Tip: You gotta be very careful in discerning the COT data. It’s a setup, not a given. Not always reliable. You need to know what Commercials are doing, even then, they are not always right. Need price action to confirm.
Typical AI tells include expressions like these and their variations : "quietly doing x", "worth stating plainly", triple negatives like not x, not y, not z. Excessive use of emdashes, semicolons, etc.
I actually ran the Substack AI check now. Scored 100%.
The ideas behind the text are still good and I agree with you. Just pointing out that the author voice is as important, if not more.
Authenticity has become a real edge these days. Doing the opposite of what everybody else does has always paid off.
You are real signal Erick, please don't fall for AI.
1) These short futures positions are imho a mostly a hedge, not an outright bet on falling markets. The dumb money (retail) isn't hedged. It is long to the hilt.
2)You correctly described the do-or-die aspect of the current AI capex madness. Yet, you fail to acknowledge that this itself speaks for huge overbuilding: when falling behind is perceived as the biggest threat by every major tech company, they will ALL err on the side of too much, too fast rather than too little too late, by definition. Add to that that there is plenty of circular deals in the space and that everyone tries to outrun the other in securing chips and power supplies and you have the receipe of an ai race that is completely ignorant towards any signals from the markets about true economic, monetizable demand. Everyone's chip investment signals to everyone else that there's even more demand - and the next capex increase is initiasted as a result.
This will end in tears and since AI capex is the major growth engine now in the US, an implosion of this bubble and a crash in new ai capex will send enormous shock waves through the economy from both financial markets (recerse wealth effect, capital losses, loan defaults) and macro activity (construction, consumption etc)
3) I'd argue thet geopolitical risks, esp. the Iran war, have actuaklly helped the ai bubble this year as many investors view AI as a save haven. Which is ridiculous if you think of it but flows into US treasuries and US equities from abroad signal exactly that.
I would not argue this will end in tears, all massive investment cycles do. I think timeline is important. We are really 3 years into this cycle, with inferencing in year 1. Not sure what retail is anymore, people love to point to it, but the vast majority of retail is passive. I was told retail got wiped out in 2022. Active management on the institutional level is mostly closet passive. So any move by algos is amplified.
Do we know what the tickers are for those nasdaq positioning lines? I can find something similar in Bloomberg but that one has swung from net short in August to net long now.
Looking again now and the net number might be TFC1RLFN - same title. And I wonder if the GS chart converts from "contracts" in bloomberg to $bn exposure on the GS chart? Could be why it looks similar, but not the same scales over time.
Contracts on bberg went from peak net long December 2025 to max net short by August 2026. Now back towards 0.
Is it? My friend wrote it. My internal checker didn't notice. I liked the point about how higher rates can be a positive not a negative. It came up at a conference I was at.
Tip: You gotta be very careful in discerning the COT data. It’s a setup, not a given. Not always reliable. You need to know what Commercials are doing, even then, they are not always right. Need price action to confirm.
Yes, I agree. The COT set up data was just one aspect of a broader bullish view anyways.
Excellent
Unfortunately.
Typical AI tells include expressions like these and their variations : "quietly doing x", "worth stating plainly", triple negatives like not x, not y, not z. Excessive use of emdashes, semicolons, etc.
I actually ran the Substack AI check now. Scored 100%.
The ideas behind the text are still good and I agree with you. Just pointing out that the author voice is as important, if not more.
Authenticity has become a real edge these days. Doing the opposite of what everybody else does has always paid off.
You are real signal Erick, please don't fall for AI.
a few remarks:
1) These short futures positions are imho a mostly a hedge, not an outright bet on falling markets. The dumb money (retail) isn't hedged. It is long to the hilt.
2)You correctly described the do-or-die aspect of the current AI capex madness. Yet, you fail to acknowledge that this itself speaks for huge overbuilding: when falling behind is perceived as the biggest threat by every major tech company, they will ALL err on the side of too much, too fast rather than too little too late, by definition. Add to that that there is plenty of circular deals in the space and that everyone tries to outrun the other in securing chips and power supplies and you have the receipe of an ai race that is completely ignorant towards any signals from the markets about true economic, monetizable demand. Everyone's chip investment signals to everyone else that there's even more demand - and the next capex increase is initiasted as a result.
This will end in tears and since AI capex is the major growth engine now in the US, an implosion of this bubble and a crash in new ai capex will send enormous shock waves through the economy from both financial markets (recerse wealth effect, capital losses, loan defaults) and macro activity (construction, consumption etc)
3) I'd argue thet geopolitical risks, esp. the Iran war, have actuaklly helped the ai bubble this year as many investors view AI as a save haven. Which is ridiculous if you think of it but flows into US treasuries and US equities from abroad signal exactly that.
I would not argue this will end in tears, all massive investment cycles do. I think timeline is important. We are really 3 years into this cycle, with inferencing in year 1. Not sure what retail is anymore, people love to point to it, but the vast majority of retail is passive. I was told retail got wiped out in 2022. Active management on the institutional level is mostly closet passive. So any move by algos is amplified.
Do we know what the tickers are for those nasdaq positioning lines? I can find something similar in Bloomberg but that one has swung from net short in August to net long now.
I’m not sure. It’s a Goldman Chart. I’m always uncertain about which CFTC series to use. Do you want to share what your chart looks like?
Looking again now and the net number might be TFC1RLFN - same title. And I wonder if the GS chart converts from "contracts" in bloomberg to $bn exposure on the GS chart? Could be why it looks similar, but not the same scales over time.
Contracts on bberg went from peak net long December 2025 to max net short by August 2026. Now back towards 0.
I did not do the Substack AI check in this, but it's AI written. There is no you in the piece Erick, please come back!
Is it? My friend wrote it. My internal checker didn't notice. I liked the point about how higher rates can be a positive not a negative. It came up at a conference I was at.