The Rise and Fall of Mechanical Trading Systems

hey, if you know systems which are constantly making money, tracked by a 3rd party in real time AND without drawdowns deeper than $10k, post them here.


I'm sure we will all want to know such systems!
I may expand the number of systems to monitor.


ScaleOutBarracuda NG-2
https://strategynetwork.tradestation.com/StrategyNetworkStore/StrategyPerformance.aspx?ParentId=1012

this one is up $10,700 since publishing. Let's see how long it will continue making profits.

please keep the journal professional.
 
I don't trade a pure mechanical system, but I do program systems and believe it is possible to do so and make consistent money.
Any legitimate money making mechanical system will not be posted on the internet for free I doubt.
But, this thread will turn into a bun fight, guaranteed, it's like saying Ford is a shitty car, well they can be a times, but I'm sure there are also good Fords.
So we'll end up with the 'for' and 'against' gangs slugging it out.
Bottom line, very hard to PROVE any concepts when it comes to trading, (due to randomness etc) but there are tradeable concepts which can be coded. It's just a matter of probabilities, win some, lose some, but have the probabilities in your favor over the long term.
The machine will one day run much of our lives, soon we'll have robots driving our cars, it is programmable.
PS: Talk to Maestro
 
This is induction from a very small number of observations. The statistical significance of any conclusion is extremely low, if not zero. It is also a biased analysis because it is very rare that anyone would post in detail the results of a system that works, unless the benefits from doing that outweigh the monetary gains from trading it, something that is unusual.

As a result, any conclusion of such study is already predetermined and no general rule can be obtained by induction regarding mechanical trading systems in general.

I think students should not be allowed to graduate from high school unless the comprehend in depth the problem of induction.

http://en.wikipedia.org/wiki/Problem_of_induction

It is quite disturbing to hear everyday people reaching general conclusions from a limited set of observations. This creates all sorts of distortions in public thinking, business and can even become a major parameter in a crisis.
 
Mechanical systems do work if done properly. By properly I mean it must match the traders temperment and of course be robust. Correlations are constantly changing and expect drawdowns to be @ least 2x your tested models.

Attached is my latest program. It consists 6 models trading several LOW correlated forex pairs, it's 100% automated and
runs 24 hours a day. Red curve is backtested results, Blue signifies real trading for the year.

The benefit of several models is that you can monitor and turnoff
and backfill if necessary. So far I have only replaced 1 strategy because it became too correlated to it's sisters.

I assume that the only reason someone would sell a strategy
is because it no longer works or the creators temperment is unfit
to follow it.
 

Attachments

Quote from Now is Now:

Now...to answer my question...what is wrong with this trade?

NiN
More...


The market can massively turn against this trade.
The system was probably optimized so the most probable scenario is that EURUSD will sharply rise in the coming days ;-)
but if the price will fall....

it will be:
- a miracle
- just a lucky string of events helping this system capitalizing on the trend
 
Quote from intradaybill:

This is induction from a very small number of observations. The statistical significance of any conclusion is extremely low, if not zero. It is also a biased analysis because it is very rare that anyone would post in detail the results of a system that works,
More...


I disagree that we are dealing here with low number of observation.
On TS website we have about 360 trading systems currently. Its quite large number, don't you think?

A large drawdown is a single event, but there is no need to make a requirement for a large number of trades during a drawdown to make it significant. Loss is a loss and it's BAD.

So far, I've chosen the best 9 systems (in my opinion, of course) to track.
There are a lot of other systems, which are slightly profitable to date. And about 60% under water. My point is: in the long run, most (95-99%) of all public systems will drift towards zero profit or end up losing.

Sure, there can be other non-public systems which are better. But we don't have the access to track them in the journal. However! If we will find a grup of systems clearly outperforming all other ones and making new equity highs,month after month, it will be great lesson researching how they make trades, hm?
 
Quote from DT-waw:

I disagree that we are dealing here with low number of observation.
On TS website we have about 360 trading systems currently. Its quite large number, don't you think?
More...

Oh, my God, 360 trading systems is a lot?

Just think about it. If you have just 20 indicators and you want to use only 3 in a mechanical system you have as many as 1140 combinations. If using 2 only you have 190 combinations. If you use 4 you have 4845 combinations and so on. This is just for 20 indicators or conditions in general and it did not allow for repetitions, i.e. combinations where the same indicator is used many times but with different parameters.

Actually, the possible number of trading systems in a countable infinity.
 
bill,
its true- the number of possible systems is infinite.
however on TS website we have so many developers, so many markets. i tend to think these 360 systems are very different one to the other.

i highly doubt all these people who post systems there are using the same indicator or method in general.
 
You must log in or register to reply here.
Back
Top