Thoughts on profitable trading systems
- This topic has 21 replies, 7 voices, and was last updated 4 years ago by
.
-
Hey all, this is going to be a long one, and its aimed for all the newbies out there.
I wanted to make a post that I would have loved to read when i was starting this journey.
In case you havnt seen me posting here, im still learning and still trying to create profitable systems. Im in no position to say “this is how it is” or something similar, however i thought i’d share some of my knowledge so far.
Let me start this by saying a couple of things, starting with some backstory. I got interested in trading and stocks 4-5 years ago. Been trying to trade stocks by myself, drawing in charts etc. And to be honest, that was (barely) profitable for me, but the biggest thing, that ruined all my bad trades, was myself. Somehow, i managed to fuck myself over and over again with bad stop losses, bad entries etc. I realized that what i was doing wasn’t too shabby, but ME as a trader is shit. I dont have the mental strength to pull off trading by hand. So i figured “i need to remove myself from the trading-equation” and i started reading about automated trading which eventually led me here. No more sleepless nights, no more “hoping” that the marked would go my way..
I work in IT and im very lucky to have a lot of time @ work to study automated trading. Sometimes i work with my systems and in general automated trading for 10-16 hours pr day, depending on how busy it is at work and home, and i have been doing this for over 3-4 months now. I guess im kinda hooked hehe. I’ve listened to all the (over 150) episodes of “better systems trader” where legends and pro’s are invited to share knowledge and the “how to’s” of the market. This has given me a lot of inspo and helped me grit through some hard times. You wont get a profitable system in your hands by listening to it, but it will tell you a lot about what NOT to do and how to do what you want to do!
A post in this forum that also helped me alot, and im sorry but i cant remember who said it, some swede i think, who said: “It took me about 50 hours until i created my first profitable system, and its a system i still use today”. I thought to myself, “okay, if he could do it in 50 hours, so can i!”.. I dont know how many hours it took me to create my first profitable system, but i would guess about 70-100 hours of studying and trying and failing.. My first system was a trend-following system for 10min timeframe on wall st (dow j). That system has been in demo for 3~ months and is currently 1400$ in profit with about 50% winrate, for now at least looking like backtest results (trading 1* 2$ contracts). I took that system and tinkered with it. its now customized and running live for 30 minute timeframe in wall st + dax and it has been profitable since day 1 so far. When i say profitable i dont mean that i have quit my daytime job… It makes 100$ here, looses 35$ there, wins 22$ there, looses 60$ here…. All in all its profitable and the risk is not too much at all. Basicly im not going to go broke by running the system for a full year. Even if the system would hit a bear marked, it should survive and even might collect a tiny profit. (based on backtest through a couple of financial crises.)
And this brings me to point #2. Ive seen many people have been asking/talking about these things in this forum but it cannot be said enough:
- To all of you asking “ARE THERE PEOPLE ACTUALLY MAKING MONEY DOING THIS??” Yup, yes, indeed.
- Is it easy? no.
- Will it take a serious amount of time? Yes.
- Do you need to be interested in this topic to keep up the stamina that it demands? Yes, if your passion is something else then just stay away. You wont get rich from reading this forum 1 h a day alone.
- Can i copy a profitable system directly from this forum? I would recommend that you dont do it. At least have it in demo for a month or 2..
- Will i make alot of money in year 1? Probably no, and by that i mean im 95% sure you wont make “a lot” of money. But will you be profitable? Well if your not stupid, play ur cards right, have patience and study hard, then yes, you will probably be profitable.
Expecations:
One of my first questions on this forum was something like “If you have a profitable system, why not just rock 100 contracts and become a millionaire over night” well, i realize now that this is not how shit work. You NEED to have your risk of ruin basically less than 1%… You have to think about stuff like “If i have had 1000€ as biggest drawdown the last 10 years, i need to make sure that i can loose 1500€” Just because max DD in backtest is 1000 does not mean you cant go over that, and it doesnt mean its not a profitable system. Just because it has 5 loss in a row in the backtest, does not mean it wont loose 10 in a row during the next month, but still it might be a profitable system! Backtest should give you an indication of what to expect, but you will never see the exact same numbers. You might never see a 600€ drawdown! You might see a 6000€ drawdown and a busted system..
People expect too much from automated trading / trading in general. People think they can quit their jobs in 1 year and trade from a sandy beach all day long. And i thought this myself when i was new to trading!! My biggest dream was to wake up and having earned money while sleeping. It doesn’t get better then that, right?? Well it doesnt work like that, it does take time, it does take dedication and it does require some level of passion. If you do not have passion then it will get boring real fast..
Lastly i want to add the following tips:
- You cannot predict the market, so never try to. Never say “im going to pause the system now because i think tomorrow is going to be good/bad <- If you know how the market is tomorrow then why are you not a billionaire? Because you dont know.. So dont think that you might know.. you dont..
- Dont try to re-invent the wheel!! Meaning, you as a single person, probably cant find that “hidden combination” using the same standard indicators that everyone else uses. Let me beat this point into your head: There are people (not just giant FIRMS) but single, solo trading people that use machine learning like this: “Test absolutely everything using these 10 indicators.” and the machine will try every combination possible. As a single person in PRT with no machine learning at that level, you cannot try EVERY possible combination, in fact PRT does not even have multiframe trading (you cant call a daily RSI in your 1h system..) Meaning your probably not going to get success throwing 10 indicators in a system, optimizing every variable and getting 90% winning trades..
- From listening to the podcast thers alot of talk about pros running systems with less then 50% winrate even. Momentum/trend following systems, long only, in stock indicies will only make money when the market is going up, and your going to get false signals a lot. But if you take EVERY trend going up, and trying to reduce false signals, and implement money management, a lot can be done! 🙂 You dont need the 80% winrate and 10 in gains.
- There is no 1 holy grail system. Try to create many systems, doing different things! Running 10 systems with 50% winrate is much better imo then 1 system that has 90% winrate. Because the 90% winrate is probably going to fail, but if you have 10, then 2 might fail, 1 might do VERY good, and the rest might be profitable..
- Make the systems as easy as possible. Everyones dream is that smooth 45 degrees EQ curve, but thats not happening. What you need to do is trying to get that curve profitable and robust in the long run, but zooming in will show you that the curve is full of dips and bad trades etc. It might not be profitable for a whole 3 months even! but the system it self might be profitable in the next 10 years…
I’d like to wrap up this post with one of my favorite quotes from the podcast, and im paraphrasing here:
“I actually dont believe that profitable systems “stop working”. If you make a robust momentum system, you might see periods, maybe even a whole year, where the system doesnt make you any money, but then over the next 2 years it might be your most profitable system. If you make a system that profits from every trend, but loose money on every false signal, you just need to make sure that the good trends are more profitable then the false signals in the long run.”
This was really an eye-opener for me. Basicly its telling you not to try to chase 90% winrate and every trade is a good trade, but rather that its okay to loose, its okay to see bad periods. everyone gets them.. But just make sure that statistically, say over the past 10 years, all your good trades are better then all the false signals you loose money on between the good trades.
Edit: About sharing codes in this forum, i dont think people are going to share their most profitable system, because it takes alot of hard work to get there! But i think people, myself included are more then willing to share good ideas that people might continue working on or people might give you some insight, or inpso to do something different or use a bit of your code. Sharing is caring.
Total of 25 users thanked author for this post. Here are last 10 listed.
Fantastic post Jebus89….. and very long so I guess if a reader doesn’t have the time or dedication to get to the end of it then they don’t have the time or dedication to be a successful trader!
I just kept reading and nodding my head and muttering ‘Yes – that’s right’, ‘Yes – I did that’, ‘Yes -that’s what I do now’ etc
I would suggest that anyone new to trading or just hoping to get rich quick reads your post and reads it again and again and again and makes notes and reads them every day to remind themselves as there is so much good solid advice in there about how it really is, what to expect, and how to probably get to where you hope to get to with your trading.
I went through exactly the same phase with buying and selling stocks and shares. Well I say selling but I could never actually bring myself to press the sell button which is where it all normally started going wrong. Robots don’t have that emotion problem.
I think it is going to be difficult for others to post further on here as you have said it all already but I hope I am wrong.
1 user thanked author for this post.
Haha Vonasi! Thanks for the reply and im glad that my feelings are shared by other great members here!
And yes Vonasi, I hope that i have said enough to make a newbie get a better understanding of what it’ll take to get a profitable trading system. Again, i have only had a profitable trading system for some MONTHS now so im in no position to say whats what.. but its good to hear that you agree 🙂
If i might add just a final thought its this: Even with automated systems i STILL SUCK balls sometims 🙂 I have lost money on profitable systems just because i kept pausing them because i thought that the marked would turn etc. So ill leave u all with another great quote from the forum “Press play and stay away”. You should not be having automated systems that make you nervous! You should not be in a position where your considering turning your system on/off for the day. Its either on or its off. That easy… Unless you have some god-given indicator that tells you if it should be on/off, but if thats the case you should just automate that into your system and voila hehe. Never ever trade on “hope” and “feelings”.
And i cannot stress enough how good the podcast is, just read what a random episode is about:
“Larry Williams has been trading futures and stocks for over 50 years.
In 1987 he won the World Cup Championship of Futures Trading ®*, turning $10,000 in to over $1.1 million in 12 months, that’s a cool 11,000% return and the highest return to ever be achieved in that competition. 10 years later his daughter won the same competition with a 1000% return and his students have won the competition in many other years too.
He is a published author, with a long list of best-selling books and has also created a number of market indicators including Williams %R, Ultimate Oscillator, the Williams Accumulation/Distribution Indicator, COT indices, cycle forecasts, market sentiment and value measures.”
Its an interview with one of the BEST traders ever.. If you dont think you can learn anything from that, then think again 😉
1 user thanked author for this post.
Thank you so much!
Understanding the actual risks of blowing up ur account is key to understanding how to make money in this place… If ur account blows up ur never gonna make another dollar.
1 user thanked author for this post.
Hey, writing long posts dosnt take that much time for me haha, i just write them out as i go and 10 min later its a massive fucking wall of text that hopefully someone might like hehe.
I have many different strategies now, some are just simple candlestick patterns with only 1-2 filters like MA filter, others use no candlestick patterns and just rely on pure indicators.
I honestly think anything can and will and wont work. Super simple 1-2 conditions vs more complex 6-10 conditions, i have them all, and i feel like and so far its been pretty even between “which ones work best”. My out of sample/time live isnt that much just yet so i might have just been lucky with my results so far.
That being said i have stopped 2 systems and they where both on the more “complex” side of things (using more indicators and more “degrees of freedom” which ofc increases chances for curvefitting)
I think simple is better because its less chance of curvefitting. I would say every book i have read have authors that like to use both patterns in chart combined with 1-2 filters from indicators…
I also wanna add 2 things that i rly look for when designing systems:
- that i only run systems that look good say 90-95% of the time in backtest. Like i will not accept something that has 3/10 years = red. And 80-90% of the quarters should also be green if i am to run it. (dont get me wrong, the systems has loosing days, weeks, months and even quarters! 🙂 )
- it needs to look decent in other similar markets. If i make something for Wall st, i want it to be looking OK in for example US tech 100, or DAX. If it looks really good ill run it in multiple markets, but it should definitly not go into hardcore red loosing mode, that would be a big red flag for me.
Hi Jebus,
Just interesting to know on which back test criteria’s your strategies pass.
1. Which profit/loss ratio (for me min 1,35)
2. Which TTM (Time to Market) (for me preferably less then 45%)
3. Win rate (for me above 50%)
4. Acceptable maximum draw-down (it is of course still only a back-test) versus average exposure (I think less then 6 % is too good to be true ???) (Example DAX 1 Euro +/- 10.000 exposure, 4 x 1,5% Stop loss losses)
How do you work with IS/OoS (I use 66% IS/34%OoS on a longlasting period) ?
Looking forward to hear your thoughts.
Stay healthy in those though times !
And a good weekend.
1 user thanked author for this post.
Hellon @Jan and Hello @Vonasi!
@Jan: I dont really have a firm “my strategy must be above this or that” criteria, but i guess i would say 50%+ winrate and above 1,3 ratio.However, lately ive seen that my backtest and live results might vary and that 1,3 is too low. I would say majority of my systems are 50-60% winrate and 1,5-2 ratio.
About “Time to market”, i have some thats 60%+ and i have some thats like <15%. And to be honest im not really sure whats “better”. I think that if you a high number of trades combined with little time to market, could make for a more robust system. However, one of my most robust system is more than 65 % in the market.. That being said i managed to filter out a bunch of losses and its now at 55% looking very good.
When it comes to maximum drawdown, for im trying to keep max drawdown at -1000€ (+/-200€) with 1€ contracts.. (so with a 10.000€ account this means 8-12%.) I have seen some systems thats been down (live) -1200 and are today positive, and ive had systems that went -1500 and just never recovered, turned off now.
About the In sample vs out of sample question, i have explained in another post i made a long time ago (which i cant seem to find now) how i make systems: I like to start off at a random point in time, either at a nice reversal or a nice trend, and just stare at it and try different things/ideas. I might test some candle patterns, i might make some simple indicator, i might try some indicators from this librabry etc. Im trying to extract profit from that single trade. I dont care about winrate or anything at this point, im just trying to get some profit out of the market in that single event at a random point of time. My code will be very rough, and not optimized more than what i can obviously see with my eyes (so no optimizing tools used) and very little or no money management. This is so when im using more data i havnt curvefitted for that 1 data, i got something that looks rough but profitable.
When i see that the trade im trying to capture looks good, i start to include more data. I might expand to data for that week/month/year (depending on timeframe) and ill try to filter out as much noise and losses as possible, when i have something that looks pretty good on that tiny % of my data, i then check it on 100% of my data to see what i got. If it looks like something i can shape into something very good, ill start working on it. I then use 40-60% of the data as in sample and i start optimizing. I might use the data from the middle of my data, but i like getting the 2008 crisis if its possible. I also like to include rough and hard data as well as “easy and good” data if that makes sense. Usualy tho its the first 50% of the data im using. If im optimizing on the 50%, then check 100% and it looks good, i do 1 final optimization on 90-95% of the data with just minor minor variable changes to see if i can fine tune it to more recent as well as the old data. Money management (like optimized stop loss, or a trailing stop loss, or even profit targets) are added lastly. I think that in my head money management is a scary thing that can screw up your variables for your indicators/lookback periods etc. So i would rather Optimize my money management to my already “finished” code, rather than optimizing my code to fit my money management. If that makes any sense to you 🙂
@Vonasi Im not sure if i agree with you on the “if a random guess = 54% why would u wanna settle for 50%”. Lets take trend following systems as an example, most books ive read on that says that with a very basic trend following system, your gonna see like 30-45% winrate, but some of those winners will win big and make up for all the losses and put you into profit. However, many of the smaller wins will only even out your equity curve, so your always waiting for the big ones to finally show up. At least for me this is true in many of my trend-systems, however, ive managened to include trailing stop losses in my trend systems that have upped my winrate from maybe 45-50% to 55%-60% and a bunch of the smaller wins are not just evening out the equity curve from losses, but they also get a small increase in the equity. But im still a slave for those big trends/swings that make the big bucks for those types of systems.My point being, if you made a system that has a 35% winrate but beats any underlying markets you test it on, by riding trends, you would probably run it! Many Small losses, few big wins.
Another thing about winrate is the tail risk attached to each system.. I ran a mean reverting system that had/has 75% winrate, but small wins, and big losses come back to bite you in the ass when you need them not to.. when the final global equity drop of -30% happened over the last months, i took the biggest losses i have ever taken.. I effing hate mean reverting systems because of the tail risk lol.. so i would argue that win rate dosnt actually mean alot. You can run 30% winrate systems and make big bucks and you can run 75% winrate systems that can ruin your account if your not careful. I know your no dummy and you already know this Vonasi, i just wanna highlight it to people who might not know this!
Find exclusive trading pro-tools on 
