Big Mike's Trading Blog

Day trading futures, discussing money management and trade management techniques, and more

The allure of automated trading

I think everyone can agree that day trading can easily be described as a dream job. Naturally, this is only the case if you are making money, but nonetheless trading gives you unparalleled freedom and control over your own destiny.

Having owned several businesses, I can also say that trading is even more liberating than being self-employed. If your small business requires interaction with customers (and whose don't?) or if you hired some employees to help with the work, then you've now got that "burden" which does not exist when you are trading.



So, why is it that many traders want to take this incredibly desirable job and put it on auto-pilot? Yes, I am referring to automated strategies. I believe that most people looking to automated strategies for success are doing it for all the wrong reasons.

Here are two items everyone lists for reasons to automate their trading:
  • Can't watch every setup, miss some setups because I was afraid to pull the trigger
  • Program my strategy to follow my own rules precisely, because I can't be trusted to it discretionary
Sound familar?



First, why do you care about missing some setups? How greedy are you? Or maybe more accurately, you missed good setups because you didn't pull the trigger and now you require more setups or maybe more excuses to trade. You need to be honest with yourself. Do you want to make $500 a day? $1000 a day? Fine, no problem. Those are my goals as well and they are completely realistic and can be accomplished with only a very small number of trades each day, usually just 2 or 3, for a couple points each. Don't be afraid to trade more contracts once you have a proven system. Usually people are only afraid to trade higher numbers of contracts because they have a pattern of losing. If you can't make money with 2 contracts then you aren't going to make it with 10.

Don't concern yourself with missing trades because really there are plenty of trades out there. Focus on being consistently profitable with two to three trades a day. Usually these trades can be had before lunch time, so you don't need to be glued in front of your computer the entire day, just 3 hours of it. If that is too much trouble then I think you are in the wrong profession, but please stay and keep trading so I can be on the other end of your trades.



Second, automation has the appeal of being able to follow strict rules you set forth in your strategy. Most people go down this road because they couldn't be trusted to follow their own rules in live trading. Sure, their rules work fantastic when they look at a historical chart. But when they sat down at their desk and started to trade, all those rules vanished into thin air. At least they've come far enough down the road to realize the problem is looking them squarely in the mirror, and they stopped blaming it on the market.

Automated strategies are great at following rules. But they introduce a whole set of new problems which are nearly impossible to overcome. We know that 90% of traders lose money, and 10% take all 90% of the losers money. Now when you automate those odds are even more against you. Take it from someone who spent over a year, 8-12 hours a day, working on incredibly complex strategies.

Instead of focusing all of your energy on automating, why not focus on correcting the reasons you can't follow your own rules? I think it is fair to say most people who develop automated strategies already have a very strict idea in mind, a strict set of rules and principles to follow. They just can't follow them. So that is the real problem here, and that problem is far simpler to solve than making an automated strategy work.

Do automated strategies exist that make money? Of course. In fact, each year more and more of the market is traded with automated strategies. But should that really surprise you? Each year, are there more people in this world or less? Each year, are more individuals trading the markets or less? So just because more automation exists in the market doesn't mean that overall these strategies are making money. They fall into the same 90% losers, 10% winners as the rest of the market.



My advice is simple. Seek out other ways to tackle what you perceive as obstacles in your trading, whether they are emotional, lack of time, or greed.

What's your opinion? Let me know.

Mike

33 comments:

Big Mike said...

I also want to point out that mechanical systems are not automated systems, and mechanical systems can be beneficial in my opinion.

With a mechanical system, you get alerts once something meets your criteria, then you can decide on a discretionary basis whether or not to take action on the alert.

I have several of these I use off and on which throw up alerts on my charts.

Mike

Neal said...

Mike - well said.
I know there is no Holy Grail - of course there isn't. That's one of the first things you are told as a newer trader.
But I often behave as if there is.
There is a fine line between research that challenges yourself to constantly improve, and obsessively seeking the perfect indicator or strategy.
One quote that resonates with me regarding this:
"The perfect is the enemy of the good." - Voltaire
A consistently profitable 'good' strategy that the trader can have confidence in to effectively execute, accepting the small losses as they come, while focusing on the big picture of net profits, can be discarded or ignored in the pursuit of perceived 'perfection.'
Thanks for exploring this challenge which has been a struggle for many.

Chris said...

Mike, I think your problem with automated systems is that you spent ages developing "incredibly complex strategies".

In fact a significant number of our customers who do create successful auto trade systems actually go the opposite way - using some very simple, but extremely well researched strategies.

IMO the key is to find a very simple strategy and then optimise that for time and ideal parameter settings. Spending time on that is often far better spent then combining more and more indicators into increasingly complex strategies.

Big Mike said...

Chris,

Can you possibly give an example of something that shows profit over 1 year of backtesting with at least a couple hundred trades?

Mike

Chris said...

Hi Mike

I can't be specific, because of client confidentiality, but one example may be a strategy based around opening range breakouts. You get trades most days with this method, perhaps not "at least a couple of hundred trades" a year, but not far off that number.

Big Mike said...

Thanks Chris. I tried closing bell and opening bell strategies as well but very few of them.

Appreciate your feedback.

Mike

Chris said...

Mike did you spend time optimising your strategy around the price action, time windows etc?

I know that you know what you are doing, but there are automated systems, and simple ones at that, that do work.

Of course, the banks and large hedge funds use highly complex systems that use many different parameters and correlations, but there is room out there for the likes of you and me to trade automatically, mainly by binning the incredible array of indicators that serve to tempt us and divert our clarity of thinking.

Luis said...

"I can also say that trading is even more liberating than being self-employed."

That's the point!
I listen a lot of people state that would like to be self-employed, so they don't have to deal with a boss. What about dealing with employees, and customers?

I prefer descritionary trading, but don't discard automated strategies, i see them as second source of income (Of course if it is a winning strategy).

Big Mike said...

Chris,

I have no doubt that there are systems that work, I just never found one that 1) had a draw down I could live with and 2) traded enough to convince me it wasn't curve fitted and 3) was backtested over a long enough period to convince me it wasn't curve fitted.

I know they exist. I just never created one, lol. So, I could be stupid, or I could have been so caught up in overly complex systems that I couldn't see what was right in front of me. Either are possibilities, creating a blog isn't that hard. LOL!

But seriously, if you care to send me an email with screen shots showing backtest results, I'd love to see them. No code or orders disclosed.

Mike

Sergei said...

I've read a lot in many places that simple Strategy could be profitable if it is linked with strict money management rules, even tossing a coin as an entry rule.
Another thing, market conditions vary with time, that means parameters should be periodically readjusted (optimized), That means you can't present back test results for very long period.

This is my current approach, still need to be proved.

Big Mike said...

Hi Sergei,

Where you exit the trade is the only thing that determines if it is profitable or not, too many people focus on the entry and not enough on the exit. So your point is very important.

As for re-optimizing parameters on an on-going basis, this is something that makes sense to me, and I have tried many times with self-adjusting strategies, but I am always weary of curve fitting them.

Mike

Chris said...

Mike

I can't send you any customer related screenshots as client confidentiality is critical to us.

My own automated strategy has only traded around 80 times this year so that won't meet your requirements, though it adds to my account nicely :-)

A project I'm currently working on seems to be producing very promising results and that has 2000+ backtested trades this year. I have only bolted a few good indicators together for this strat at the moment, we haven't even begun optimising it.

Continually monitoring and re-optimising an auto strategy is critical as its performance will attenuate as the market changes.

Its interesting that you spent time and effort trying to create an self adjusting mechanism. I imagine that would be more difficult to achieve than developing the original strategy.

roonius said...

I have developed a strategy for customer who wanted to enter at market at random times at random directions at keep trade open for a random generated time. It was a basic frame, there were additional MM requirements.
People are more and more believing that research must be done on MM - not on entry signals

Chris said...

Hi Mike

This thread got me thinking so I have just spent literally 15 minutes coming up with a strategy that needs a bit of tweaking, but actually should work after some tweaking and optimising for time, stoploss and profit targets.

The strat is to go long when the HMA(21) is rising on the 5min, 10min and 15min timeframes. It sells when the HMA(21) is falling on the 5 min time frame. Thats it!

There are no stoplosses, profit targets or other optimisations and the back test ran from 00:00 to 00:00 from 1st April to now on the ES.

You can see the strategy analysis results here:

http://www.medianetrix.com/images/nt_images/Big_Mike.jpg

The results are not very good at the moment, but look at the sharpe ratio, & profitable etc. It shows that with a bit of effort this really simple strategy could be traded and it only took me 15 minutes!

Big Mike said...

Chris,

I appreciate the effort. But 1500 trades would be about $7500 of commission and the profit was $900 bucks, so about a $6400 loss or so for this report right?

The average trade appears to be less than 0.01% profit?

Mike

Big Mike said...

Roonius,

Interesting -- random entries for random lengths -- I think we all agree the exits are more important than the entries, lol :)

I've tried a lot of money management ideas in strategies as well. Strategies that decide if they should keep trading "today" based on bunch of factors... strategies that adjust targets (exits) based on the profitability of the prior trades for that day, all sorts of things.

Mike

Neal said...

For consideration, John Clayburg has created a universal automated trading system that he presents at Trade The Market (Carter/Senters). It trades a basket of markets. Recently, I watched one of his free daily videos where he demonstrated their new approach to strategy optimization. Clayton has input maybe 30 or more combinations of parameters to consider, and has created a running daily line equity curve of all of the combinations together, differentiated by a different color. He runs the strategy pre-market and after the market open for 15+ minutes and determines which parameter combo is performing best in the current market environment, and assumes that one will continue to, and finish in the top combos for the day. It's a novel approach for daily optimization to an automated strategy.
I think I presented this accurately.

Big Mike said...

Hi Neal,

I watched his videos too a while back. There is also a free "beta" version of Universal out there with full source. But everything is for TradeStation only, something I refuse to do...

:)

As with most strategies, people had very mixed feelings (mainly negative) about this. Many posts about it on TL.

Mike

Chris said...

Hi Mike

Yes, the stratgey I created in a few minutes is far from perfect of course. There was no optimisation of any sort whatsoever, however, perhaps with some tuning it could form the basis of a simple, yet successful system.

My point is that simple systems such as I quickly created can form the basis of effective auto traded systems, with effective money management and tuning.

duck said...

I have come across an Indicator that has got my attention. It costs $1995 which is a serious amount of money. no holy grail promises but seems to have got my attention

and cover said...

...if someone asks 2 grand for an indicator!

Big Mike said...

lol nice one. :)

Mike

duck said...

yeah I know guys. sounds crazy huh?

Big Mike said...

What is the name of this holy grail indicator?

Mike

duck said...

here's a link.
http://backtothefuturetrading.com/Home.html
Ben from affordable indicators (you prolly know him) built it and says it's pretty amazing.
I am working on getting some discounts to those who may be interested. you still use your own entry and exit methods.

cunparis said...

Hey Mike, I agree with you on most of what you said but I think the real problem is you are making things too complicated. Like MikeTrend, your automated strategies may be too complicated as well. From what I've found, the simplest strategies are the ones that work the best. I've got two for ES, I'll email you the equity curves. I've got 1 for stocks which is nice as well. They're all very simple, however they're all currently in a drawdown but I'm confident when the market finally breaks out of this trading range they'll go back to being profitable.

One other thing.. if my automated strategies, I'm 99% sure I could do better if I was monitoring it and overriding it. However I work several days a week so this isn't possible. One example is I have a stock strategy that goes short on the open. It would be easy to wait and see what price does, if it starts shooting up then I could wait a bit. But I accept a lower return in exchange for not having to babysit it. :) That lets me focus on my main discretionary strategy.

It's all tradeoffs. I have several "systems" that I spent hundreds of hours on them only to watch them fall apart. Now I keep it simple.

sam028 said...

@duck: I've seen only one serious automated system. It was rented by an Quatari bank, for few millions $/month. So an "holy grail" for only 2k$, it's not enough... Just check their web site, are you really ready to trust people which such website ? It reminds me a video from a Forex robot dealer, so funny: "turn 500$ in 1.000.000$ in a month !!!", what a joke...

duck said...

Yes, it is hard to imagine such an indicator exists. From what I have been watching on this indicator,It does seem different in a way that it is "not" a lagging indicator. I'm a break-even to losing trader that has watched the market for 3 years now. I do not rely on a cross of a moving average to signal my entries or any other factor of an indicator. I am interested in seeing how this indicator gives information in advance before it really happens. Give me direction and I can take it from there. lol
you can email me if you have any questions.
ducktrader@gmail.com

Anonymous said...

There is one system exists that is very simple and nedds no indicator to look. just watch cycles. These cycles are generated by manuplaters. These manuplaters always working against physciology of retail traders. When retailers are buying, then at that time manuplaters are looking for shorts. Just do one pracice at opening of market . consider two markets. say Russel and ES. Where the looks in favour ,After watching overninght session. Russel is our profit target. Just for today. If we short russel for 2 contracts and long ES for 2 contracts. Then a profit target for $ 200 hit in 1st 15 minutes. This pracitice only possible with Strategy runner plate form. Because we have a option there for daily profit and loss targets. Once the hedge difference occur for profit. The profit target triggers and system close all positions. For June out of 7 working days only one day was loosing. If some one find some mistakes in writing, then sorry I am poor in english.

HPT said...

Mike,
I share your frustration with developing an ATS, but dismissing ATS's altogether as an impossible endeavor is not the answer. How can you compare the average newbie trader that supposedly loses 90% of the time to a trader that develops a profitable backtested ats?

"We know that 90% of traders lose money, and 10% take all 90% of the losers money. Now when you automate those odds are even more against you."

Roughly 30-50% of trading on the NYSE is automated, and the definition of automated trading in that regard is when a basket of 15 stocks or more is executed via a computer algorithm. Considering you can have a computer setup to monitor 1000's of stocks and execute a trade within 20 milliseconds (5x faster then a human can execute manually), how are humans at an advantage?

Why does James Simmons from Renaissance Technologies hire people with Phd's to create algorithms for his high frequency quant fund? Because ATS's dont work?
You said you've backtested many complex strategies. I assume you've tried making multi instrument/multi time bar based strategies and backtested on accurate tick data, as opposed to just backtesting strategies based off of 1 min bar data and a strategy based on a set of indicators that have "optimized" parameters. Having said this, I know successful automated traders, however these traders are skilled programmers(very smart might I add) and they spent several years developing there own ats. The first year they couldn't find a profitable system, however, they continued researching (doing market analysis, stats on specific markets), and they found a contract/stock that had an inefficiency yet to be exploited and now they trade this ATS everyday and make money consistently.

HPT

cunparis said...

Anonymous - can you explain how you know to go long Russel and short ES and not the other way around? Do you have criteria for this?

your english is readable, thanks for the effort. :)

shodson said...

I appreciate what Mike says, and can't dismiss his years of experiences/frustrations. As a professional software developer the allure of automated strategies is, well, alluring! I have some entries and exit strategies (scalping ES, initial SL of 8 ticks, sell 1/2 lot at 4 ticks profit, move SL to 4 ticks when 3 ticks is reached, let other 1/2 of lot ride to 12 ticks of profit with 8, then 6, then 4 tick trailing SLs as more profit is reached) that I've been executing on my own with a measure of success that I'd like to automate, starting with small lots and scaling up with success. If anything I find the exercise of developing the strategy with NT interesting and stimulating. But I take Mike's cautions to heart, not to get my expectations too high, yet I'm not looking for a "holy grail" 100% correct system, I'm willing to take some losses as long as they are managed well.

Michael Lomker said...

I trade some automated systems with MiniAnalyst but the draw downs would make most people's hair stand on end. It has been profitable over the last 18 months but there are long dry spells between equity highs.

Post a Comment

Note: Only a member of this blog may post a comment.