The author lays out all the math and practical considerations and then…skips to the conclusion.
Did he put it to the test? Seemingly not. And now that the methods are explained, there will likely be too many other people trying this to make any money from it.
This article is essentially a white hat hacker disclosing a vulnerability.
He implied the buy-in was about 21k for one attempt with about 5-7% chance of winning. Mocking up some quick analysis for a blog is one thing, taking out a loan for 100k+ is quite another...
You can also split the cost of the 21k across many people. You walk away with less money if you win but you also get a chance to play without selling a kidney. The hard part wouldn't be the finding a fraction of the 21k but finding enough people that would agree on the subset of 20 matches from the overall 38 that you should bet on. Depending on what matches they have listed for the week, you might even play a smaller subset than 20. If most of the matches are strong teams vs weak teams, you stand a better chance than many evenly matched teams. Although his data seems to suggest that splitting winnings somehow results in even less winnings than you would normally think so it really does seem to be that high buy-in is the best course of action.
He also asked the question about citizenship to play the lottery which he never answered in this case. I wonder if at the least he was flying to Australia to purchase tickets.
In the early years or Ireland's national lottery, a syndicate was opportunistically assembled (1992) to guarantee a lottery win and positive return on investment. They succeeded [1].
But in a general lottery (random) it makes more sense to put $50 into a single lottery than $1 in 50 lotteries. That's the only way to improve your chances.
If there's any chance at all that the lottery does not draw from an equiprobable distribution and some numbers are privileged - but you don't know what they are, it's best to spread your bets to increase the chance that you bet on the hot number by accident. This is true if a single number has any edge eps>0. Therefore you should almost always (with probability 1) spread your bets.
$21k per play
5% chance winning
= at LEAST 4m bankroll?
It’s still a fun read (except the ending) but that should’ve been near the top to separate casual readers vs imma-get-rich-by-applying-similar-method readers.
Thanks for the feedback - included 'Results' section with an idea of worst bankroll draw down
Also, as mentioned by @falsedan, 20 * 20k = 400k (if that's the math you were attempting, which is not a sophisticated way to determine bankroll, but a good starting point)
