How is your experience with ‘renting’ trading strategies on sites like Collective2?
For several weeks I have been monitoring several trading strategies available for rent on the Collective2 site. Recently I have subscribed to a fourof them. Mostly I have taken the signals and traded them manually. One especially promising strategy I have put on ‘auto-trade’. My trading results thus far are slightly positive.
Is it a viable approach to trade strategies available online? What sites other than C2 have people used and had good trading results?
There’s one thing you have to pay attention when using such systems. You have to apply the same money management as the “provider”. Otherwise you will not have the same results as them. If they are using 2 contracts, for example, you have to do the same and not only trade the first target…
This mean that you have to have enough capital to be able to trade the same way. Also, be careful to things like the maximum draw down the system have. Your account should be able to “swallow” it…
I personally would recommend you to use some algo trading strategies you can backtest yourself. In order to be able to understand the different money management aspect.
I can advise you some…
All the best and feel free to ask any questions you might have.
Other sites are StrategyRunner, Zulutrade, Tradency.
In these sites, im sure there are good trading strategies but there’s also a lot of crap so you need to be careful and investigate/test them before you try out real money. Be careful with:
– Backtesting that is tampered with. If the sites allow HTML uploading of backtests, these could be manipulated.
– Backtesting that is overoptimized, fits like a glove to the past but will not work in the future. They need to have forward testing to be trustworthy
– Look for real live trading results that match the performance expected in backtesting.
– Beware of systems that dont loose, because when they do, they could wipe the account (systems that win 1 pip per trade and win 99% of the times but when they lose they loose 200 pips)
– In some of these sites developers earn money for each trade they do, multiplied by number of users on it, so they are motivated to make many trades, not necessarily in your best interest.
(im leaving my strategies out of these sites for now and making an independent trustworthy portfolio www.forexperiment.com)
Two sites which were not mentioned are Currensee.com and myFXbook.com. I would not trust the latter, as some strategies there are blatantly bogus.
I would recommend diversification, i.e. committing small parts of your account to different strategies (after, of course, screening each of the strategies individually).
These sites have so many strategies that one needs to remember that even good live performance can be result of pure luck. Collective2 for example shows that they analyzed 15,206 strategies over 7 years. Let’s say that each strategy has 50% chance to outperfom the maket and if it is underperfoms it is closed then after 7 years you are going to have 118 strategies that outperformed just based on pure luck (15,206 / 2 / 2 / 2 / 2 / 2 / 2 / 2 = 118). This concept is descibed very well in “Fooled by randomness” book by Nassim Taleb.
I am also curious why people would publish successful strategies instead of just trading them using their own capital (this is what I do).
in myfxbook you cannot activate strategies to trade them, its just a showcase of systems. One one side there are the systems which are pure backtests and many are manipulated… but also there are live accounts running which cannot be altered and this is what many people use to publish their results and show other people their systems, for example:http://www.myfxbook.com/members/forexperiment/forexperiment-demo/93956
Interesting the “randomness” concept BUT although those 118 systems that might have ended positive 7 years in a row, they most probably wont be stable and with a smooth curve if its because of randomness. So the risk will be high on them and wont be attractive… well there might be 1 case in 15,000 where its smooth AND random… and yes, it’s probably the one you’re going to choose and start loosing the moment you get in HAHAHHAH
What attracted me to Collective2 is the actual trading results they publish. I don’t trust anybody’s reported backtest results – luck, overfitting, fabrication – too many ways to go wrong. But C2 monitors each strategies actual Buy/Sell signals and posts the results, allowing you to filter strategies by max drawdown, longevity, Sharpe etc.
It’s true you have to do your homework and dip your toe in cautiously. One of the strategies I tried is now experiencing its ‘second worst’ drawdown. Glad I was only paper trading it most of the time!
Good point about the strategies with high percentage winning trades. Those tend to be scalpers who are in and out of mostly Forex trades in a few minutes, taking a tiny profit on almost each trade. Then every so often the ‘fat tailed’ distribution rears up and takes away most of the profits in one ugly trade.
Some of the other sites offering trading strategies are strategyrunner.com, strategynetwork.tradestation.com, zignals.com, ninjatrader.com, activetradermag.com
I have not traded any strategies from these sites because they don’t have as robust tracking of trading results as Collective2 and some of them are tied to particular trading system products that I don’t have. That is not to say the strategies on these sites aren’t worthwhile.
take a look at IntelliTraders. You can view the setups and trading strategies. If you would also like to try it for a week or two I can make that happen for you. Let me know what you think.http://www.intellitraders.com/trading-setups
I looked at a lot of these sites & systems a few months ago. One of the major things that concerned me was that the sites weren’t actually running the actual systems themselves but just reporting the results based upon signals entered by the strategy owner. As such it is a record of signals entered and not the system itself because there is no way to audit that the signals entered were 100% accurate for the system described. This could give a very distorted view of the the strategy if the owner for any reason doesn’t enter 100% of the signals accurately and in a timely manner. It also concerned me that the owner could stop updating the website mid-trade and you might not know until it was to late.
Imagine somebody registers with the site with 64 different user ID’s. Let’s assume they enter a signal once a week, and have half their strategies use that signal, and half of them the opposite. After 6 weeks they have one strategy that has been up for 6 weeks in a row and I believe 6 more that have been up 5 of the 6 weeks. If they then take those 7 strategies and their next best, and apply the same theory for the next 3 weeks, at the end of 9 weeks they are likely to have at least 2 very good looking strategies.
you honestly believe these types of silly tricks would fool a buyer long enough for them to part with serious money? Remember, the strategy developers on these sites are there to get long term subscribers. There is a community of users and browsers who share analysis and experience with the strategies and the developers. Guys pulling stunts like this would quickly be out’ed. And, few people would pay money to trade a strategy with only a 9 week lifespan.
I did check out IntelliTraders.com. I have to say that I’m not very impressed with the content there. It seems largely like a front to market eToro brokerage.
I have had good experiences with collective2 .
I am providing signals on Zulutrade and just begin with Collective2, for the time being I am using it as a way to control my own realtime simulations.
When following such signals you have to be very carfeful not to be “fooled by randomless” as Andrew said. Indeed strategies that underperform market are deleted quickly by their providers.
If you use an MT5 ExpertAgent Zulutrade interface is very reliable, but collective2 looks like more professionnal.
Thanks all for the list of other sites that are like Collective2 (C2) – I’ll have to check them out one day. I have a question:
Has anybody tried this using historic data from C2 to predict future performance?
I have in mind spidering the C2 site (say once a month) and using statistical pattern recognition techniques to estimate future performance based on the C2 statistics.
The best performing systems on C2 are a mixture of those that are fundamentally very good, and those got lucky and survived. One way of getting closer to the Holy Grail is to find the C2 systems that are likely to work well in the future.
Whenever we evaluate systems we use a whole load of stats, such as max drawdown and Sharpe ratio to come to a subjective opinion, it would be good to use the same stats to come to an objective opinion.
It’s clear that past performance is some indication of future performance, but it would be good to model it and find out how much we can rely on the C2 statistics. There’s a load of other useful information there apart from the standard stats as well – such as the number of systems the author has on C2 (the more submitted the higher the performance of the top one).
To make things concrete, as and example I’d like to maximise my returns with a Sharpe ratio of 2.0 or better. I could do this by estimating my statistical model then evaluating all C2 systems, discarding those with a predicted Sharpe ratio below my threshold, and sorting the rest for highest predicted gains.
So, has any one done this yet?
This sounds like a formidable task, to me. I don’t have training in mathematics.
Assuming you could devise appropriate analytics for this objective, the small sample size might make it difficult to achieve statistically significant results. For example, there are only fourteen strategies with 1 year or more of history on the site. If you discard those with Sharpes < 2, you are left with only two. --- This sounds like a formidable task, to me. I don't have training in mathematics. That's okay, I do and this is the sort of thing I do all the time. Dan: there are only fourteen strategies with 1 year or more of history on the site I make it 260 strategies that have been active for more than 1 year (seehttp://www.collective2.com/cgi-perl/system/grid.mpl). So that's probably about 10,000 data points, that'll do for starters (there are over 1000 active, so that's 1000 more data points every month). Okay, there are only 20 strategies that have a current Sharpe ratio > 2, but it’s possible to learn something from all examples.
I don’t immediately see how to get a list of all current strategy ID’s but when I sort this out it’ll be easy to grab pages likehttp://www.collective2.com/cgi-perl/system57938439 every month and extract the data.NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!