Idea for exploiting Collective2

Discussion in 'Automated Trading' started by braincell, Nov 16, 2011.

  1. A commercial such as a big retail bank with an investment division creates a daytrading algo that works consistently. It only works with a limited capital, like most algos. Too much capital and it becomes detectable and countered by other algos. The bank takes it's algo and puts it on collective 2. It allows no subscribers for 2 years. The algo is proving profitable with very low risk. All of a sudden, the automated trading system is accepting subscribers at a low cost. People jump on it. What happens next. The algo works as it always has, sends out buy/sell signals and they are profitable. It becomes popular enough that the C2 subscribers make it's moves more detectable. Thus it starts becoming less profitable over the months. The commercial however earns more. How? They post the signals a short time (20 seconds) after they are generated. This makes the subscribers push the price up/down a bit in the favour of the algo, in a way making the algo the front-runner. The subscribers end up getting near to 0% returns over time, but the commercial is happy. The premise is of course that it gathers enough subscribers and that the algo knows how/when it's best to publish a signal and still make no major losses for the subscribers. With a sustained 0% subscribers won't bail so soon, because hey, after all the algo has 2 years of success behind it, so maybe if they just wait a little while longer it will make them rich.

    Just a thought...
  2. I think you overestimate exactly how much capital is behind the 'subscribers' of c2, and underestimate the legal risk involved in something like this;

  3. Well ok, collective2 is small but imagine selling this to prop firms. Legal risk is minimal short of a subpoena.
  4. Pekelo


    I don't get your idea. If the bank has a low risk, profitable algo in the first place, why mess with C2 at all? What is the advantage??
  5. And subpoenas is exactly how the legal risk will arrive;

    Not sure what you mean by "selling this to prop firms".

    As far as scams go, this one is pretty needlessly complicated: you gotta have a working algo first right? If you got that, and can get investors to believe it, why not just run away with the money?

    If you are going to do something illegal (this idea is based on a fraud), a competent criminal would take the easier route

  6. Ok C2 or prop firms, doesn't matter. Some prop firms will buy signals (in case you didn't know) and have their traders interpret them or act upon them in a slightly different way, thus dispersing the signal and making it harder to detect.

    Yes, the algo is low risk and profitable, but as i said, it reaches it's full potential very quickly. In other words it can't be made more profitable due to the liquidity and due to other algos detecting it's actions. So basically, since it can't be made any more profitable by normal means (increasing capital or leverage) you do what I suggested you do. If the algo is going long, and there is a point at which it will take profits (target), then under normal market circumstances it would go up by 2 points and stop, and that is your target. If prop firms or subscribers or whoever acts upon your signal also, then you are sending the Buy signal to them, and they further push up the price by hitting the Ask (market orders), so the result is the algo can adjust it's profit taking at 3 points instead of 2 points, because the buying from the algo (signals sent to subscribers) actually pushed the price up 1 point. If it's 1 point or 0.2 points doesn't matter, it's just for illustrative purposes.

    It would work best for an algo that works on 100+ instruments with medium liquidity, such as stocks of larger (but not oo large) companies. They would be best because they are illiquid enough to be pushed by "smaller" capital, and liquid enough for the algo to take profits as soon as a target is reached.
  7. Then you are in breach of your fiduciary duty to your 'subscriber'/'client'/whatever;

    There will definitely be a civil suit - possibly criminal

  8. How will they know? The bank can simply say the algo underperformed. Is that reason enough for a subpoena you think? I think not. The SEC is easily corruptable on these cases.
  9. Okay - I see we are playing the 'how will they know' game... which is fine - since this whole idea is a prima facie scam - and like all scams, the subsequent discussion is related to the likelihood of being caught.

    Good luck with that.

  10. Name one firm that "buys" HFT or LFT signals... and has been stated repeatedly, wtf would anyone enter into a massive fraud with a working algo? Your moronic assumption is that it will be traded flat which necessitates fraud to continue to earn?

    Beyond a mental (masturbatory) exercise; are you concerned that your signals will be delayed and you'll be on the losing end of a scam?

    And now you have the SEC involved. God this site is retarded.
    #10     Nov 16, 2011