let's be honest: who is making money trading "mechanically"?

[fantastic]

I am really curious to find out who is making money trading mechanically.

automatic trading sounds very sexy, but what is the reality for all of you out there?
do you live off the profits of your computers?

 

[donaldduke]

The big profits you can make and the ease of trading sound very sexy.

However, the thousands of hours of research and development time you need to spend finding and building a robust automated system are not very sexy. And no, you cant just buy a system or copy someone elses system either. But there are many entry/exit/positions sizing ideas in countless trading books or on the internet. But to turn these into a production automated system that you can actually trade and make decent returns with takes a long time, years of feedback and refinement, just like any other business.

The large drawdowns you will have to take, sometimes several times a year, are not very sexy either. So if you want to live off your mechanical trading system you need to manage your cash flow carefully.

Unless you can employ trustworthy and very reliable people on your behalf, then the daily tech support monitoring is not very sexy either.

 

[SvenFoster]

From my own experience I can backup what DD has just said its tough. very tough.

There were a couple of fellas over on oanda that were trading fx automated and one "chaffcombe" took a portfolio based approach to attempt to smooth the returns.

He basically had several simple systems on different pairs and timeframes. He would increase and decrease each systems allocation of funds over time, maxing the winners and pruning the losing systems. He was careful not to change the allocations to agressively based on the premise that a market would come in line with a system for a reasonable period of time.

This approach seems v.sensible to me as isn't it basically how a trading floor would work? different people trading different things, differently. You'd put more dosh with the ones that seemed to have a clue.

Also, a few simple systems can behave like a complex system but easier to write.

So when i come back to automation, that will be an approach i'll investigate further.

 

[GammaJammer]

I went to a meet and greet / mini seminar / roundtable sorta thingy recently, hosted by one of the large algorithmic systems providers. The place was full of representatives from the larger algo firms in London, and discussion centered around the need for the ability to manage your deployment in real time, and the fact that no-one these days is just developing a model, deploying it, and sitting there for years just sticking their feet up and coining in the dough.

These guys spend absolute fortunes - millions in many cases, on R&D, cutting edge technology, proximity hosting, hiring smart cookies etc etc. The high frequency shops look upon roundtrip times to the relevant exchange etc in terms of milliseconds.

Some really good points were made, one of which was that all these strategies, and any edge they may generate, are basically a decaying asset. So there's a delicate balance to be surfed by these firms between R+D speed, amount of forward and backtesting, how much time and effort they put into operational factors etc.

Bottom line is - the systems stuff we see here on this site is really truly the tip of the iceberg compared to what else is going on (and while the thing I went to recently was pretty heavily dominated by high frequency traing shops, they were by no means the only people there. Plenty of macro quant funds too). And at the pro level the money and resources thrown at this stuff is unreal. Which is all the more reason to have a realistic view of what can be achieved with a small scale mechanical system.

Not to say that no-one will ever make money doing that, it's just that if there truly were a quickfix shortcut to riches, don't you think people might have thought about it by now on the wholesale side? Like em or hate em (and there are people on this ste who REALLY have the wholesale community in pretty low regard as far as I can tell), they're certainly not all stupid.

 

[NetTecture]

I am not sure I would buy that. Do not forget that those large systems move large amounts. If you invest a million, you need to make back a lot more. And that means moving lots of funds, lots of contracts.

What most people do here would be under the radar. I would be more than happy having one system earning 60.000 euro a year, then work on the second. With three such systems I make good money.

The large shops would not even consider such trades.

 

[littlemog]

So versus the big boys, most of us would not have even say, thousands to waste in 'R&D', so how viable can automated trading really be then? (at least not for a single person?)

 

[donaldduke]

These guys spend absolute fortunes - millions in many cases, on R&D, cutting edge technology, proximity hosting, hiring smart cookies etc etc. The high frequency shops look upon roundtrip times to the relevant exchange etc in terms of milliseconds.

Some really good points were made, one of which was that all these strategies, and any edge they may generate, are basically a decaying asset. So there's a delicate balance to be surfed by these firms between R+D speed, amount of forward and backtesting, how much time and effort they put into operational factors etc.

Bottom line is - the systems stuff we see here on this site is really truly the tip of the iceberg compared to what else is going on (and while the thing I went to recently was pretty heavily dominated by high frequency traing shops, they were by no means the only people there. Plenty of macro quant funds too). And at the pro level the money and resources thrown at this stuff is unreal. Which is all the more reason to have a realistic view of what can be achieved with a small scale mechanical system.

Not to say that no-one will ever make money doing that, it's just that if there truly were a quickfix shortcut to riches, don't you think people might have thought about it by now on the wholesale side? Like em or hate em (and there are people on this ste who REALLY have the wholesale community in pretty low regard as far as I can tell), they're certainly not all stupid.

Obviously the small retail trader cant compete on execution, r&d budget, operations etc with the big boys.

To compete I have to trade in a style where those things arent as important to success. This neutralizes their performance and budget edge over me.

A bit like driving to work in the heavy rush hour traffic in a ferrari or a ford focus, the extra performance and budget of maintaining a ferrari wont get you to work much faster than me.

 

[NetTecture]

A lot. Just make sure you know your nice.

* Skalping / hig hfrequency - forget it, or make sure you have very short ping times. I ping to CME around 60ms from my servers, so I CAN NOT compete there.
* But I can move. One can get in, out with 10 contracts at will. Make a little money, run. A 100 million fund needs to identify trends big enough for it to enter large positions. Which in themselves move the market. No "fast in, out" deals.

Technically you will find that the money you can do from scalping and day trading is limited per approach. Not because one is a bad trader, but because your volume soon approaches "disturbing mass" for the market (on a very short scale), at least for stop orders (i.e. getting out is more the problem in case things go wrong).

Remember, you make half a million - you are VERY happy. A 100 million fund makes half a million, they - suck.

 

[donaldduke]

So versus the big boys, most of us would not have even say, thousands to waste in 'R&D', so how viable can automated trading really be then? (at least not for a single person?)

I spent five years developing my trading systems, sometimes spending a hundred hours a week of effort.

If i paid a quant for five years r&d we are looking at around a million in costs.

As for execution, i use Interactive Brokers (IB). They provide me access to the same hardware and software infrastructure they use to generate almost a billion dollars a year in automated trading profits. I plug my automated systems into their infrastruture.

So i have spent the equivalent of a million quid on R&D and i have access to a professional execution infrastructure which probably cost IB in the region of 100+ million to develop.

So i would say automated trading for a single guy against the big boys is not completely unviable...

 

[pedro01]

Right now, I have a system on the NQ that looks to be set to make me around about $2000-$3000 a month. IT's backtested & the forward testing is on track. It doesn't make many trades per month & the above includes slippage & transaction fees.

It's not mega-money but it is feasible. It won't scale up to where I can make $200,000 a month and it would be of zero interest to any investment hours with a million dollar R&D budget.

 

[zupcon]

Mechanical trading is an absolute nightmare. Thankfully I've got a far less stressful job now, assistant to an alcoholic blind knife thrower, my therapist says I'll get over the trauma of algorithmic trading in a decade or so.

 

[GammaJammer]

lol zup. I'm not saying that it's not possible for a small retail trader to make money trading systematically. I fervently believe that there are discernable relationships and patterns in the market, and therefore that it is not entirely random. Ergo, I believe that this can in theory be exploited. All I'm saying is that there simply isn't free, easy dough out there for the retail trader, it's just that this is what everyone who comes on here having plotted a chart with a couple of crossing moving averages seems to believe.

GJ

 

[littlemog]

Thanks for pointing the way. It sounds daunting, but I guess if there's hope, we can find a way to squeeze through. I've looked at IBs too, though I'm not yet considering a switch until I do get a few good systems up and thoroughly tested.

Do ring if you ever start a thread somewhere on the results and findings in detail

I spent five years developing my trading systems, sometimes spending a hundred hours a week of effort.

If i paid a quant for five years r&d we are looking at around a million in costs.

As for execution, i use Interactive Brokers (IB). They provide me access to the same hardware and software infrastructure they use to generate almost a billion dollars a year in automated trading profits. I plug my automated systems into their infrastruture.

So i have spent the equivalent of a million quid on R&D and i have access to a professional execution infrastructure which probably cost IB in the region of 100+ million to develop.

So i would say automated trading for a single guy against the big boys is not completely unviable...

 

[francisfinley]

Interesting post. I have been following a few of the EA's for metatrader for over a year, nothing of note has come to light as obviously if it had at our level I think you would be nuts to release it.
That said the metatrader side seems to be dominated by our eastern european/russian friends. There is a guy from the metatrader competition selling his EA for $1000 but you really have to build your own so you can tweak it.

There is a firm in the city which has been running its own system for 15 years which has constantly produced a return, whilst not fantastic it has been solid - a profit is a profit.

I agree the big four spend oodles on this stuff and on this last crash you might not see the same enthusiasm for quants.

Most trades are automatically generated and what with the dark pools of liquidity stuff im sure it will become a bigger and bigger topic.

But at our level ( im a piddly retail spreadbettor ) i think you would spend at least 3 years developing your own system before it made returns.

As for the "try my system for 20 days and if you are not completely satisfied" tripe - listen matey if its so good why do i have to pay for it first on your dodgy looking website??

There are so many blogs and ridiculously bad websites/single landing pages with systems on its unbelievable. Im starting to wonder if all the vendors have people on here pointing people to affiliates and referrals - most vendors are part of someone else etc.

 

[Citizen2007]

Developing an automated system is hard work, there is no doubt about that. The amount of detail that one has to go into to get the process in order is immense. However that is not to say that it is not possible.

It requires a lot of patience to go through all the testing etc, let alone the monetary costs involved. The testing for bugs and improvements can take months.

The platform base you have developed or the approach you take is another factor that is crucial. How flexible is your base? How much additional work is involved in making smaller changes to your system? Do you have to rewrite the whole thing or just alter one aspect of the program?

The initial structure of the program is important and the mismatching aspect here is that most programmers are not traders and most traders are not programmers. The ones who have a combination of the two are doing quite well for themselves.

The best approach that newbie’s can take here is to begin developing a system but do it in phases. Testing and using each programmed phase to determine its effectiveness. For example, if you have developed a trend indicator, use it with your manual trading and see how it performs under live conditions. One it can assist in your manual trading and secondly it allows you to validate its performance. Then as one performs the way you need it, you make additional add-ons until you get to complete the exit or re-entry

The most important part is understanding technical analysis. This is all that the markets are. Whether you look at price or use certain indicators, the markets will follow a certain protocol. It is understanding this that will make all the difference. There is a time for just longs; there is a time for just shorts, a time for scalps and a time to hold a bit longer, a time to abstain. All these factors have to be understood and clearly defined before the process begins. Understanding how to work with all the different timeframes is another important factor that has to be understood initially. Once you have this mapped out only the should you go down the route of automation

And yes I do make money from my automated systems and am continually trying to develop new ones. All this would be so much easier if I was also a programmer.

 

[StratOpt]

I have been running full mechanical systems in full automation since about 2001. It is entirely a doable venture if a person puts forth the time, effort and energy. It doesn't take being a quant type of person to make it happen either. Fortunately, simple still works.

 

[fantastic]

For you who are doing well with it, do you make more than 48% per year ?

 

[donaldduke]

For you who are doing well with it, do you make more than 48% per year ?

A number like 48 percent, on its own is meaningless.

With a mechanical system you can risk 0.5% of your account on each trade or you can risk 5% of your account on each trade, its just a parameter in the system you need to set.

Risking the latter will give you 10 times the returns (assuming no compounding) of the former, ofcourse your drawdowns will also be 10 times larger too.

This is why metrics such as Mar ratio and Sharpe ratio are used to measure performance.

 

[fantastic]

A number like 48 percent, on its own is meaningless.

With a mechanical system you can risk 0.5% of your account on each trade or you can risk 5% of your account on each trade, its just a parameter in the system you need to set.

Risking the latter will give you 10 times the returns (assuming no compounding) of the former, ofcourse your drawdowns will also be 10 times larger too.

This is why metrics such as Mar ratio and Sharpe ratio are used to measure performance.

OK. When comparing systems, one should precise the Mar and Sarpe ratio as well.
Anyother metrics to compare the different systems?

 

[thierry911]

I do use mechanical system only. I spent about 3 years developping them. They are trend following, breakout and countertrend. I do use 3 of them in real trading. (that s for my own account)
I do trade for the past 20 years for banks and trade on a discretionary basis... I moved to mechanical because I feel more confortable to trade that way.