Press play and the two runs walk the session together, one window at a time, from
12:00 AM Pacific. The left panel is the always-up run, the right is always-down, and
both are placing their bets into the same 96 windows. Both accounts start the day
with $200,000, the same number, so the two panels can be read
against each other. Each shows the running total and what is actually in the account. The
second one is the one that dies and recovers.
Both runs buy at the price the window opened at. That price is not a coin flip and it is not the same twice: across the 96 windows the YES open ran from 40 cents to 60, with a median of 49. Buying up at 62 cents and buying up at 38 cents are different bets, and the run takes whichever the window offered.
The assumption is the fill. The open is one print at the top of fifteen minutes. A real order placed into it can fill a little worse, and the spread and the depth of the book are not in these figures at all. So here is the size of that: move every single fill 1 cent against the trader and always up finishes +$4,495.34 instead of +$4,936.51, while always down finishes −$2,008.74 instead of −$741.25. Read that as the size of the guess, not as a margin of error.
Every figure was settled when the page was built. Nothing is simulated in your browser,
and with scripting off the full ledger below is still the whole record.