Digesting the DeepSeek Asset Shock
My initial reaction to the DeepSeek asset shock on Monday was to see the stocks down/bonds up reaction as classic “flight to safety”. Many of asset flows are systematic and happen because signals like VIX are embeddded into trading and risk management systems. CTAs, for example, add leverage when VIX declines and decrease leverage when VIX advances.
I’ve had some time to further digest the DeepSeek catalyst and intermarket moves on Monday. DeepSeek can be seen interpreted as a hugely disinflationary signal, ie. the cost curve of the development and runtime of compute environments for AI applications has just collapsed.
It means growth expectations should increase and inflation expectations decrease. Growth increases because both the production and consumption of AI applications become more accessible. Remember economic output is measured of as the sum of transactions in the economy, and economic throughput could ramp with the aid of AI. Think about what has happened as high-speed broadband. Internet has become cheap and abundant; there would be no Web 2.0 and no "web apps", Zoom or work from home.
Inflation should decrease because AI applications are destined to separate humans from their employment; the next decade will be the story of immense technology induced pressure on wages which will be countenanced by politics that protect and preserve the value of voters, I mean labour. We haven't seen anything like this in our lifetime; it is analogous to the Industrial Revolution and advent of steam powered factory plant that could perform mechanical labour at scale.
The disinflationary impulse from AI is the biggest threat to the growth outlook. I assume that with some friction and lag displaced people will find themselves in new careers. Government programs and the private market will move to facilitate the reeeployment of human capital. Over the medium term destroyed jobs != destroyed income. Lost income will be transitory, a necessary friction within the context of a sweeping economic revolution that has huge ramifications for productivity ergo wealth creation. Remember in the long run growth = demographics + productivity.
Back to Bonds
We are testing a key area in bonds, and how the market responds has me contemplating a flip in my directional conviction on bonds.
Prior to the DeepSeek shock, Bleeding Edge was on record as saying that with the Fed firmly on pause, yields were on an endogenous path which should see long rates continue to rise until this tightening of financial conditions causes a slowing in the economy, and consequently a halt to the rise of risk assets like equities.
Inflation has been stickiest in services and shelter. And whilst DeepSeek (and many progeny, soon to follow) will do nothing to ease shelter inflation, it could well smash services inflation over the medium term. Markets are forward looking and will begin to discount this impact.
DeepSeek has soundly shaken and perhaps invalidated my thesis of endogenously expanding rates. As Keynes quipped, "When the facts change, I change my mind. What do you do sir?"
Let me tell you what I see on the charts.
Big picture is an uptrend on the daily and down trend on the weekly when looking at Parablic Support and Resistance (PSAR). This trend indication inconsistent with my proprietary momentum indicator; reason is my indicator is linear whereas PSAR is exponential.
The region from 113'08 to 113'21 is a pivot zone. You can see on the chart, the upper/lower boundaries of this zone are previous PSAR levels that were invalidated by strong market moves that later faded. On the 1h chart the action has been choppy.
The lower boundary of the pivot area lines up with one of the key levels I observe when a momentum flip occurs on my proprietary momentum indicator (refer to arrow shown).
When a new bull trend is signalled by a flip to positive momentum, my technique is to pencil two potential levels of support
the close of the signal bar
the low signal bar (or high of the signal bar in a new bear trend)
On this basis, the absolute level of support for the current bull trend on 1hr chart is all the way down 112'17, which was the low of the signal bar at 9:00 EST on Friday, 24 January.
When I started drafting this note my decision was to hold all my ZN & ZB positions and wait for further evidence I am on the wrong side of the market. My plan was to wait for two daily closes above 115 before scratching.
Instead I've decided to cut my exposure in half immediately, and will scratch entirely on two daily closes above 115.