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)
Just curious, how many lines of code is your ATS? (minus deadspace?) I feel that we're comparing apples/oranges. My strategies don't require thousands of lines.
Again, if you're able to generate a strategy in a few hours that works for you, great. That's not been my experience. Then again, I have a little bit of OCD when it comes to performance. I can always make the strategy better, if that's simply reducing drawdown or reducing the total number of trades (increasing profit/trade) or time in the market.
A strategy that features a 5%/day return might have a lot of volitility and risk as opposed to the same 5%/day strategy.
I'm always looking to make my strategies better. If nothing else, it helps me to stay atop of the varying market conditions. Again, if you have one that works under all conditions, has acceptible risk/drawdown, gives you desirable profits, then my hat goes off to you....
What kind of adjusted profit ratio are you seeing? What kind of RINA index values?
Generally the programmimng is the easy bit and that takes only a few hours because all the major work has been done beforehand.
I'm happy to give an example of the sort of ATS that I use and the thinking behind it:
Here is a of the top of my head trading strategy. I don't use this and don't know if it is profitable or not so please don't trade this example without researching it deeply.
Take simple price action around VWAP and answer these sorts of questions:
How does price behave when it approaches VWAP from above?
How does price behave when it approaches VWAP from below?
What is the percentage of times it bounces off VWAP? If so, how far does it bounce?
What is the percentage of times that prices goes straight through VWAP?
Does price pass through but then retrace? If so by how much?
Does time of day make a difference?
Does anything else make a difference (ie above/below open price)?
etc etc
So you research and answer questions around simple price action like that and if you find more winners than losers and can apply suitable money management and trade management, then you know that you have a strategy worth developing into an ATS.
Lets now look at the ATS:
Does it use indicators?: one VWAP
Does it need optimisable parameters?: No unless time is one, there is nothing to optimise with VWAP
Is it simple to code?: Almost certainly
Is it complex?: No
Does it need backtesting?: No because you have already researched the strategy
Does it need forward testing/paper trading?: definitely to ensure it works as intended and to make sure that it delivers in reality what your research has found
I hope that helps show what I mean. There is nothing complicated there. In fact there is a real sense of anti climax if it works because there is nothing to play with or tweak. You turn it on and let it do its own thing. Boring, but if it works you make money and that is what it is all about (to my mind, at least).
Why are you limiting the question to manual traders when the topic is 'allure of automated trading'
Jon
Writing to you from the wonderful province of Ontario, Canada. Home to the world's biggest natural negative ion generator, the Niagara Falls, and to those that dare to know how to go over it in a barrel. SALUTE!
I can imagine that many auto traders are fairly well versed in backtesting and have tested over pretty large sample data sizes. I'm sure there are some who don't test properly, but I'm more interested in how manual traders/testers do their testing, since, for one example, it can't be too easy to manually backtest the last ten years of the S&P.
I trade manually. My view of backtesting is that the results are very fraught and its really easy to arrive at biassed or wrong conclusions from it because of hindsight bias and subtle or not so subtle curve fitting.
To develop a trading method I follow something like this:
1 come up with the method rationale - it has to make sense to me, has to fit with how I view price action
2 write up some initial trade rules/guidelines
3 manual backtest a sizable block ( say 100 trades) and refine rules if required
then I go to forward testing with no further backtesting - so basically the backtest is simply to establish the initial trading rule set. Much more time will be spent forward testing/simming and it is here that the method has to prove itself.
One thing I dont like about ATMs is how much of a time trap they can become - and personally I think my time is better spent elsewhere. Thats just my view of course.
I don't speak for all auto traders, but I think you're assuming that people use backtesting alone. Backtesting simply screens strategies that aren't worth forward simulating. I know that if a strategy doesn't do well in a backtest, it surely isn't worth forward testing, so it saves me time by screening lost cause strategies and optimizes my time. I then take the strategies or modifications that seem to work and then forward test them.
Unfortunately, most of my strategies aren't able to backtest, as I trade inside the bars and tick by tick (if nothing else than money management).
A) I think you're assuming others are at the same coding capability as you. For some of us, it's not that simple.
B) I have several strategies that seem marginal to ok at first, but then with some massage, they can become viable.
For example, I have several strategy variations that feature nice string qualities. A couple of them feature a 65% win rate and above an 80% win rate immediately following a loser. By simply increasing the trade amount following a losing trade, I can take advantage of a higher win%. That's a "simple" modification, but an example of how you test, modify, retest in order to manipulate certain performance factors.
That type of modification won't increase my daily profit %, (actually decreases it as a function of account size required) but helps the profit and adjusted profit ratio and can under some circumstances, help to reduce drawdown (or increase it). Similar analysis can be done on strings of winners.
It's not that the original strategy is complicated, mine are fairly simple concepts as well, but the execution can get complicated. Money management is usually the most complex part of my coding, as I prefer to execute trades (or at least manage them) tick/tick rather than end/beginning of bar.
I also like to have the flexibility of executing limit or market orders for the same strategy, which can have an effect on slippage.....again, all these tweaks and variations are the same general concept of when to enter a trade, but vary by trade amount, trade time, profit goals, loss tolerance, etc.
A dog of a strategy can be profitable if you simply mess around and optimize a simple bracket order. Then you might find that for certain market conditions, a larger profit goal is preferred and/or loss limit (or the opposite).
So basically, what starts as a very simple strategy can be tweaked and optimized and become more and more sophisticated/complex (for execution).
I fully agree that price-based strategies works better than indicator-based strategies (especially the standard momentum indicators). I have also found that the profitability of price-based strategies can also be greatly improved by looking at when and how to sell. Using an appropriate trailing stop exit strategy can improve results tremendously.
These questions have been nagging at me so I thought I'd post a quick reply. I am a discretionary trader but also an empirical thinker who normally prefers to rely on hard evidence when making decisions.
I have often struggled with the thought that my trading should be based on a verifiable "edge" that can be demonstrated statistically. At various points in my trading career I've stopped trading completely in order to give myself time to gather data and demonstrate this edge. I've logged thousands of hours into the manual analysis of a bunch of different systems.
In the end, I don't feel that this testing has been particularly helpful to my trading. When real money is on the line, I simply don't trust the results of the testing. I've observed that market conditions change much more frequently than I thought when I first started trading, with has created this huge nagging cloud of doubt every time I'm trading a tested system that starts to run into a drawdown phase. I can't get past the eternal question for mechnical traders: Is this just a routine drawdown or the beginning of the end? Psychological weakness maybe, but I've eventually come to the conclusion that I couldn't trade a mechanical system even if I wanted to.
My discretionary trading is too nuanced to be tested in any meaningful way. I do think that some degree of testing is helpful, but the purpose served by this is more to develop tactics that work in different conditions than to develop statistically significant evidence of how any given trade is likely to turn out. I've given up on that latter question as a pipe dream.
Not sure if this makes sense to anyone else but I guess my basic answer is that my discretionary trading is not testable, and even if it were testable I wouldn't trust the results of the test. Admitting this in a public forum gives the empiricist in me an identity crisis but there it is.
Like I said earlier, the one thing I've learned from trading is that strategies, mindsets, rules, philosophies, etc are about as varied as people are. The traders who are successful, stick to what works for them.
You and I have the opposite fear. I tried trading manually. I first simmed and no matter how hard you try to accurately replicate live trading emotion/actions, it's impossible. Once real green paper is involved, it makes it VERY stressful for me...particularly for money management and particularly AFTER a losing string. I begin to question if what I'm doing is correct. I stay in trades too long (instead of taking profit and leaving), etc. Only through reducing it to a systematic DISCIPLINED approach, could I not drive myself crazy.
Now I know it's just a numbers game. If I stick to the plan, I know it comes out in the wash. Where I got hurt when I first started was changing strategies. If I'd lose a trade and the next started out as a loser, I'd start wondering if I should get out now...I'd get out, only to see it come back my way and I would have made profit. You can get really off kilter and get into a bad "frequency" or out of tune and spiral into oblivion.
I think the most ABSOLUTELY tortorous thing in the world is to A) Watch the market dance just below your profit limit or B) just above your stop for what seems like hours. As with most of you guys, I've actually had a trade where the price was at my limit order for almost a minute before it filled, the whole time I'm yelling "just do it already!" (and wondering, should I just move it down, should I just move it down?")
It's too un-nerving for me. I know now why so many guys smoke a pack a day.