Intel and Gold Both Plummet While US Treasuries Rise! The Culprit Isn't Earnings Reports, It's the Dollar Pipeline! SOFR Inversion, Reverse Repo Hitting Bottom, Reserves Running Low, Hedge Funds Liquidating Positions—Even Selling Gold! The Drop Magnified 3.34 Times, Circuit Breakers Are No Longer a Zero Probability! The Flash Crash Is Just the First Manifestation of the Conflict Between the Fed and the Treasury!

Last weekend, a global massacre took place in the capital markets, with tech stocks, gold, and Bitcoin plunging in tandem while Treasury bond prices rose. By analyzing the overnight repo market, the spread between SOFR and IORB, reserve balances, and TGA account data, the host pointed out that the root cause of the crisis lies in US dollar liquidity tightening and the Treasury's debt issuance draining liquidity, which forced hedge funds to liquidate positions for cash. The flash crash revealed the policy conflict between the Federal Reserve and the US Department of the Treasury.

Video link:

Full transcript:

Observe things as they are, reject casual perceptions. Last week, the capital markets experienced a global massacre. I saw that there were quite a few confused friends in the comment section, so I discussed it with members of the badass gang and released the performance program from last weekend, hoping it can help clear up some of the confusion. As I said in the performance program released on Sunday, seeing the price inversion being so severe and the entire market structure floating like that, the first thing to think about is definitely not an aggressive or offensive strategy, but rather that it is time to pick up cheap insurance and protect the accounts. Also, it seems like it was the week before last, or maybe not last week's show, I put all the data in the program. The fucking national team of the Chinese Communist Party, who knows where that rumor came from, was dumping goods themselves, and there are still people playing in this fucking market. I really fucking don't know what they are thinking. Today, I will use data structures to tell you what actually happened last Friday. Right on last Friday, because

Intel once plummeted by 28.58%, which was the worst day for Intel since 1947. The Philadelphia Semiconductor Index also fell by 7.22%, the Nasdaq 100 ETF, which is QQQ, fell by 3.47%, and the S&P 500 fell by 1.82%, closing at 7580.07 points. Of course, tech stocks falling is due to poor earnings reports, which is understandable. But on that exact same day, spot gold also fell by 3.62%, closing at 3875.84 USD per ounce, which is the lowest price in a year. Silver was even more miserable, falling by nearly 6% and breaking below 50 USD. Bitcoin also broke below 77000 USD. At this point, everyone might be confused: safe-haven assets and risky assets jumping off a building in the same direction within a single day is not so easy to understand. Of course, if you cite news and explain gold with the Iran war and oil prices, and explain stock prices with poor earnings reports, it might seem understandable at first glance. But there is a larger market, the bond market, that disagrees.

Because on that same day, US Treasury yields across all maturities experienced a widespread plunge—meaning Treasury prices rose. This contradicts the surface-level trends of the Iran war and inflation expectations. Today, our goal is not to understand a single event, but rather to recognize a paradigm. When gold and the Nasdaq plummet in the same direction on the same day, the root cause is not in the earnings of the molecular end, but where? It is in the US dollar liquidity at the denominator end. To put it bluntly, it is not that no one wants to buy; it is that someone must get US dollars immediately. Today, I will break down this chain for everyone step by step. First, let us present the story on the surface completely. The story itself makes sense, but where does the problem lie? It lies in the other half that cannot be explained—which is precisely the Treasury bond part I just mentioned. Let us first briefly review the foreground of the process of this event. After the market closed last Wednesday, Google's parent company, Alphabet, released its third-quarter financial report. Google's single-quarter free cash flow has now been driven into a negative USD 9 billion, which can be said to be very, very rare. Historically, continuing from the previous second-quarter report which turned negative for the first time, this trend is still extending. However, Alphabet's capital expenditure guidance of USD 195 billion to USD 205 billion for this year has not been cut by a single penny, and it has clearly stated that it will continue to increase significantly next year. Then, Google's stock price fell by more than 10% the next day. Immediately following that, after the market closed on Thursday, October 29, Amazon released its third-quarter financial report. Amazon's free cash flow over the past 12 months was a negative USD 42.4 billion, and its full-year capital expenditure guidance of USD 220 billion was not cut by a single penny either. On the same night, Intel provided its fourth-quarter revenue guidance, with the figure being USD 15 billion to USD 16 billion, while Wall Street's expectation was USD 173 billion. Not only was it significantly worse than expected, but Intel itself also dealt another blow to its own stock price, stating that the global PC market would shrink by 12% this year. Up to this point, the story is still generally coherent. AI's capital expenditures have eaten into cash flow, and tech stocks need to be repriced. But when you look down another layer, things start to feel wrong. The S&P 500 equal-weight ETF, namely RSP, only fell by 0.79% on Friday, meaning that the average drop of each stock in the S&P 500 was less than half of the index's decline. For the entire month of October, the S&P 500 only fell by 1.2%, while the Dow Jones rose by 1.5%. Only the Nasdaq fell by 5.2%. This is not a recessionary market. So where did the money flowing out of tech stocks go? On Friday, someone was indeed buying Treasuries; the 10-year US Treasury yield dropped by 6 basis points to 4.93%, and real yields and the US dollar index were also falling. What are these two things? They are entirely a tailwind for gold prices. Yet gold unexpectedly plunged by 3.62%. With this price chain, the directions of outflow and inflow can be determined: money came out of tech stocks, did not enter gold, and even gold itself was being sold off. Then there is only one explanation left: someone is desperately short of cash. No matter what they are holding in their hands, anything that can be converted into US dollars is sold first. This is a script suffering from a liquidity short squeeze.

To see who is in a rush for US dollars, what must we look at? We must look at the most fundamental wholesale market for US dollars: the overnight repo market. Although we have already discussed this indicator many times, I still need to introduce it first. Let us start with SOFR, which is called the Secured Overnight Financing Rate. What does this thing measure? It measures the price at which financial institutions borrow money for a single night using US Treasuries as collateral. IORB is the Interest on Reserve Balances. What does this interest rate mean? It means the interest commercial banks get by keeping their money in that account at the Federal Reserve. As of the time I am making this program, it is 3.9%. The Federal Reserve also has a tool called the Standing Repo Facility, with the English acronym SRF. What is the purpose of this tool? It is a window where primary dealers and banks can go directly to the Federal Reserve to pledge Treasury bonds and borrow this overnight money. And this window carries a punitive nature, so this window price serves as the ceiling of the entire interest rate corridor, which is 4%.

I mentioned that under normal circumstances, banks and primary dealers are reluctant to knock on this window. Why? Because not only is the interest rate for borrowing this money high, but it might also invite a series of investigations from the Federal Reserve. Let me explain this mechanism so everyone can better understand why we need to watch this figure. I remember giving you all an observation dashboard before; SOFR minus IORB is the most important figure, bar none. Think about it: in a normal year, the interest rate for borrowing money outside is SOFR. It is collateralized, and the collateral is the highest-grade Treasury bond, so under normal circumstances, SOFR tracks closely with the Federal Reserve's interest on reserve balances (IORB). Think about it: money left at the Fed only earns an annualized rate of 3.9%, while the outside market is willing to pay 3.95%. Since the collateral is Treasury bonds, there is nothing to fear. Therefore, banks will immediately pull money out of the Fed to lend it out and earn that risk-free spread. Once the money floods out, the outside interest rate is immediately leveled. That is exactly how it works.

That is the principle, but look at the six consecutive days in late October. On Thursday, October 22, the SOFR was 3.9%, which was on par with the IORB. After that, it went higher day by day. On Tuesday, October 27, it jumped to 3.95%, a full five basis points higher than the IORB. Then, on October 28 and 29, it stayed stubbornly at 3.94%. This was five consecutive trading days of inversion. What is even more severe is the tail end: on October 27, the 99th percentile transaction rate for SOFR had already hit 4.04%, which is higher than the Federal Reserve's 4% window rate. What does this mean? It means that those most in need of money would rather pay a price more expensive than the window rate to borrow this money. The trading volume also tells the story: the overnight repo volume rose from 2.945 trillion on October 22 to 3.147 trillion on October 27, and hovered around 3.1 trillion for the following two days. And these days, it was neither the end of the quarter nor the end of the month. It indicates...

What does it indicate? It indicates that every day, 3.1 trillion US dollars in Treasury collateralized loans need to be rolled over, and the cost of rolling them over is getting more expensive by the day. This is the liquidity stress test we talked about last month. Why is this happening? Because money is being drained from three consecutive accounts on the liability side of the Federal Reserve. First, look at the overnight reverse repo, the ONRP; this is where money market funds park their idle cash at the Fed. On December 30, 2022, the peak of this account reached 2.55 trillion US dollars. Over the following three years, the Treasury mainly issued debt by buying it in advance from money market funds. At that time, bank reserves didn't move much because the overnight reverse repo cushion was still quite thick. By October 28 of this year, only 176 million US dollars remained in this account; the buffer was completely gone. So, from this day on, every time the Treasury sells Treasury bonds, it can only come from the reserves of commercial banks, and thus this pressure shifted to the Treasury's general account.

Yes, that is the TGA account. On Thursday, October 29, this account hit 1.04 trillion US dollars. The Treasury's quarterly refunding statement in August had already forecasted that it would hit around 1.05 trillion in late October. The process is as follows: the Treasury auctions Treasury bonds, and buyers must pay on the settlement day. How does the money flow at this time? Money is transferred from the buyer's bank account at the Federal Reserve into this TGA account. As a result, the reserve account decreases by one dollar, and the TGA account increases by one dollar—a one-to-one relationship. Only when the Treasury spends this money will it flow back into the reserve account. So, in the end, what takes the hit is the commercial bank reserves themselves. According to the H.4.1 weekly report released by the Federal Reserve on October 29—the weekly report on the Federal Reserve's balance sheet—the reserve balance has already fallen to 2.89 trillion US dollars. Although it has not yet pierced the 2.85 trillion "life-saving" critical value, it dropped by 74.3 billion US dollars within a week.

This is also a matter that deserves attention. The outside world often uses the comfortable range of 2.8 trillion to 3.2 trillion; where does this come from? Well, this actually has a formal name called the Lowest Comfortable Level of Reserves (LCLOR), which is estimated by Wall Street's money market teams. What does it calculate? They calculate a threshold even higher than mine. Their estimated lowest comfortable level is between 3 trillion and 3.07 trillion. Some friends might wonder, why is this number called "comfortable"? Because after falling below this line, the same withdrawal of 50 billion would have left interest rates unchanged before, but now interest rates would jump. Some might ask, since SOFR is already at 3.95%, why don't the big banks lend out these reserves? Can't you arbitrage by lending them out? No, because big banks have a bunch of regulators watching how much liquid assets they must hold at the close of business. When reserves reach 2.9 trillion, each bank is not far from this regulatory bottom line.

It's far, so they can only choose to hoard it and not move it. And the Federal Reserve's 4% window is not something you just knock on casually, because there is too much trouble involved, so under normal circumstances, big banks are also unwilling to knock on this window. Whoever knocks on this window first is equivalent to admitting to regulators and peers that they have no money. Consequently, this 3.1 trillion in repo rolling demand is all pressed onto the balance sheets of primary dealers. What will the dealers do at this time? Dealers can only raise the repo rate and the collateral haircut for hedge funds. At this point, hedge funds can no longer hold on, because there are two types of major players involved: one is a super fund with hundreds of trading teams tied to the same suture line, and the other is what? It is an arbitrage fund team that scrapes the price difference between spot and futures Treasury bonds. These two groups usually play in a high-leverage range of 20 to 30 times, but now overnight borrowing is both expensive and tight. To prevent

preventing liquidation, their choice is to cut positions first, meaning on one hand, they dump risky assets in their hands to raise cash, and on the other hand, to close short positions or grab the safest collateral, they frantically buy short-term Treasury bonds in the market at any cost, all to desperately protect their margin lines from being breached.

There is also a policy gap here. Previously, we mentioned that after the Federal Reserve ended quantitative tightening at the end of last November, starting in December of last year, it bought $40 billion of short-term Treasury bills per month to replenish reserves. This action is called reserve management purchases, which is the RMP tool. By this June, it dropped to $10 billion per month, and starting in mid-August, it was simply paused until mid-October. The two months when the Treasury Department was draining the TGA just happened to be the two months when the Fed stopped replenishing reserves, and the combined effect was that reserves broke through the red line. So, last week on the same day, financial reports were crashing and the plumbing was leaking—how much did each of the two events account for? We use a counterfactual model to say, hey, what is the purpose? It is to fix the impact intensity of this financial report and only change one variable: the spread of SOFR minus LRB. We look at how deep this index would be hit under different spreads for the same financial report. The sample consists of 250 financial report earnings shock days and liquidity shock days from January 2018 to October of this year. In the formula, R zero is when the plumbing is loosest, meaning what is the loosest situation for a pure baseline drop? It is a state where reverse repos are still above $2 trillion and reserves are above $3.5 trillion. Under this state, the earnings reports of Google, Amazon, and Intel combined would only be worth a 1.05% drop. And this MS is the amplification multiplier, and S is the spread. When the spread is below zero, this multiplier barely moves. Rising from minus 5 basis points to zero, the multiplier only inches up from 1 to 1.14, because in the formula, this 1 and this cube are negative pulling downward, while the square and the fourth power are negative times negative equaling positive pushing upward, so the positive and negative just cancel each other out. Once the spread crosses this zero axis, all four terms will superimpose at the same time. When the spread is positive 1 basis point, the first-order term accounts for 6% of the increment, which is simply the rise in financing costs. And when this spread reaches positive 2 to positive 3 basis points, the second-order term takes over and accounts for 35%, which is the acceleration of spread widening after primary dealers' balance sheets are full. And once the spread reaches positive 4 to positive 5 basis points, the third-order term accounts for 34.5% to 38.5%, and the fourth-order term accounts for 13.2% to 18.4%, with both terms combined exceeding half. What does the third-order term correspond to? The third-order term corresponds to dealers raising discount rates and adjusting client loans, while the fourth-order term corresponds to margin calls and market makers' mechanical dumping, as well as gold and silver being sold off for liquidity. Look at this, from October 27 to October 29, the weighted average of the spread over these three days was positive 4.5 basis points, and plugging that in, the multiplier is 3.34 times, so 1.05% multiplied by this 3.34 equals a 3.5% drop, while QQQ's actual trading on Friday was a drop of 3.47%, and everyone can ponder this error. Conversely, if reverse repos still had money and reserves were still above $3.2 trillion, then the exact same three financial reports would at most be a 1% consolidation. Let's look at this uncertainty again. The standard error is simply the scatter radius of the prediction bullseye; the larger the number, the more out of control the market is, and the easier it is to miss the target. The spread went from minus 5 to plus 5, and the standard error rose from 0.085% to 0.58%, a nearly 7-fold amplification. In other words, at positive 4.5 basis points, the point estimate is a 3.5% drop, while after considering negative skewness and fat tails for the 99% single-day extreme tail, the drop is 5.21%, and at positive 5 basis points it is 6.26%, while the U.S. stock market's level-one circuit breaker threshold is 7%. Why am I saying this? Because it heralds something—what does it mean? It means that above this spread level, a single-day circuit breaker is no longer a zero-probability event. The mechanical dumping by market makers in the fourth-order term can also find coordinates on the mechanical chain. We pull up the options that had not been closed out by the close of trading on October 30, and add together the market makers' gamma exposure according to strike prices. Gamma exposure simply means that for every 1% the index moves, market makers must buy or sell a certain amount of index futures or spot shares to hedge the risk in their hands. The S&P 500 index between 7,705 and 7,715 is the zero-gamma flip line. On Thursday, the S&P closed at 7,720.39, hey, still above the line. At the opening on Friday, it gapped down directly from 7,631 past this line, which is the process of how the liquidity shock was transmitted.

When below the flip line, market makers' net gamma is negative, so what can they only do? They can only chase ups and sell downs. The closing price of 7580 corresponds to a net gamma of negative 54 billion dollars, which means that for every additional 1% drop in the index from the closing position, market makers are forced by seasonality to dump 54 billion dollars worth of S&P futures and heavy-weight stocks. The intraday low on Friday was 7527.56, and further down, the 7500 level is the largest put option wall, where the existing net gamma of this wall is negative 48.5 billion. Then why did it rebound by 52 points after stopping at 7527.56? Because a large number of institutions holding 7550 and 7500 put options took profits and closed their positions at that level. At this point, the market makers' short put options were hedged away accordingly, and the market makers bought back the short futures positions they had previously established. It was this buying that pulled the broader market back from the edge of the cliff.

You mentioned that the decline in tech stocks includes factors from earnings reports, but gold has no earnings reports and is not affected by that part, so in the case of gold, we can isolate the macroeconomic component. So what is left? It is the purest form of liquidity. More than 80% of gold's daily price fluctuations can be explained by two factors: one is the real yield of 10-year Treasury Inflation-Protected Securities (TIPS), and the other is the US Dollar Index. How did these two factors move on Friday? The real yield of 10-year TIPS fell by 3 to 4 basis points from 2.63% on October 29 to around 2.59%, while the US Dollar Index fell by 0.15%. In other words, both were tailwinds for gold prices. Based on historical elasticity, the theoretical fair gain for gold on Friday should have been around positive 0.5%, corresponding to a fair price of around $4,042 per ounce. However, the actual spot gold closed at $3,875, down 3.62%. GLD, the world's largest gold ETF, fell 3.47% to close at $355.78. This means there was a huge gap of $166 per ounce between the actual market price and the fair value. I measured it using the standard deviation of gold's daily residuals, and it was negative 2.86 standard deviations. In other words, the probability of this happening under normal circumstances is less than 0.5%. So how was gold hammered down within a single day on Friday? I pulled the tick-by-tick transaction data for GLD and looked at it. At 9:30 AM Eastern Time, the market opened. At this point, the centralized risk control of the multi-strategy platform saw, holy shit, the repo rate was still set at 39.4%. Then, with tech stocks plunging in pre-market trading, the portfolio volatility and total leverage were in a state of alert. At this moment, the risk control algorithm would do one thing: indiscriminate deleveraging. Sell the most liquid assets first, and gold is the most liquid one. GLD was hammered from $368.58 in pre-market trading down to $364.20 by 11:40 AM. This happened to coincide with the London Gold Fix. As major European and American traders tightened their overseas dollar borrowing quotas, arbitrage funds had no choice but to sell gold and exchange it back for dollars before the London market closed, which then hammered GLD down to $358.90. Where was it truly brutal? It was at 3:45 PM in the afternoon. The prime brokers on Wall Street—the investment banking departments that lend money to hedge funds—have a deadline for daily margin calls that is set 15 minutes before the market closes. Those funds that could not borrow money in the repo market were also out of options and had no choice but to place sell orders regardless of cost between 3:40 PM and 3:45 PM, frantically dumping the GLD they held. The goal was to get dollars to fill the margin holes in their stock accounts. Therefore, GLD hit its daily low of $354.67 exactly at 3:45 PM. Once this wave of forced liquidation selling stopped, it immediately rebounded by $1.10 in the final 15 minutes, closing at $355.78.

Once we understand the root cause of this problem, we can determine whether the market can stop falling this week. What should we look at directly? We should look at whether the Federal Reserve and the Treasury Department take action. The earliest data available is the SOFR released by the New York Fed at 8:00 AM yesterday, but that was the figure for October 30th, which was still at 3.94%. What does this indicate? It indicates that even after the end of the month, there was no easing. Next, we need to pay attention to another data point, the Standing Repo Facility (SRF) usage. With SOFR inverted like this, the daily usage of SRF is still zero, because the median of SOFR is still five basis points away from the 4% window price, and large banks are afraid of the stigma of hitting the window. More importantly, the hedge funds that are truly short of cash are not qualified counterparties for the SRF. Therefore, when the daily usage of SRF breaks through 15 billion to 30 billion, or the New York Fed lowers the SRF interest rate by 5 to 10 basis points, what does this mean? It means that the Federal Reserve's interest rate ceiling has begun to take effect. Furthermore, there is another pressure this week: the 69 billion in two-year Treasury notes and 70 billion in five-year Treasury notes auctioned this week need to be settled and paid for. This is another round of money moving from reserves to the TGA, or let's call it a pressure. Therefore, I think there is a high probability that the interest rate spread will be pushed up again yesterday. However, this figure has not been released yet. What should we look at to see if the liquidity can really be released back? We mainly look at tomorrow, which is the Treasury Department's quarterly refunding meeting. The Treasury's current year-end cash balance assumption is 850 billion. If it is lowered to around 700 billion, then 150 billion to 300 billion will flow back into the market over the next month, flowing back into bank reserves through government spending. At the same meeting, the scale of Treasury bond repurchases will also be determined. The Treasury Department is supposed to use 38 billion USD each quarter to buy back old Treasury bonds from the primary dealer market, as we have mentioned before. But faced with a basis trade stock of 600 billion to 1 trillion, this amount is simply not enough. Then, the Treasury Department will have to look at the Federal Reserve's attitude, specifically whether the Fed's reserve management purchase tool will be restarted. If it is restarted, the monthly scale is estimated to be around 30 billion to 50 billion, pushing the reserves back to between 3.15 trillion and 3.25 trillion. This is the fundamental way to push the interest rate spread back below zero. By the way, let's take a look at the foreign central bank repo facility, which is the FIMA, the window where foreign central banks exchange US Treasuries for US dollars with the Fed. If the weekly usage jumps directly from zero to 5 billion to 10 billion, it indicates that the repo crisis has spilled over into an offshore dollar shortage.

The real difficulty lies ahead. The Federal Reserve just raised interest rates by 25 basis points on September 16th, and the core PCE inflation, which is the price indicator the Fed monitors, has been above 3% every month this year, so the Fed is also having a very difficult time. To make matters worse, the market is still betting on another rate hike in December. However, you have to treat the illness in the repo market. If you want to treat it, you have to buy short-term bonds again and expand the balance sheet. If this plan is restarted, Walsh will face pressure: on one hand, he says he wants to tighten liquidity to combat inflation, and on the other hand, he has to print money to buy short-term Treasury bills to replenish reserves. Fed Chair Walsh cannot justify these two things simultaneously. It is the Treasury Department that stored 1 trillion in the TGA, and it is also the Treasury Department that is frantically issuing short-term debt to borrow money, expecting the Fed to take over. The Fed wants to tighten monetary policy to curb inflation, but the Treasury is forcing it to print money to provide a backstop by frantically issuing debt, which creates an irreconcilable contradiction. So what is the situation now? It is either that monetary policy is forced to give way to government borrowing, or the Fed stands its ground and ignores any fiscal demands. The middle path, which is also the most probable, is to provide emergency relief first. The way to provide relief is for the Fed to resume the purchase of short-term bonds without abandoning its long-term monetary policy goals. In fact, the flash crash on Friday was just the first time this contradiction manifested on the market. What we need to watch next is the refunding meeting tomorrow, and what we are looking at is not the numbers, but which of these two parties will concede first, and to what extent. Therefore, if you have positions in your hands now, you must ensure you are protected.