You can register 150,000 domains for $1m. Right?
So the break-even odds is 1/150,000.
However if you sell 150k domains for $1m, you will have $1m to register another 150k domains.
So the break-even odds now becomes 2/150,000
If you sell the second 150k domains for $1m then you will have another $1m to register 150k domains for the third time.
If you register 150,000 domains for 150,000 times then one of your hand registered 22,500,000,000 domains must be sold for $1m.
If you don't have money to start with 150k cycles, you can start with any number of domains you can afford. The only difference will be the number of buy-sell cycles until you register 22,500,000,000th domain.
For instance you can register and sell 10 domains each time. You will need to repeat this cycle 2,250,000,000 times.
If you have 40 years to die, it's 14,600 day, approx. 15k days.
To do something for 2,250,000,000 times until you die, you must do it for 1,500,000 per day, or for 1,000 times per day to do it for 150,000 times.
If you sell 22.5b domains you will likely sell couple of domains for less than $1m that are actually worth $1m or more.
Therefore the point is not to sell 22.5 Billion domains. The point is to know which domains are worth $1m. One the other hand, if have this knowledge, then you don't need to register 22.5 Billion domains. You can register 1 domain and can sell it for $1m. So there is a mutual relevancy between knowledge and obtaining the knowledge from trials and errors. All you need is to adjust the balance between your current knowledge about $1m domains and the required number of trials and errors to obtain the absent part or parts of your knowledge.