Comment Re:You cannot do that (Score 1) 310
I apologize for being unclear, but apparently you really did not understand what I was saying in that last post.
Whatever phrase you finance guys have come up with involving "algorithm" apparently has very little to do with the CompSci definition of the term. In CompSci an algorithm is any repeated process that can be approximated by a series of instructions. Any process. If you only data blondes your mate choice algorithm includes a a step about hair color. If you think Apple's long-term worth per share is $125, and therefore whenever you see that it's below $125 you buy and whenever you see it's above you sell; that is an algorithm even if you don't set up a computer program to do it automatically.
Thus, if the market price is what the last guy paid, then the algorithm is the price is what the last guy paid.
If the algorithm is that simple then it's harder to prove because you can't just turn the steps into a computer program that nobody could argue with. You could turn the steps into a computer program, but that would apparently involve guesses about how his unexecuted trades affected actual executed trades, which makes it hard to prove beyond any doubt. But the standard is not "any doubt," the standard is "reasonable doubt."
But that kind of hard is typical in financial crimes, which tend to be things that are perfectly legal except under certain arguable circumstances (ie: Martha Stewart would have gotten off if she'd had a diary that gave some other reason for selling her shares that day).
So I suspect they've got something besides the trades. Perhaps his latest business included him boasting about making money off a crash he'd help create, or an email to a snitch said something about actively manipulating market prices, or they think they really have isolated the effect of a single unexecuted trade on the market price.