Success with 100% Systematic Trading?

[Allistah]

Hi there..

I was wondering if anyone is having any success with trading using a 100% systematic approach with automated trading?

I think that just getting any system and backtesting it so it's profitable and running it forward is asking for a disaster. I do believe that a quality strategy with proper backtesting and forward testing will yield a positive outcome. Problem is - I've never had the tools to do this and have never had anyone share their real life experiences with me on it.

I would be really interested in hearing from people that are making this approach work and what they found to be the biggest reasons it is working for them. I believe that the actual strategies that are used are not a big part of it like most people think. I think it's -how- you test and employ those strategies that will make you or break you. Thats just my opinion at this point. I have zero data to back that up so this is why I want to hear from others that are making it work.

Thank you for taking the time to read this post and again if you decide to help me out by sharing your experiences with me and the others that will read this.

-Allistah

 

[money_13]

Hi there..

Thank you for taking the time to read this post and again if you decide to help me out by sharing your experiences with me and the others that will read this.

-Allistah

I have been doing this for over 3 years now. Day trading eminis. But I disagree that actual strategies are not so important. They are, your insight is vital and your tools as well. I am a physicist turned trader, so I have a good background in analytical tools, but I still think your insight matters more than backtesting, the latter is what most people can do, the insight is something that only you can contribute.

Well, anyway, good luck. You mileage may vary, but it may be worth trying.

 

[Allistah]

I have been doing this for over 3 years now. Day trading eminis. But I disagree that actual strategies are not so important. They are, your insight is vital and your tools as well. I am a physicist turned trader, so I have a good background in analytical tools, but I still think your insight matters more than backtesting, the latter is what most people can do, the insight is something that only you can contribute.

Well, anyway, good luck. You mileage may vary, but it may be worth trying.

Hi there..

Thanks for taking the time to reply. I have worked in the software industry for 21 years now and I have done a lot of programming over the years. I currently program all of my strategies but have not had access to any tools that would allow me to back/forward test them correctly. I am much more of a scientific type of personality and would prefer to take that sort of approach to my trading.

Is the majority of your trading been with automated strategies? If you're successful, what do you think have been big reasons that have led to your success?

 

[WinAtTrading]

Hiya
I trade 2 systems, one is a systematic trend following, but with some discretion. I am averaging about 40% a month.
Another is a Expert Advisor that is created by a programming friend, that has had wild swings, but is currently up around 10%.
I believe it is possible, only becuase I have experienced it.

Somesh

 

[Allistah]

Thanks for the reply. I appreciate you taking the time. Can you explain what an EA is? I have done a lot of software programming as well as programming of strategies.

How do you use your discretion to influence your 40% strategy? Do you think you would be able to take it 100% automated or do you feel like that bit of discretion is what makes it a winner?

 

[WinAtTrading]

Hi Allistah
An EA is an expert advisor, which is a strategy that when programmed into MT4 runs on auto pilot. Therefore I have the program running on the Mt4 account and it takes trades, sets stop losses, sets profit targets based on the the settings we have programmed into it.

My trend following system is rule based, but I also use my discretion to take trades and to take profits. Sometimes I close 50% etc depending on the market conditions.

What strategy are you using?

Somesh
winattrading

 

[kimo'sabby]

Hi there..

I was wondering if anyone is having any success with trading using a 100% systematic approach with automated trading?

I think that just getting any system and backtesting it so it's profitable and running it forward is asking for a disaster. I do believe that a quality strategy with proper backtesting and forward testing will yield a positive outcome. Problem is - I've never had the tools to do this and have never had anyone share their real life experiences with me on it.

I would be really interested in hearing from people that are making this approach work and what they found to be the biggest reasons it is working for them. I believe that the actual strategies that are used are not a big part of it like most people think. I think it's -how- you test and employ those strategies that will make you or break you. Thats just my opinion at this point. I have zero data to back that up so this is why I want to hear from others that are making it work.

Thank you for taking the time to read this post and again if you decide to help me out by sharing your experiences with me and the others that will read this.

-Allistah

The markets are in a constant state of flux, therefore you can only gauge the market, as opposed to actually reading and trading it 100% systematically, for any long periods of time.

 

[resistance]

If the market would be moving according to technical things all the time then we can make a system that will work 100% but these are the fundamentals that drive the market and set the new levels, so 100% automated system is not possible.

 

[random12345]

If the market would be moving according to technical things all the time then we can make a system that will work 100% but these are the fundamentals that drive the market and set the new levels, so 100% automated system is not possible.

It is possible, it just can't be left on for x years from day zero. It needs constant updating and adapting over relevant samples. However core libraries and principles tend to stay the same... my systems are regularly optimised but I haven't altered some of the core code in years.

 

[Intel_Bvc]

Hi there..

I was wondering if anyone is having any success with trading using a 100% systematic approach with automated trading?

I think that just getting any system and backtesting it so it's profitable and running it forward is asking for a disaster. I do believe that a quality strategy with proper backtesting and forward testing will yield a positive outcome. Problem is - I've never had the tools to do this and have never had anyone share their real life experiences with me on it.

I would be really interested in hearing from people that are making this approach work and what they found to be the biggest reasons it is working for them. I believe that the actual strategies that are used are not a big part of it like most people think. I think it's -how- you test and employ those strategies that will make you or break you. Thats just my opinion at this point. I have zero data to back that up so this is why I want to hear from others that are making it work.

Thank you for taking the time to read this post and again if you decide to help me out by sharing your experiences with me and the others that will read this.

-Allistah

Allister,

From my personal experience, the ability to successfully backtest my system and quantify its "past performance/profitability" was a big turning point.

I have created a Neural Network based system that is 100% automated and for nearly 3 years whilst i learnt about FX and developed the system I continually lost money despite "thinking" the system was amazingly accurate.

I was always looking at the output charts that mapped the movement of price perfectly but somehow I always seemed to lose money.

It was only when a Futures Trading friend of mines put all my trading signals in an excel spreadsheet and worked his magic (calculating drawdown, std dev, cumulative returns etc) that I made some massively important discoveries that turned things around for me.

Firstly the stats proved that instead of me picking and choosing what signals to follow, and when, the stats suggested that i needed to trade every signal was generated.

Secondly, I was able to see the maximum drawdown (run of losses) that I would have experienced had I traded over the last x years. This is important because you can get an idea of how much a "winning" strategy can actually lose you should it experience a rough patch.

Thirdly, I was able to see that adding Stop-Losses and take Profits to my trades were not always the optimal things to do. At the end of the day, stop losses can limit your losses but they can also lock in a loss. If you do the analysis and you're confident in the system and its proven stop losses reduce your profits then don't use them.

Finally, I was able to calculate the optimal hold time for each strategy when applied to different currencies. Not all strategies work on all currencies. I have the cold hard facts that enable me to make an informed decision on what strategies to run on which currencies.

Obviously everyone has their way of doing things, I've been fortunate enough to know someone who works for a Quant Fund and he's been able to pass on some of his knowledge and I've applied it to what I've been doing and its definitely helped turn around my fortunes.

I'm now armed with an idea of exactly how much I'd expect my system to generate me over the next year, and I've got a decent idea of the maximum drawdown I could potentially fall victim too. This gives me confidence in the system and ensures even after taking some losses I don't get panicked and turn it off.

Hope that helps!

 

[fatowl]

I have created a Neural Network based system that is 100% automated and for nearly 3 years

Intel, this rocks. I'm a big fan of neural networks and have had exposure to them in engineering applications outside of trading. I want to play around with some low-level neural networks for price prediction, but can never find the time to really delve into it. Kudos for getting a system developed yourself! I've got my own mechanical system I'm happy with for right now, so neural networks would be a fun side-exploration.

Can you give some stats about your system? Things like win/loss rate, average win percentage, average loss percentage, expectancy, max drawdown (backtested or live), etc.

If you're skittish of the public forum, a PM to me would be fine.

 

[Brumby]

From my personal experience, the ability to successfully backtest my system and quantify its "past performance/profitability" was a big turning point.

.... except it doesn't reflect an outlier. Trading without stops may improve the trade performance but combined with an outlier is what may kill an account - eventually.

 

[Intel_Bvc]

Intel, this rocks. I'm a big fan of neural networks and have had exposure to them in engineering applications outside of trading. I want to play around with some low-level neural networks for price prediction, but can never find the time to really delve into it. Kudos for getting a system developed yourself! I've got my own mechanical system I'm happy with for right now, so neural networks would be a fun side-exploration.

Can you give some stats about your system? Things like win/loss rate, average win percentage, average loss percentage, expectancy, max drawdown (backtested or live), etc.

If you're skittish of the public forum, a PM to me would be fine.

Fatowl, its been a long hard road and tbh im still trying to understand the system fully and I'm yet to really start reaping the rewards. However I feel I'm getting there.

The system generates trading signals on the hour, every hour. So for this reason its only going to really work on a micro account (i.e 0.01 lots per trade). At any given point in time I can have over 350 trades open. Suffice to say only an automated system can manage that many trades at once. I had to ask my broker to customize my account slightly to allow micro-lots, unlimited size and max 500 open trades.

I have several strategies which work to varying degrees against different currency pairs. I cherry pick the best based on the stats and ensure that I use the optimum hold time for each strategy/currency-pair combination.

Below is the Year-to-date System Stats (incorporating 4 strategies across 9 currency pairs):

MEAN PER TRADE 0.58%
STD DEV 1.34%
T-STAT 35.8
WIN RATE 65.90%
TRADE COUNT 6771
MAX DRAWDOWN -66.71%
TOTAL CUMULATIVE RETURNS 3939.13%

This is back-tested performance, although I have traded some of this on a live account over the last 2 months. The real power of the system is derived from trading every single signal and not trying to fiddle about with it by closing trades early which has hurt me in the past. I know the optimal hold times so I need to stick with it because over time the analysis has statistically proved out the best hold times to have.

The system actually builds up a complex web of hedges (whether on the same currency pair or on correlated pairs) and this has caused me problems because by not trading every signal in the past I've had trades disabled that would have hedged against a set of trades that were on and eventually went against me.

Across my basket of currencies, a 1% move averages to approx 120 pips, so the system return over the last year would have been approx

3939 * 120 = 472680 pips

Ive been trying to do this on a £6k account but really I'd say I should have at least 12k account trading 0.01 lots per trade. Actual peak to trough max draw down is actually approx 100% (12000 pips) although looking at the cumulative returns chart for the system (attached) I think most would agree its pretty impressive!

 

[Intel_Bvc]

.... except it doesn't reflect an outlier. Trading without stops may improve the trade performance but combined with an outlier is what may kill an account - eventually.

Agreed.

I can only speak of my personal experience. I'm not advising anyone to trade without stop-losses. I'm just saying that with my system and from my experience they hurt me more than they help me. And I'm advising people to put themselves in a position where they can quantify the effects of using them (or not) so that they can make an informed decision.

However, I've actually statistically proved out that all my GBP/JPY trades need to have a stop-loss, so they do.

 

[random12345]

Fatowl, its been a long hard road and tbh im still trying to understand the system fully and I'm yet to really start reaping the rewards. However I feel I'm getting there.

What's with that d/d %? That's some exposure no if you intend to trade the same relative size live? Did you use an arbitrary amount for initial capital or something?

 

[Intel_Bvc]

What's with that d/d %? That's some exposure no if you intend to trade the same relative size live? Did you use an arbitrary amount for initial capital or something?

These calculations are done using "cumulative returns" (Cumulative Return Definition | Investopedia), so the value is independent of starting capital etc.

If you consider that 1% cumulative returns across my portfolio of currencies = approx 120 pips, then that max drawdown is 66.71 * 120 = 8005 pips. Remember the system itself generated over 470,000 pips. So this drawdown is actually amazingly good...

The smallest size I can trade is 0.01 lots which is approx £0.06 per pip. So Max Drawdown would have been £480 (8005 * £0.06).

I reckon the system should really be traded with an account size of 12k so actually in that context the drawdown would be 4% of Initial Capital.

I actually started trading with £6k in which case max drawdown as a percentage would be 8%.

 

[random12345]

These calculations are done using "cumulative returns" (Cumulative Return Definition | Investopedia), so the value is independent of starting capital etc.

If you consider that 1% cumulative returns across my portfolio of currencies = approx 120 pips, then that max drawdown is 66.71 * 120 = 8005 pips. Remember the system itself generated over 470,000 pips. So this drawdown is actually amazingly good...

The smallest size I can trade is 0.01 lots which is approx £0.06 per pip. So Max Drawdown would have been £480 (8005 * £0.06).

I reckon the system should really be traded with an account size of 12k so actually in that context the drawdown would be 4% of Initial Capital.

I actually started trading with £6k in which case max drawdown as a percentage would be 8%.

Interesting - what commission is this using? Your equity curve is very flat for a small level retail trader, normally such a thing is only possible with a 0.1 basis point commission rate (where there is no error in the code anyway!) - was live performance within expected norms? And have you budgeted for limit exits only along with added slippage for the ~40% losers or have you dealt with the variable spread in some other fashion?

How many parameters for entry do you have and what was the optimisation sample or are the entry and exit criteria not numerical?

 

[Intel_Bvc]

Interesting - what commission is this using? Your equity curve is very flat for a small level retail trader, normally such a thing is only possible with a 0.1 basis point commission rate (where there is no error in the code anyway!) - was live performance within expected norms? And have you budgeted for limit exits only along with added slippage for the ~40% losers or have you dealt with the variable spread in some other fashion?

How many parameters for entry do you have and what was the optimisation sample or are the entry and exit criteria not numerical?

Avg spread across all the trades is approx 4.4 pips so to consider trading costs for the last year then its simply 4.4 * 6771 = 29792.4 (or 248.27%).

Basically trading costs account for 6.3% (248.37/3939*100) of the total cumulative returns.

I do not use stop losses except for GBP/JPY trades so they remain open until the designated close time. Through analysis I know the optimum time each trade must stay open dependent on strategy/currency-pair combinations.

Its a neural network system... it take a whole lot of inputs, train the nets and then gives me a prediction for the week. I created my own strategies based on the price predictions. Over time I saw consistent behaviour from price when I saw certain predictions from the neural nets. So I've automated the system searching for these patterns and I place a trade according to how price would usually move prior to seeing a specific prediction from my Neural nets.

This is a 100% original concept.

For 5 weeks straight I made approx 8333 pips every week... which after trading costs and the fact that I kept foolishly closing trades before their optimum hold times is about spot on what the stats suggests I should make.

Then prior to the last NFP data release I made an error and thought that I needed to de-risk and closed a few trades and also disabled some strategies altogether... But because my system essentially builds up a complex web of hedges I essentially had positions open that weren't hedged as they would normally be and took a bit of a hit.

But my accounts still very much intact and I've learned my lesson that I cannot play with the trades and must simply trade all signals, all the time, because end of the day there were 12 NFP data release in the last year and the stats have shown my system would have made money regardless.

 

[random12345]

Avg spread across all the trades is approx 4.4 pips so to consider trading costs for the last year then its simply 4.4 * 6771 = 29792.4 (or 248.27%).

Basically trading costs account for 6.3% (248.37/3939*100) of the total cumulative returns.

I do not use stop losses except for GBP/JPY trades so they remain open until the designated close time. Through analysis I know the optimum time each trade must stay open dependent on strategy/currency-pair combinations.

Its a neural network system... it take a whole lot of inputs, train the nets and then gives me a prediction for the week. I created my own strategies based on the price predictions. Over time I saw consistent behaviour from price when I saw certain predictions from the neural nets. So I've automated the system searching for these patterns and I place a trade according to how price would usually move prior to seeing a specific prediction from my Neural nets.

This is a 100% original concept.

For 5 weeks straight I made approx 8333 pips every week... which after trading costs and the fact that I kept foolishly closing trades before their optimum hold times is about spot on what the stats suggests I should make.

Then prior to the last NFP data release I made an error and thought that I needed to de-risk and closed a few trades and also disabled some strategies altogether... But because my system essentially builds up a complex web of hedges I essentially had positions open that weren't hedged as they would normally be and took a bit of a hit.

But my accounts still very much intact and I've learned my lesson that I cannot play with the trades and must simply trade all signals, all the time, because end of the day there were 12 NFP data release in the last year and the stats have shown my system would have made money regardless.

Well you are heavily outperforming me if that's the case and I make a very good living from trading.

Seems like your future is paved with gold if you continue to replicate this in live.

 

[Intel_Bvc]

Well you are heavily outperforming me if that's the case and I make a very good living from trading.

Seems like your future is paved with gold if you continue to replicate this in live.

Lets hope so mate.

Like i said before I'm still not there yet, I'm still making mistakes because there is no precedent. No-one can really advise me on the system because i built it and I don't even I truly understand how the nets work and how they can "see" the unseen, underlying forces at work that are driving price.

I just have to learn through experience and hope i don't lose too much money learning the lessons before I'm able to become consistently profitable.

My friend who taught me how to do the statistical analysis is currently going through the spreadsheet I have with every single one of the trades/signals from last year because he cannot believe how out of this world the stats are.

Will be sure to keep you updated with how it goes, take care.