SI423: Why Trend Following Might Outlast the AI Arms Race ft. Mark Rzepczynski

Mark Rzepczynski · AMPHI Research and Trading

Bond yields are switching regime after decades of decline, so the bond market is now two-way.

Yields peaked with double-digit US inflation in 1981, then fell for decades in a secular rally. Mark argues that the prior view that trend followers could not profit from rising rates was an artifact of that long bond rally. Yields are now switching back, and the market is more two-way, so shorting bonds can make money, which would have seemed anomalous ten years ago.

“we're seeing that the market is more two way and you can make money shorting bonds which which that was something that ten years ago, we'd say, well, that can't happen.”at 11:26

Mark Rzepczynski · AMPHI Research and Trading

The current bond selloff is normal: yields are near fair value given growth, inflation and term premium.

Mark builds fair value from first principles: nominal rate = real rate + expected inflation + term premium. The real rate is roughly long-run growth plus productivity, about 2-2.5%, and the Atlanta Fed GDPNow for Q3 is above 3%, which supports a growth and productivity boom. Current inflation is above 3% and has been above the 2% target for over six years. Adding that inflation gives about 5-5.5% before any term premium. Term premium should also be positive now because the stock-bond correlation has turned positive, which raises bond volatility. Recent auctions near 5.5% therefore look close to where the underlying economics put yields, so the selloff is a return to normal rather than an oddity. Private-market demand for debt may crowd out public-market issuance.

“So it's not that we have an oddity, it's that this is normal.”at 42:13

Mark Rzepczynski · AMPHI Research and Trading

Hedge funds' basis-trade bond demand is more price-sensitive, so bond moves could go either way.

Hedge funds are now much more active in bond markets through basis trading and financing, and government buyers such as QE are less important. Mark's guess is that hedge-fund demand for bonds is far more elastic than government demand. Multi-strategy funds cut positions after even modest trading losses, so a change in their expectations could trigger a large move. Mark's concern is less that bonds are far from fair value and more that the composition of debt buyers has changed, which raises the chance of a big fallout or a big rally.

“my guess is that the elasticity of demand for bonds from hedge fund is much greater than what you're gonna find from the government... we could have a big fallout or a big rally. It could go both ways.”at 46:06

Mark Rzepczynski · AMPHI Research and Trading

Rising Treasury supply is eroding the convenience yield, already gone at the long end.

The convenience yield is the premium investors pay for a liquid, deep, safe asset. It rises in crises when safe-asset demand jumps. As the government keeps issuing more Treasury debt, the supply of the perceived safe asset grows and the convenience yield shrinks. Mark says this advantage has already disappeared at the long end of the Treasury curve and survives mostly on the short end, in T-bills. He suggests this is one reason Treasury is issuing more bills, because that is where the convenience yield sits.

“the government such as US Treasury continues to issue more and more treasury debt, then the supply of what is perceived as the safe asset increases by more and more. And that means the convenience yield is gonna be shrunk.”at 48:29

Niels Kaastrup-Larsen · Top Traders Unplugged

Equity gains are earnings-driven while multiples contract, and the buyback tailwind is fading.

Niels says the 'FIMO' (earnings momentum, measured by actual earnings per share) has been soaring while FOMO, reflected in P/E multiples, has been contracting. He is sceptical that the way some companies account for higher valuations, with the gains showing up in earnings, can hold up indefinitely. He also argues that the buybacks that supported equities over the past few years are coming to an end and that new equity supply is rising, so competition for capital is fierce.

“FIMO is measured, by actual earnings per share, and that's been soaring... FOMO... reflected in the [P/E] multiples, and they've been contracting a little bit.”at 5:20

Mark Rzepczynski · AMPHI Research and Trading

Stocks are rising despite higher rates only because earnings growth has outpaced discount-rate pressure.

Using a dividend discount model, price is earnings divided by the discount rate, and the discount rate cannot fall below Treasuries plus a credit spread. Discount rates are rising with interest rates, yet stocks keep climbing, which Mark explains by earnings growth exceeding the discount-rate drag. He argues that once earnings stop growing and earnings momentum is lost, the discount rate becomes the dominant force, and that is where the real fear comes in.

“But we see that the discount rates are all going higher because interest rates are going higher, yet stocks are still going up... Once earnings stop to increase, you lose earnings momentum, then... the discount rate will then become more dominant.”at 6:13

Mark Rzepczynski · AMPHI Research and Trading

Commodities may be entering another super cycle, driven by underinvestment and a grind higher in prices.

Commodity super cycles have recurred since the 1980s and are treated as major dislocations rather than crises. Mark argues another may be starting because periods of underinvestment in commodity markets mean less capital expenditure. That shortfall should produce a grind of higher prices across several markets until a reversal, as happened around 2007-2008 at the end of the last super cycle.

“we may be in another super cycle again because what we'll find out is that there are periods where there's underinvestment in commodities markets, so there's less capital expenditures than there's gonna be, you know, a grind of prices moving higher across a number of different markets”at 36:14

Mark Rzepczynski · AMPHI Research and Trading

Low inventories, not production, drive commodity spikes; a cold Europe winter risks a stock-out squeeze.

Mark argues that in commodities, the key risk is inventory levels rather than output. A stock-out is the danger: a producer with no inputs pays any price to avoid shutting down. Europe's winter inventories are below expectations and have been low for five years. A cold winter could therefore cause a stock-out. Hoarding then amplifies the move, pushing prices parabolic. Markets with few close substitutes, such as diesel for diesel trucks, are most prone to these nonlinear spikes.

“the real problem in commodities is not what is produced, but it's what the inventories look like at any given time... If you have a real cold Europe winter, you're gonna have a stock [out] problem.”at 58:03

Niels Kaastrup-Larsen · Top Traders Unplugged

French sovereign debt is being stress-tested, with 10-year yields near 5%, reminiscent of Eurozone crisis.

The 10-year French yield is about 5%, up roughly 165bp over the last twelve months and about 50bp in the last month, a 24-year high. Niels compares this to the Eurozone sovereign crisis in the South and argues that large countries whose sovereign debt is being tested can see this kind of stress. He says it is hard to see how the situation goes away once it reaches that stage, so markets may have to deal with it for a while longer.

“I think we're trading now the ten year is just about 5% now in France. That's up 1.65% in yield in the last twelve months... it is reminiscent a little bit about some of the crisis we've had in Euroland”at 2:35

Mark Rzepczynski · AMPHI Research and Trading

Shortage of bank intermediation capital raises crisis and market dislocation risk.

Primary dealers, who are banks, warehouse risk between buyers and sellers. If they are capital constrained, through repo market problems or balance-sheet restrictions on risky assets, they cannot hold positions. Sellers then have to find buyers extremely quickly and in a disorderly way, which makes dislocations more likely. Mark says commodity firms act as dealers too, so the same dynamic can occur there, and in uncertain, illiquid markets participants scale in and out of positions in ways that can produce trends.

“if there's a shortage of... intermediation capital... Then we're gonna be more susceptible to a crisis.”at 22:50

Mark Rzepczynski · AMPHI Research and Trading

US inflation will not return to 2% near term; it has run above target for over six years.

Mark argues the talk of inflation returning to 2% does not match the data. Current inflation is above 3%, and inflation has been above the 2% target for well over six years. Even if inflation eases slightly, it probably will not fall back to target in the near term. He uses this inflation level, plus a roughly 2-2.5% real growth rate, to derive a nominal yield near 5-5.5%.

“We've been above the 2% target for now, well over six years. So we're not gonna probably see it in the next... In near term even if we could see inflation abate slightly.”at 40:51

Mark Rzepczynski · AMPHI Research and Trading

US public debt-to-GDP is rising because budgets are not under control; interest costs are crowding out spending.

Private markets and consumers have cleaned up their balance sheets and reduced leverage, judged by BIS credit-to-GDP early-warning measures. Public debt is still rising. Mark says the US and France have not got their budgets under control. Countercyclical deficits in recessions should be followed by primary surpluses, but those are not occurring. On top of primary deficits, the cost of debt is becoming a major part of the federal budget and dwarfs many other spending categories.

“overall debt to GDP ratios are increasing for the simple reason is that they haven't really sort of got their budgets under control.”at 43:33

Mark Rzepczynski · AMPHI Research and Trading

The dollar is still dominant, but its reserve regime is breaking as central banks diversify.

The dollar is still the dominant currency. On the margin, however, central banks have reduced their dollar exposure. Mark sees a move toward multipolarity, with reserve managers holding a range of alternatives. He ties this to Treasury convenience yields: the reserve-currency convenience yield may not be in jeopardy, but on a relative basis it is being diminished. He cites Ken Rogoff's earlier appearance on the show as a reference.

“Yes. But on the margin, what we're seeing is that people have reduced their exposure, central bank holdings have reduced their exposure to the dollar and what we're seeing is that the dollar regime is breaking and there's more of multi polarity from central bank reserve managers to hold a number of different alternatives.”at 49:31

Mark Rzepczynski · AMPHI Research and Trading

Geoeconomic fragmentation from tariffs and sanctions should raise cross-market diversification.

Mark argues that tariff wars, sanctions, capital controls and industrial policy are fragmenting the world economy. Under that fragmentation, markets should decouple, which should raise diversification across markets. He adds that the amount of geoeconomic fragmentation will ebb and flow. When fragmentation falls, markets become more correlated and diversification shrinks. Trend followers therefore need to track geoeconomics because it changes the correlation matrix their models use to diversify.

“if the world becomes more fragmented and dislocated, we should sort of see the diversification increases in markets. Markets are gonna become more decoupled.”at 54:26

Mark Rzepczynski · AMPHI Research and Trading

Machine-learning crowding is shortening alpha half-lives and making hedge fund positions more similar.

Survey work suggests most hedge funds use machine learning to find alpha. Quants trained at the same schools use the same well-known techniques, such as transformers and boosting. As more funds use these methods, models adapt and alpha is crowded out faster, shortening its half-life. Strategies then migrate to the next model, which speeds up market adjustment and forces firms to spend more on computing to keep pace, a Red Queen problem. Mark cites 13F filings showing growing similarity in hedge-fund positions in major stocks, which he describes as strategies eating their own. Trend following is different because trends are not a stable half-life, though it still can use machine learning.

“all these strategies that have used more and more non trend following techniques, machine learning techniques that actually they're... we'll call it they're eating their own.”at 1:06:15