Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Why do people assume that automated systems must be "incredibly complex"?
My ATS is consistently profitable no matter what market volatility is like, trades every day and yet it took me less than two hours to program.
If I develop an ATS that starts to layer complexity on complexity it goes straight into the bin as the basic principle must be wrong and adding fix upon fix just results in curve fitting.
I accept though that finding a simple system is not easy and my experience is that researching, analysing and monitoring an ATS is still a lot of hardwork.
IMHO people go quickly : Oh, you worked for years on an automated strategy, and you made it incredible complicated, and it didn't worked , so: It's not working for a retail trader.
I think following this thinking is wrong.
I can't make money on the 6A for example, does that mean nobody makes money on the 6A, of course not, but does that mean that there are more losers on the 6A than on for example the 6E because I didn't made money on it ? Of course not!!!
Platform: TradeLink, OpenQuant, considering anything that works...
Trading: if it trades...
Posts: 94 since Oct 2010
Thanks Given: 24
Thanks Received: 39
Mike,
Thanks for sharing your view so concisely.
Why are you so anti-automated trading? And I don’t mean the logical reasons you have listed on your blog post and in the first post this to this thread. There appears to be something personal to it?
I couldn't enter a trade unless I'd tested the system.
The sort of systems I trade are only tested to my satisfaction on 2-3 years of data and hundreds of trades
There's no way I have the time or patience to do that by hand
So I automate the system for testing
(you can see what's coming)
Once I'm happy with it, the only way to guarantee I do the same thing live is to run it automated.
Given all that, it's rather obvious why I have to auto trade.
Now, I could choose systems that are manual, and I used to do that. But I soon realised any success may have been largely down to luck. After all I didn't have much testing behind me, didn't have the patience to paper trade it for six months, so what was I really basing my expectations on, and how could I say each month was better or worse than expected, since how did I know what the expectation should be? Also, how could I say what my max drawdown might be?
So, my questions to the manual testers/traders are (and I would really really like some people to answer as I'm genuinely interested)
How much testing did you do?
How many trades in your testing?
Are you aware of how to calculate the error resulting from sets of data? Here's a simple example
Let's say you tested with 60 trades, and you've been trading for three months, successfully, at 15 trades a month. Everything looks rosy right? You have 105 trades and your average success rate is, say 60% A simple calculation of the error factor in your average is 1/sqrt(n) which is 1/sqrt(105) = 9.5%. A crude confidence interval is plus or minus double the error factor. So your 60% average is correct for your system, plus or minus 19%. So, three months in, it's going well and you double your stakes. The rest is history.
I'm on the side of automation. If I haven't automated it, I don't trade it. That said, I don't think HFT is valid as an argument by example, for the simple reason that the techniques the 'big boys' use can not be profitably applied by us retail traders. (Due to slippage and commissions they don't have.)
Complexity comes not just with the entry/exit strategy, but adjustments for volume, methods for money mangement, etc. Autotrading allows you to employ very sophisticated money management strategies that would be too consuming to calculate manually. I came up with the basica principle of most of my strategies very quickly.....the hard part comes after....
Like most auto traders, strategy performance is all about optimization....
Let's say you have a strategy that's profitable, both forward sim and backtesting, but it has too much drawdown, or maybe it spends too much time in the market (i.e. you'd prefer a strategy that trades 10 times a month, rather than 100 times a day), then you "tweak" the strategy to optimize those features.....
Additionally, I find that the true secret to raw profitability with any strategy is not increasing gross profit, but reducing losses.....how do I remove the losers but keep the winners. THAT is what ends up making the typical auto strategy very complex. I can test, squeeze out more profit, choke down more loss, reduce drawdown, increase entry and exit efficiency, etc, etc, etc all with a few minor modifications to entry signals (or modify how/when to exit).
For instance, you might notice that your strategy is breaking down under certain circumstances and see a consistent pattern and then code to make sure that it skips entry under those conditions (and take your profit factor up a few notches). Or, you might notice that your biggest loser was because of a certain set of conditions, so you tweak that.....
If you've been able to simply develop a strategy that works and makes money reliably, then kudos to you, most of us have to work and tweak it. I can tell you that I have strategies that started out pretty modest (adjusted profit ratios around 1.00 and RINA indexes on order of a hundred or so) and by tweaking and improving them, i've squeezed out a little more profit, but I've really reduced risk and drawdown and upped the performance/risk indexes.
I was thinking more about this topic on the way to breakfast...
I think one reason this tends to become a 'religous' argument stems from a failure to distinguish between two different kinds of 'edge'.
Using 'Big Mike' as an example (because I suck as a discretionary trader), I think he would have both an objective edge, and an intuitive edge. His objective edge is his 'rules'. These can be taught, and they can be programmed. I believe I am a good enough programmer that if I was taught his objective edge I could implement it in a strategy, and could execute it (the objective component only) better than he can!
However, I think he also has an intuitive edge, which comes from experience and possibly innate talent. I believe this can not be taught (but can be learned through experience), and more importantly to this topic, can not be programmed.
If this dichotomy is true, the trader who has a strong intuitive edge will not benefit from ATS vs discretionary trading. The programmer who has no intuitive edge (me) needs to either develop one, or stick to ATS trading.
I completely agree. I have a good friend who's a very successful equity guy and understands the complexities of options. His brain works differently than mine. He's been successfully trading since the mid 90's and he's seen a lot of different styles, etc. One of his big wisdoms is that there's 1000 ways to skin a cat, the trick is figuring out which one works for you.
As you pointed out, some guys have an intangible "intuition" which is really a way of describing someone who can recognize a lot of different indicators/conditions and have a feeling based off experience about how to act (or not act).
I partner with a professional trader with 20 years experience in trading and managing large trading teams. He describes his skills as "trading savvy". He loves my automated system and in fact I trade some of his capital with it through an IB financial advisor account.
I benefit because he has natural talent as a trader and he encourages me to become less risk adverse and helps me understand the market and how institutions place and manage trades.
By combining our talents we are on the verge of something that could grow quite big.
I am not so sure about this statement. My main strategy has nothing to tweak. Parameters are time to trade and order quantity that is all. If the stop gets taken out so be it, but I know that statistically the system wins more than it loses and the money management is fine.
Over time it makes money, hence my partnership with the guy I mentioned above. There is no way he would risk his capital in an over optimised system. In fact he encouraged me to take out complexity that I was tempted to add to reduce stop outs.
We were discussing how locals trade index futures just this morning as I want to automate their methods. They use a really simple approach, coupled with good (but not complex) money management. That is what our next ATS will do. It will not take many hours to program.
My approach to developing an ATS is to look for price action that keeps recurring with more wins than losses. Build in some money management - stop loss size and placement - and trade management - when to move to break even, trail stops etc, - and that is that.
As long as your research into the price action is accurate and comprehensive and your system works as intended then you are set. I do not spend hours backtesting.
Try not to use indicators as that is when the fun starts with optimisation and your simple price based strategy suddenly becomes complex and on the path to failure.