How does Collective2 work?

Discussion in 'Automated Trading' started by braincell, Oct 18, 2011.

  1. I've been developing some systems for a while and looking at Collective2, I see some interesting ones I could compare to. But how interesting are they really?

    Namely, I can't seem to find answers to some key questions that determine how accurate the hypothetical results of various systems really are. I've sent e-mails to their customer service, but no reply. Maybe someone here can enlighten me on these questions:

    1-Once a system is introduced to C2, it starts generating data for a portfolio which then shows up in the hypothetical results charts. Can a system be changed (recoded/improved) after introduction and if yes, what changes are allowed to be made?

    2-Can the system admins turn the system on/off so that it stops generating signals and thus creating a "flat line" on the chart?

    3-Can system admins cancel signals manually (since i've seen some signals cancelled) or do the systems do it automatically and there is no other way?

    4-Some hypothetical results charts look like they are going up, but then they have a few downward spikes. I'm not sure what they mean. When I looked very carefully at the "Recently closed trades" table there were no negative trades reported at those times. Could it be the trades were open for a longer period and thus visible on the chart, and in the meantime the PnL of the portfolio was hurt by the underlying oscilations?

    These are fairly important questions for me and my research right now so any answers would be appreciated. The C2 forums seem to not generate much activity so I thought I'd ask here. Thanks.
  2. fjpenney


    Of course a system can be changed. Trading systems on C2 aren't meant to be hard coded forever. Some systems are mechanical and there is nothing to prevent the system designer from changing the algorithm.

    A system administrator can close all open positions and therefore turn off the system if that is what you mean.

    Trade signals can be cancelled before they are executed.

    The recently closed trades are delayed at the system administrator's discretion (e.g. two weeks) but the equity curve shows up-to-date performance of the system.
  3. Thanks fjpenney, I was waiting 2 days for somebody to answer. ET to the rescue!

    It's really odd that they allow systems to be changed after introductions. That means that you can start out with one, and pretty much convert it to an entirely different system later while it's still showing the historical performance of the other. That's not much consistency that we can count on, but i'm guessing system admins will only really introduce improvements. Also, with such input allowed by a human it seems more like it's semi-automated trading, and the historical charts displayed are thus also a rating of a trader not just a system. This to me sounds even more appealing, it's like getting both - a system and a trader at the same time.

    I think I'll test some subscriptions out with a limited risk capital while i'm still working on mine.
  4. Although you got the concept of C2 right, you have to mind that anyone can create a "system". Most of them trade theirs on a discretionary basis but claim to follow strict rules. The entire site portrays the image of professionals running sophisticated algorithms but this is not the case. Many charts actually have extreme spikes (drawdowns followed by rapid recoveries), hinting at martingale trading. Since system administrators cannot cheat on performance they size up on their position to recover fast to attract new (or retain) subscribers. The system recovers because it could still meet margin requirements while you blew your account in no time because you couldn't. Those are most dangerous to your money. The very fact that you need to execute a trade over their interface skews the impression of a sophisticated computer behind every system (even though you can connect your broker to C2 for a fee). Everything is about attracting subscribers, so the faster and higher the chart goes, the more attraction a system will receive. Administrators don't have the patience to establish a long-term record because every 6 months without a subscriber costs a whopping $100 fee. Time is running out. The fee is not a fortune, but we all know how irrational we are with regards to money. Why waste time and money hoping for subscribers if your system doesn't even show up in any ranking after months? It would have been a much smarter concept to let people trade their strategy and establish a track record for free, then let them publish it when ready. Of course C2 would not bother considering it as their revenues would collapse. If you wish to have someone trade your money, choose a performance that is boring but smooth. Best yet, learn to trade on your own.
  5. fjpenney


    C2 is a platform for trading system designers to post their trades. As far as I know, there is nothing on the site to suggest that the trading rules will not be modified after a system designer starts posting trades. How many of us are using the exact same trading rules that we were five years ago? None. It is natural for the trading rules to be modified with experience and varied market conditions.

    Each trading system designer is ranked on a proprietary scoring system. I suspect you will find that the best system designers have scores above 990 (out of a possible 1000).

  6. fjpenney


    One way to deal with Bomb's criticisms would be for system designers to demonstrate that they are trading their own accounts using their trading system as per the posted trades on C2. There has been discussion on the C2 site about this but I haven't seen any interest by the site's owner.

  7. Right, I see, and that's what worries me the most. I've seen those spikes and looking at some systems I'm thinking maybe I can place my stops according to the frequency and depth of those spikes, then re-enter when the system tries to recover. However, often the recovery requires 3x the margin so it might not be quite possible at all. I suppose this also means that the "% trades profitable" figure isn't realistic either. For this reason it's only reasonable to trade smaller amounts of risk capital using any of the systems and try to figure out over time how to best make it work in the real world.

    yes, seeing system admins using their own systems on their own real accounts would be encouraging. It would probably make me subscribe to more systems (risk capital allowing).
  8. If you are a "developer" and have an account with OEC or a couple others you can make it just track what you do in your real account but I have no idea how to prove it to the public. I just created a "system" using an OEC account and it works perfectly but if I was to subscribe to a system I don't think there is any way I would have it auto trade for me it just seems that to much can go wrong. But I wouldn't trust what anybody says they are doing. Half the time I forget I even have a trade on :D

    Trades 25
    # Profitable 22 (88.0%)
    Avg trade duration 14.4 hours
    Annual return (compounded) 787.6%
    Average win $683
    Average loss $425
    Profit factor 11.8:1
    Max peak-to-valley drawdown (historical) 3.97%
    drawdown period Sept 10, 2011 to Sept 13, 2011
    Correlation w/ S&P -0.026
    Sharpe ratio 8.41
    Keep after worst-case slippage 90.0%
    Probabilities of future account loss
    Chance of 10% account loss 0.0%
    Chance of 20% account loss 0.0%
    Chance of 30% account loss 0.0%
  9. In my system description I tried to make it say I have no fing clue what I'm doing but I bet I can destroy most of these system that claim to be so scientific. And C2 refused to allow it :(
  10. Collective2 is idiotic. Their TradeStation interface flat out does not work, and their "tech support" makes no effort to help you fix it. I'm about to submit a chargeback request to Visa to get my developer's fee back -- a pittance, I know, but I guess it's a matter of principle.

    I'm annoyed with myself for wasting any time with them
    #10     Nov 5, 2011