On US repo market

Recently we have seen news in WSJ and FT about the "turbulences" in US repo market after years of relative calm, manifested by widening of the SOFR / EFFR spread (fig. 1) and resulting FED's temporary OMO intervention.

 
I will try to balance the "breaking news" by making this post more educational, explaining why this is happening and why I think the reasons are more prosaic than the fast news suggest.

First, let's define the relevant rates in question:

🔵EFFR (Effective Fed Funds Rate) is the rate at which banks lend money overnight to each other without any collateral.

🔴SOFR (Secured Overnight Financing Rate) is used by many more institutions (not just banks). The loans are made against collateral (typically Treasury bonds). This distinction will become important later.

Because EFFR is unsecured, it is expected to be slightly above SOFR due to a small credit risk premium. This is a textbook knowledge and has indeed been empirically confirmed (2021-2024).

However, in the second half of 2025, SOFR started being jumpy and rose above the EFFR. This stirred the Fed's reaction (fig. 2) and the aforementioned news.

💡Why is SOFR jumpy now?

During quantitative easing, the FED supplied market with plenty of liquidity by buying bonds from the Treasury Department and other participants. Once there is a lender (e.g. the FED), finding borrowers (Treasury and others) is rarely a problem. The result of this dynamics is a smooth, uneventful EFFR-SOFR spread.

However, everyone who has ever borrowed money knows that it is much harder to quickly find another lender to step in, when the original one demands to be paid back. After years of quantitative easing, FED started a period of quantitative tightening. Cash is being drained from the system and flooded instead with treasury bonds.

In other words, there is a scarcity of cash but plenty of collateral. More Treasuries on the market also mean more demand for repo financing (e.g. hedge funds buying bonds and selling bond futures, therefore funding bond purchases by borrowing at SOFR).

💡Why is EFFR not affected?

Remember what we said previously - the FED Funds rate is unsecured and therefore is not affected by the aforementioned repo market dynamics.

💡What is the significance of all of this?

The important thing is the quantitative easing or tightening itself, because it profoundly affects money supply, inflation, employment and long-term economic prospects. The daily noise in the repo market is just a natural second-order effect of reducing FED balance sheet. In grand scale of things, those wobbles don't mean much if managed properly.

The central bank is aiming to smooth this by applying temporary open-market ops, for the first time since mid-2020. This fact alone may indeed sound intriguing. However, in the 2010s, such operations were occuring more frequently (fig. 4).

On "Voice Trading comeback"

FT recently published an interesting article on the "voice trading comeback" in the Treasury bond market. The article claims that reason behind this comeback is the increased demand for Treasury basis trades, which are hard to execute electronically. However, it stops short of explaining why. In this post, I will try to fill that gap.

Trading of Treasury bond basis goes back to the late 70s. However, it became popular again in late 2023, driven by rising Treasury issuance and structural demand for futures in a high interest rate environment. You may have noticed a few articles about Treasury basis occasionally popping up here and there.

🟩 How to trade (long) treasury basis ?

• Borrowing cash from the bank via repo transaction
• Buying bonds in the cash market and simultaneously posting them as collateral in the repo market
• Selling bond futures.
• On the delivery date, the cash bond will be delivered into the short bond futures contract
• Invoice proceeds from the short bond futures contract will be used to "repurchase" the bond at the maturity of repo transaction.

🟥 How to trade short treasury basis ?

• Similar to the above, but instead of borrowing cash, one borrows bonds via a reverse-repo transaction and sells them in the cash market.
• For your bank to be able to lend you bonds, it needs to have them on its balance sheet first. And the balance sheet is a scarce resource.

🟧 Why it is traded by voice ?

While trading of equities, FX, or options has been largely "democratized", you cannot simply execute a Treasury basis trade via your Interactive Brokers account.

Buying and selling the basis always involves borrowing cash and borrowing bonds in a highly leveraged manner via repo and reverse-repo transactions. That means access to cheap repo funding and the balance sheet of Treasury bonds is a critical component of the trade, which differentiates winners from losers.

Being a member of the earlier group typically requires a seat in a hedge fund with an established relationship with a bank willing to grant access to its balance sheet available for repo trading. This balance sheet is a limited resource, carefully allocated to the most valuable buy-side clients. This is the reason why this type of trading is done by voice and not electronically.

📙 One of the best books on Treasury bond basis is by Galen D. Burghardt et al., with the same name. This kind of trading is not for everyone for the aforementioned reasons. However, the book certainly is.

On euro swaps, dutch pensions and "the most important thing"

If you follow markets, you may have noticed recent steepening of EUR swap curve on long end. This topic is now getting traction and everyone started paying attention to what is about to happen next. But why?

It all boils down to the Netherlands passing a law called the "Future Pensions Act" more than a year ago. According to this law, Dutch pension funds will gradually abandon defined benefits (DB) and move towards defined contribution (DC) pensions. The largest funds, including BpfBOUW, covering almost 1 million workers, agreed to do so starting 1 January 2026. If you live and work in the Netherlands, you most likely already know.

Defined benefit pension funds typically have floating-rate assets and fixed-rate liabilities (aka defined benefits). The duration mismatch between them is usually bridged by buying lots of (fixed-rate) receiver swaps.

Removing these fixed-rate liabilities will shift demand from institutional buy-side clients from receiver towards fixed-rate payer swaps, especially in the long end of the curve.

What happens to the 30Y swap points when demand for these receivers by pension funds is gone?

Many players have already started betting that 30Y swap rates will go up and have begun positioning themselves by entering into 10s30s curve steepeners.

Bidding up the fixed rate in the 30Y region is currently being anticipated by sell-side and hedge funds, who are trying to get ahead of the January 2026.

☝Last but not least, if you are aiming to capitalize on this idea, think of the most important thing: "second-level thinking." As Howard Marks explains in the chapter of his book by the same name, being right when everyone else is right too often yields no financial gain. That may well be the case with this idea.

Further things to consider:

1️⃣ Current shape of EUR swap curve + peers
2️⃣ Historical evolution of 10y30s basis, compared to USD and GBP
3️⃣ Current and historical basis between swap market and treasuries

⚠️ Disclaimer: This is not a trade idea, investment advice, or product/service solicitation. Opinions are my own

On most recent rate cut (It's a surprise to have a surprise)

Each time the FED raises or cuts the policy rate, it is printed on the front page of major news outlets. However, it is rarely breaking news itself. By design, the FOMC makes sure to give the market enough transparency and hints, so the cut is largely anticipated ahead of time.

The consensus in the US is usually strong enough to make even ChatGPT hallucinate in the past tense when enquired about an upcoming rate cut.

However, when it comes to other countries, transparency is less of a norm, and sometimes the policy change catches the market by surprise. This is often true, for example, of Central European currencies or South Africa.

There are dedicated services such as CME's FedWatch that estimate decision probabilities. The simple framework that works best for me is to look into spreads between the current RFR and forward-looking swap rates for various tenors (Fig 1). I like this approach because it is simple, can be applied universally to any currency (not just USD), and doesn't depend on proprietary probability models.

The attached example shows that on Wednesday, 29th October, just hours prior to the FED announcement, the SOFR fixing to the 2W SOFR IRS was -34 bp, fairly high by this year's standards.

It is worth noting that this methodology combines the magnitude and certainty of a rate jump into a single number. Looking into swap tenors longer than 1M encapsulates the effect of multiple rate announcement.

🔔 Follow me for more on markets, banking and quantitative finance
⚠️ Personal opinions only, no investment advice or product solicitation.

On FX Carry (Part 2)

Previously, we explored basic mechanics of FX carry trades and reasons behind the existence of the risk premiums. Carry trades in managed currencies can be profitable, yet inherently very risky (think of fat-tail events).

In this post, we will use a carry/vol framework to assess viability of carry trades in G20 currencies (see attached table). We are looking for currencies that exhibit high IR differential with a low probability of adverse FX movements. This probability can be analyzed in multiple ways:

1️⃣ Historical and Implied FX Volatility

Historical FX volatility is a go-to measure for assessing the risk of adverse currency movement. The drawback is that it is backward-looking and not well suited for pegged/managed currencies or currencies with large spot/vol correlations (which are typical for emerging markets). An alternative measure is volatility implied from FX options, because incorporates forward-looking market consensus.

2️⃣ Risk Reversals

Positive risk reversals indicate that currency is likely to experience high volatility, if it depreciates. Risk reversals tend to be higher for currencies with asymmetric, event-driven, or political risks. This applies to both managed and free-floating currencies.

🟥Managed Currencies

Pegged (AED, SAR, HKD) or managed (TWD, CNY, SGD, MYR, INR, and to some extent also JPY) currencies typically offer lower volatility when measured using historical data or implied from ATM option volatilities. This can sometimes mislead traders into a false sense of security when executing carry trades.

It is not a coincidence that most recent headlines were made in managed currencies such as JPY and TWD.

(See graph showing managed currencies in red and free-floating in green)

🟩Free-Floating Currencies

Free-floating currencies such as GBP, AUD, or EUR offer less potential for profitable carry trades, since their central banks manage interest rates in a fashion resembling synchronized swimming, leaving their FX movements to market forces. On a flip side, these currencies exhibit less tail risk.

🧨FX Carry and Political Risk

While currencies like the Turkish lira (TRY), Russian ruble (RUB), and Brazilian real (BRL) may initially appear attractive for carry trades, they carry significant political risk that can undermine their appeal. In particular, the TRY is prone to persistent depreciation, driven by the country's high inflation environment. As a result, purchasing power parity considerations are essential when forecasting its future trajectory.

Attached figure shows evolution of TRY between Oct 2024 and Oct 2025. The actual depreciation is around 20% due to inflation, while the interest rate differential is above 30%.

It is worth noting that the resulting 10% premium exists for a good reason - FX returns are not normally distributed, and there is significant fat-tail risk due to political events. For example, consider the 19/3/2025 arrest of Istanbul's Mayor E. İmamoğlu, which resulted in a 5% drop in value within a single day.

⚠️Disclaimer: This is personal post, no investment advice. Please read full disclaimer in the footer.

On FX Carry

Carry strategies are as old as the markets themselves, and almost everyone at some point has come up with the idea of "borrowing low and investing high".

Macro hedge funds love this idea too, mostly because of its ability to deliver solid market-neutral returns with low Beta and vol. The drawback comes with the tail risks, and the main skill in running this strategy lies in the ability to avoid and manage catastrophic events. More on that later.

💡The idea is simple:

- Borrow money in a low-interest rate currency
- Invest in a high-interest rate currency
- Profit from the interest rate differential
- Avoid periods of high volatility and regime changes

🌐Risk-neutral and risky worlds

In the risk-neutral "Q" world, expected future FX rates are equal to today's FX forward rates. Therefore, one is expected to earn a risk-free rate with no risk simply by means of the covered interest rate parity - buying FX spots and selling FX forwards.

However, we are not living in a risk-neutral world. By not covering the risk exposure, one enters the real world of risk and the "P" pricing. In this world, FX rates follow paths prescribed by trade balance, NIIP and other macroeconomic relationships.

FX risk is the reason why people are willing to flock into low-interest rate currencies (the risk buyers) and why others are willing to exploit the rate differential by engaging in the carry trades (the risk sellers).

💵 Taiwanese Dollar Example

For example, consider USD/TWD chart between May 2024 and May 2025. Quarterly deposits on USD were trading in range of 450-500 bp, while Taiwanese were lingering around 160-170 bp.

For Taiwanese investors (mostly life insurance companies), it was far too attractive to miss the opportunity to convert TWD into USD and invest in a high-yielding currency abroad rather than at home.

Red line (🟥) shows the term structure of USD/TWD FX forwards as of May 2024, predicting the currency path in a "risk-neutral" world (Q), where market risk is not priced in. In such a world, one would expect to buy USD at 32.474 and sell it one year later at 31.166, realizing an FX loss on USD, offset by its higher deposit rate.

However, that's not what happened in the "real world" (P). Thanks to the Taiwanese semiconductor industry and a positive trade balance with the US, the Taiwanese dollar enjoyed levels indicated by the blue line (🟦), above and beyond what was "deserved" by the interest rate differential.

Surplus on Taiwanese capital account was offset by deficit on investment account, thanks to all the lifers allocating capital to US Treasuries. This created tension in the investment position, which ultimately snapped in May 2025, resulting in a rapid unwinding of carry trades.

Another example is the JPY carry trade, which began in the early 1990s when the Bank of Japan slashed interest rates amid concerns over long-term economic growth. This strategy has remained popular ever since, though it occasionally backfires. Most recently, it made headlines in August 2024 when the Japanese yen experienced significant strengtening.

And that's the moral of today's story - those who are attracted by superior risk-adjusted returns on carry trades must be prepared for hidden fat-tail risks.

On end-of-month interest rate turns

When trading interest rates, most people are familiar with instruments such as deposits, FRAs, futures, or swaps. These inputs are directly observable in the market, e.g. via Bloomberg or Reuters feeds.

However, there are instances where rates traders must apply discretion when constructing the yield curve. In such cases, they often estimate certain inputs using statistical or judgment-based approaches. One such input is the concept of "Turns".

What are turns?

Turns represent a trader's view that the anticipated overnight rate on a specific future date, typically at month-end or year-end will be X basis points higher or lower than on regular trading days. There is empirical evidence supporting the existence of turns, as illustrated by historical rate data.

Why do Turns exist?

Turns exists due to a variety of factors, which differ from one economy to another. For example, the presence and magnitude of turns may depend on whether the economy is rates-driven (like G5 countries) or FX-driven (such as export-oriented city-states like Singapore or Hong Kong SAR).

Take, for example, a chart of the EUR €STR rate (attached). This rate regularly spikes downward toward month ends. In Europe, €STR represents the rate at which a large number of banks are able to lend directly to each other at overnight unsecured rates.

At month ends, unsecured lending is either reduced or shifted toward the secured repo market to improve balance sheets (window dressing). This causes the perceived spikes shown in the chart.

In fact, there is empirical evidence of turns taking opposite sides on secured and unsecured rates (see the attached table).

Turns on secured and unsecured markets

What happens if turns are ignored?

IRS trading is the bread and butter of many investment banks. Capital is always tight, and trading spreads are competitive. A cash flow on 31 December 2030 will clearly have a different value than on 1 January 2031. Failure to recognize this when making IRS markets will clearly lead to suboptimal decisions, which will accumulate over time and hurt P&L.

Which markets exhibit the most significant turns?

One of the most notorious examples of money markets with turns is the HONIA rate in Hong Kong, where turns regularly print in high double digits.

HKD is pegged to USD, and there is strong demand to hold USD for end-of-month reporting purposes by a wide range of market participants. Through the effects of LERS and HKMA operations, this results in the HKD HONIA rate often spiking upwards.

Hong Kong

Other markets that experience strong turns are Singapore (SORA), Korea (KOFR), Poland (WIRON), and Mexico (F-TIIE).

Mexico

Korea

Poland