> I learned some things about the market during this time, but did not succeed in my detection algorithm.

What failed? Was it too late to follow the trend by the time one was identified or something else? It seems much more transparent than trying to reason about say dark pool trades.

▲

hx8 4 hours ago | parent [-]

I am unfamiliar with working with such signal-to-noise ratios.

* I only had the example trades in the news as 100% confirmed positive trades.

* There are hundreds of millions of dollars in trades a day in Polymarket.

* In Polymarket you can just spin up a new account. If an account spins up, makes a $50k bet and wins, and then has no other activity, was that an insider trader or just someone with a behavioral pattern of spinning up new accounts? Just following up on these types of trades didn't provide a very big edge, as the nature of the trade adjusts the payout percentage.

▲

jldugger 3 hours ago | parent [-]

> In Polymarket you can just spin up a new account. If an account spins up, makes a $50k bet and wins

You'd probably want to use some form of bayesian ranking, like say add 1k of total bets and 500 in total winnings to the raw scores.

But your bigger problem is just that people spinning up new accounts may be doing it to avoid your tracking. The kind of person with lots of evidence that they're good should be smart enough to know about copycat traders. The evilest among them might use a small bet to lure copycats and then trade against them in alts.

	▲	pessimizer 3 hours ago \| parent [-]
		> The kind of person with lots of evidence that they're good But what evidence is there that anyone is "good"? The fact that I am an insider to one event does not make me an insider to all events. If I were an insider to many events, I would probably have better ways to profit off of them. Placing a big bet is something I would do if I knew one thing.