Who wants to make a killing together?
The Bet of the Decade
I'm working on something that I think a lot of people would find interesting. As many of you know, CRE is in BIG trouble. And it's just starting. But heavy losses of property values and massive increases in defaults present huge opportunities.
The problem is accelerating. Here are a few recent articles:
https://sfstandard.com/2023/07/18/san-francisco-commercial-property-values-could-plummet-40-by-2025/
Now, whether banks are bailed out or not after loans stop being paid, technical defaults are still going to happen which will trigger credit default swaps.
Now, think Big Short. We identify which buildings are going to default and buy CDS against them. This presents two questions:
How do we identify which buildings are going to default?
Who would be willing to be a counter-party in this current atmosphere?
How do we identify which buildings are going to default?
Well, that's easy. AI. They didn’t have that in 2008 but we do today. You can buy cell phone data, building access data, and property owner financials, feed it into AI, and it'll tell you which buildings are occupied (leased), which are not, and based upon the owners’ financials, which are likely to default. It's really not hard. We can find even more data to tell us with a ton of accuracy which to bet against.
Who would be willing to be a counter-party in this current atmosphere?
Well, like 2008, banks. They are greedy and think they are smarter than everyone. The CDS likely won't be cheap on a bps level, but with an ace up the hole like AI, we'd be smarter. We can also run the AI for a while and see how accurate it is. If it’s right 80% of the time and banks are willing to sell us CDs for 20 cents on the dollar, you could literally make billions on this.
We could either do this with private capital or turn it into an ETF and let the whole world bet against CRE.
So who wants to do this with me? Reach out to me.
Sounds like a scam to me.
Does this AI already exist or you mean build one? I'm assuming build one as otherwise the banks would have access to it already (or after a few bad bets). I do suspect something like this already exists behind closed doors for identifying vacant buildings because some insurance companies have gotten suspicionsly good at identifing them without inspection.
I don't know how familiar you are with AI/ML but the problems it has a chance of solving generally fall into 2 camps:
1. Supervised: Needs a ton of labeled data (if a building is actually vacant or not) to find complicated relationships (Most useful AI)
2. Unsupervised: The solution/correlation is so obvious that it almost always could have been done faster with basic statistics.
If this problem is in the Supervised camp, then my concern is where are we getting the labeled data as vacancy is quite closely guarded secret by owners (as far as I know?). If its in the unsupervised camp, then that's great for the short term (assuming banks are not already aware) but also means they will likely figure it out quickly.
Aside from concerns about relying on an AI silver bullet so heavily, I'm not trying to put down the idea as even if in the short term, it could be massively profitable. I'm personally betting on real estate in a similar way but with less risk tolerance (and less profit potential). Personally, I would be interested in helping build out this AI (assuming it's of the supervised type) but investing in something like this is too far outside my risk tolerance (at least from the information I have so far).