My brother-in-law is a blue collar man, he has a high-paying job in road construction and also owns a number of rental properties. He does most of the work himself so if a tenant has a water heater blow up he will replace it himself, he has friends who owe him favors so he can get low-cost help too.
If you are flipping homes yourself you will do it on the same basis and work for equity. If you are a corporation you're going to (1) have to hire people to do the work, and (2) probably hire some supervisor to drive around in a $90,000 truck to watch the first group. They all get cash before you do.
The individual home flipper working out of their own money is going to only choose projects that look very favorable. The corporation is going to hire a manager and that manager will probably be told to close a certain amount of deal flow and will feel pressured to close deals that aren't very profitable and not have such a financial incentive to keep the deals profitable.
The corporations trained an AI on data that showed a strong upward trend in selling prices. The result is the AI predicted that most flips would be profitable and thus almost always said "buy."
This worked until the housing market slowed down, the upward trend in selling prices slowed (and reversed in some cases), and the AI predictions of profitability became wrong.
① Insufficient or poor quality training data (e.g. better an old algorithm on good data than a cutting edge algorithm on poor data)
② No calibration. Calibration seems to be the best kept secret in ML (people know need good training data but they are lazy or not brave enough to insist on it... I've made that mistake, but https://scikit-learn.org/stable/modules/calibration.html seems to be truly obscure) In the case of a trading strategy or other commercial action you would be calibrating on expected return or possibly something that balances risk and return like Sharpe ratio.
③ Non-stationarity. Distributions are changing all the time on their own, but in markets they get changed by your own actions and those of people following the same strategy of you.
This book has a great study of a hedge fund strategy that burned out the way many algorithmic strategies do
https://www.amazon.com/Hedge-Funds-Perspective-Financial-Eng...
I think it's key that the burnout happens because of greed. A given strategy can absorb a certain amount of capital. When more capital gets attracted to the strategy the returns inevitably go down, market participants can try to make up for this by increasing the leverage but you can see how that goes... So I would still blame too much greed and too much capital because it inevitable that markets fluctuate.
If you're talking about building new homes, what you're describing is pretty much "real estate development," which is a profitable but risky business in the U.S. A real estate developer has to work with local authorities to site the project, which is itself highly politicized and can take years. Then they have to secure capital, hire architects, engineers, builders, and comply with environmental laws. Only once the project is mostly done can they market the homes to potential buyers.
I suspect a lot of the properties that were bought were bought to fill some kind of goal for the period rather than they having profit potential.
The better investment is buying, renting and then selling a decade later. The idea is that the renter pays the mortgage and other expenses for the property. But that's too long for most companies.
And "we are geniuses at 10,000 feet!" financial industry schemes, especially ones in cyclical sectors, have a strong tendency to scale up as the tide is turning against them.