If you look at the graph near the bottom, he's offsetting short-term gains with loss. But it's a loss, he's losing money. And looking at the two, many years have bigger losses than gains.
They even state in the article "Of course, if you consider the trade as a whole, it makes no money."
> One strategy, in simplified form, works like this: Make two bets that should move in opposite directions. Think of, say, both betting on and against Coca-Cola’s stock. Towards the end of the year, one bet will be up, and one will be down. At 365 days, the last day a trade is considered short-term, sell the one that’s down. A day later, sell the one that’s up.
> Of course, if you consider the trade as a whole, it makes no money. But that isn’t the point. You’ve found a risk-free way to generate two valuable commodities: short-term losses and long-term gains.
They're picking which gain/loss to realize - in both directions, on the same stock - in a way that means they always win.
A loss is not valuable.
Edit: see my follow-up comment before you make another post re-explaining the scheme. The point is it's bad explanation because the merits have nothing to with "losses" being generally "valuable commodities".
The scheme works as a whole, but the dynamics of it have nothing to do with losses being valuable commodities. It's just an over-clever writer trying to jazz up the explanation without realizing they're saying nonsense in the process and introducing a bad model.
This is like those Saturday morning commercials that say Cocoa-puffs is part of a balanced breakfast, and tyring to explain that by asserting how each component, including the Cocoa-puffs, is balanced. That's not how it works, the aggregate property is not piecewise reducible.
Suppose during 2023, you make $1B from short-term trading. If that's all that happened, you'd be taxed for short-term capital gains on it. But also, this happens:
July 1, 2022: Place a large bet "A" on S&P 500, and a separate large bet "B" against the S&P 500.
July 1, 2023 (365 days later): Suppose S&P500 is up. Sell your bet B. You lose $1B in short-term gains.
July 2, 2023 (366 days later): Sell your bet A. You make essentially the same amount that you lost yesterday, $1B, but in long-term gains.
Total for 2023: $0B net short-term gains, $1B net long-term gains, taxed at a low rate.