The best of Collective2?

Being a bit of an analysis freak (I’m a statistician, so I can’t help myself!), I thought I’d put on my “fund managers” hat and go searching for systems on C2 I’d seriously be interested in.

The results can be read at:

… including the process, criteria, and calculation details.

As I state at the top of this page:

“Like beauty, “the best” of anything is very much in the eye of the beholder.” So I don’t claim this to be a definitive list or anything, just my view of what’s good here at C2.

The systems that rate a mention (and links back to their system pages) are:

Holly FTSE

Team Aphid Bird

Have Fun


Gold Swing Survivor Futures

DataGlobe FX II (now has a DD that will drop it from the list), and

GoldenStocks Short Only

Take a look and let me know your thoughts re the process, method etc.


What matters to subscribers though is if past performance is informative for future performance. I looked into that for long-only end-of-day stock systems, and had a very hard time finding a relation between past and future performance for these systems. You can find the full analysis and discussion here:

Science Trader,

Yes, I think your conclusion (on your blog) that the Sharpe ratio isn’t sufficient on its own as a predictor is very true. In fact it puzzles me why the Sharpe ratio is an industry standard (other than the fame of its creator - Nobel prize etc, but the fact he worked for LTCM should ring alarm bells!).

The Sortino ratio overcomes the major issue with Sharpe by not penalizing upwards thrusts in system performance, but even it on its own isn’t likely to be a good predictor.

The long-tailed nature of system performance distribution (positive kurtosis, or skew) is likely to link to the similar distribution of market volatility: those rare, extreme volatility events knock out systems that until that point had performed superbly (perhaps for many years). So increasing the look-back period only brings marginal improvements in predictability.

I just wish I was as clever as you and knew how to use that API stuff to extract data from C2. I looked at it briefly and got bewildered by the prospect … perhaps I’ll have to head back and take another look as I’d love to be able to do the time-based breakdown like you did. Not sure if it would provide all the data I’d need to test my broader model.

Going forward I certainly plan to analyze the models performance and make adjustments to improve it - much in the same way as I do with my trading systems.

Thanks for your response. I think achieving predictability for system performance will prove as challenging as it is for the markets themselves. Still, I like to think I’ve made some progress in the latter area, so … it’s tempting enough to continue the journey!

Murray - what your paper stats do not tell you are the real world experiences. At least one of these systems is extremely difficult to trade manually and two individuals have had their accounts obliterated autotrading it. Not necessarily the fault of the system vendor but the high leverage nature in conjunction with need to autotrade can be a recipe for disaster.

As for Sharpe ratio, Sortino may be superior but there is something to be said for Sharpe. High Sharpe implies low volatility and thus reduces emotions involved with trading. This is an important aspect for me. For example, when I have a really good period of trading my feelings of greed take over and I have a tendency to want to overextend myself - overleverage, put too many eggs in one basket, etc.


Hi Steve,

Thanks for your feedback.

I know what you mean about the greed emotion taking over after extended periods of good trading. I had that issue big-time for 18 months or so. Better trading discipline has helped, but the big thing for me this year has been taking little trading holidays for a few days after particularly good runs. I helps ooze the adrenalin out and get me re-focused and re-grounded before getting back into the markets.

It’s a very good point, and maybe a plus for AutoTrading?

Re use of too much leverage: That’s something I’d like to add into the analysis, and I’ve placed a separate post asking for leverage stats to be available: highest ever used, this months average, and increase % over last month. Feel free to add your vote for leverage stats at that post!

You’re right, paper-based analysis will never uncover all the issues experienced with real-life trading, but I’m hoping the leverage data will get me one step closer to a useful tool.

If Matthew can’t/won’t deliver it (I really think subscribers should be able to view this info), then I’ll look to see if it can be gleaned via one of the API’s.


Our system is on Murray’s list and we are aware that Team Aphid Bird is difficult to trade manually if you cannot follow FX-Market 24 hrs, because signals are random. We do it for our own accounts since we work as a team and team members live in different time zones and we can cover FX-Market 24hrs.

But regarding to autotrading and obliterating the accounts, last week we were told that there are some subcribers using the same trade size as we are, but with an initial capital of $5,000 or $10,000. Because they think we recommend 2 standard lots or 20 minilots. When this is the case it is obvious that the account will be obliterated, because we are using 20 minilots trade size for $100K initial capital. If they use one tenth of our tarde size with $10K initial capital, there is no way to obliterate the account with such a drawdown figure of ours.

We have been trading the very same strategy for our own accounts since November 2006 and our return is 120% in the last 12 months.

As a result, in order to help these newbies (or wanna be “get rich quick” dreamers if they are choosing the 20 minilot trade size with that kind of initial capital consciously), we are considering to scale down our initial capital and hence trade size so that they can autotrade the same trade size we use on C2.

Best Regards,
Team Aphid

In the particular case I was referring to I never heard back regarding the root cause of the problem. But it sounded like a screw up on the broker’s part (stop orders were dropped or something like that). If it was in fact subscribers trading with too much leverage then that is their own fault of course.