of your side of the skill gap. This bubble was famously fueled by margin trading. Leverage doesn’t just amplify returns — it amplifies whatever skill differential exists between you and your counterparty. If the data says your trades predict returns negatively, leverage means losing faster.

The Real Lesson

Markets do not merely reflect wealth inequality. In their most dramatic moments, they manufacture it — quietly, legally, and at scale. Over 18 months, one bubble moved roughly 250 billion RMB from those who could least afford it to those who needed it least, and the mechanism was nothing more exotic than a difference in skill, multiplied by volatility.

But the conclusion is not to flee the market. Fleeing only guarantees that the harvest continues without you — and the gap widens either way. The conclusion is that skill was the weapon, and skill is not hereditary. The wealthy prepared before they participated; most retail investors participated before they prepared. That ordering — not income, not connections — was the difference the data measured.

No one can, or will, hand you a twelve-month curriculum or a step-by-step system. The way cannot be given; it can only be taken — by those with the desire to go and seek it. The path back across this skill gap runs through a wide range of subjects and demands a mentality of steel, and no article can walk it for you. What an article can do is what this one has tried to do: turn on a light, and point at the door.

The rest is yours. And the encouraging truth — across both the traditions I was raised between — is that the door opens for those who genuinely go looking:

"Seek, and you shall find; knock, and it shall be opened unto you."

— Matthew 7:7

「千里之行，始於足下。」  "A journey of a thousand li begins beneath one’s feet."

— 老子《道德經》 (Laozi, Tao Te Ching)

The next bubble is already forming somewhere. The only question is which side of it you will be on when it breaks — and that question is answered not on the day it breaks, but in all the ordinary days you spend preparing before it does.

 

Data source: Li An, Dong Lou, and Donghui Shi, "Wealth Redistribution in Bubbles and Crashes," Journal of Monetary Economics, vol. 126 (2022), pp. 134–153. All figures cited are from the paper’s Tables 2, 3, and 5. (An earlier 2019 working-paper draft used slightly different group definitions and figures; this post follows the published version.)

For those who want to read the original paper:

For those readers who want to read the original paper:

The authors' own copy on Dong Lou's LSE page — this is the exact published version I pulled all the updated numbers from: https://personal.lse.ac.uk/loud/AnLouShi.pdf

Other access points:

·         SSRN (abstract + download): https://ssrn.com/abstract=3402254

·         LSE Research Online: http://eprints.lse.ac.uk/113766/

Personal Finance (22) The Lesson from Historical Charts — Market Can Fall, and It Will, Periodically...

Every market falls. Not “might fall.” Falls. Periodically. Without exception. 

Why I Am Writing This Now

Something is happening around the world that every experienced trader recognizes instantly, and every new investor cannot see at all.

Well, I came across the above news a few days ago... And, I recall another story that I read before:

Here is the old real story, nearly 100 years ago…

According to the legend, before the Wall Street Crash of 1929, Joseph Kennedy was having his shoes shined when the shoeshine boy enthusiastically began recommending stocks.

Kennedy supposedly thought:

"If a shoeshine boy knows enough to give stock tips, then everyone must already be in the market."

He then sold much of his stock portfolio before the crash.

So, I get AI, Fable 5, to give me some data for this post, and here they are:

The Record: No Asset Is Exempt

Finally, here comes the MOST IMPORTANT CHART.

What This Means — and What It Does Not Mean

This article is not trying to stop you from making quick money, nor trying to tell you to buy or sell, and it is not predicting a crash next month. Parabolic markets can run far longer than skeptics believe — that is exactly how they trap the most people at the top. I have no idea whether the AI rally ends in 2026, 2027, or later. Neither does anyone else.

Note:  All the Horizontal Histogram Charts are generated with Fable 5, while the Price Charts(in Log) plot by Amibroker, Data are from Yahoo Finance, which does not cover all historical period, except China Data from TongDaXing. 

Bless You

KHTang

Personal Finance (21): Benchmarking Warren Buffett's Berkshire Hathaway Inc. (BRK-A) Returns Against the S&P 500 Index (^GSPC) and SPY_ADJ (Adjusted for All Dividends) Over the Past Few Decades

 The attached table and charts are self-explanatory.

CAGR: Compound Annual Growth Rate (%)

Tickers:

BRK-A: Berkshire Hathaway Inc. Data available from 17 Mar 1980 (Yahoo Data)

^GSPC: S&P500 Index

SPY_Adj: S&P500 ETF with all the Dividend Restored, Starting date 15 Jan 1993 (YFinance)

(Click to Zoom in the Charts)

Apparently, BRK-A has shown no clear advantage over the S&P 500 index since 2008 — especially in the past few years. Is this because quantitative finance has already triumphed over the traditional wisdom of trading? Or, as rumors and news floating around suggest, is it because the wise man is holding a great deal of cash, waiting to buy at the next bottom?  ONLY TIME will tell — perhaps a year from now...

Self-Development Idea (35): (当代周处除三害)The Last Monster Is You: Zhou Chu and the Three Harms of Our Age

Every culture has its redemption stories, but few hit as hard as the tale of Zhou Chu. It's about hunting monsters—only to find the last one is you. And it isn't really an old story. It's happening again, right now, to all of us.

During the Jin Dynasty, the town of Yixing was tormented by what locals called the "Three Harms." Two were beasts: a tiger prowling the mountains and a flood dragon in the river. The third was a man—Zhou Chu himself, strong beyond reckoning, quick to violence, feared by everyone.

Too afraid to confront him, the villagers tried something clever. They praised his strength and begged him to slay the tiger and the dragon—secretly hoping all three terrors would destroy each other.

Zhou Chu agreed at once. He killed the tiger, then plunged into the river after the dragon, fighting three days and three nights until the water went still. He dragged himself home expecting a hero's welcome—and found the streets celebrating, because the people believed he had died.

In that instant, the truth landed: of the three harms, he had always been the worst.

He could have answered with rage. Instead, humbled, he sought out scholars, remade his life, and became a man of justice—slaying the third and final harm, his old self.

Now look at our own time. The shape repeats. Three harms are terrorizing ordinary people again. 

The first is a beast loose in the mountains: artificial intelligence. As it sharpens by the month, it swallows whole categories of work—jobs people trained years for, gone almost overnight.

The second is a dragon beneath the water: quantitative trading. Markets that once rewarded patience and common sense are now ruled by machines that calculate faster than any human can. For the ordinary investor, the river has turned against them.

The third is hardest to see, because—just as in Yixing—it lives inside us: the old map in our heads, a picture of the world that no longer matches the territory. We work harder, push longer, follow the routes our parents trusted. And still we get nowhere, because those roads no longer exist.

Like the villagers, we want to blame the beasts outside. But the legend already told us where the real battle is.

The modern Zhou Chu doesn't begin by fighting AI or the markets. He begins with himself—tearing up the old map and drawing a new one fit for this age.

He stops fearing AI and learns to walk beside it. He turns the first harm into an ally, and lets it help him build what the old world never offered an ordinary person: his own quantitative trading system—his answer to the second harm.

And in doing so, he slays the third harm: the outdated self. Exactly as Zhou Chu did seventeen centuries ago.

The monsters were never only outside. Neither was the victory.

Now, some of you are thinking this isn't realistic. So let me show you.

I asked an AI to write a quant trading system for my AmiBroker platform. Within minutes, it produced several—complete with backtesters and exploration programs. I ran the backtest, and here are the results:

This is a monthly rebalanced system built on Cross-Sectional Momentum Z-Score. A trend-following approach that you only touch once a month—monitor at the start of each month, take whatever action is needed, and walk away.

A word of honesty, though. I've marked the results in red, because most people couldn't stomach a drawdown that deep or a Sharpe ratio that low in a live trading system. And there's a second caveat worth stating plainly: the tested database doesn't include delisted stocks, which means survivorship bias is flattering these numbers.

But that was never the point. The point is this: the barrier to entry for understanding—and improving—a quant trading system is no longer years of study. With AI beside you, it's one step away.

And flawed as these results are, they're not a dead end—they're a starting line. A baseline to measure against, a benchmark to beat, a clear direction to push in. You're no longer staring at a blank page; you're standing on a vantage point, with somewhere to climb from here.

The dragon in the river hasn't disappeared. You've just learned to swim.