Anyone interested in sharing WORKING strategies
Anyone interested in sharing WORKING strategies
Hi Guys,
I have a few WORKING strategies for futures and forex.
I'd be interested in seeing other strategies which are also working.
My suggest if someone is interested, then we first exchange the system stats and graphs, and then if that is acceptable to both parties, then we exchange code.
Let me know your thoughts if anyone is interested.
My strategies are usually > 700% Return on account, so i would only be interested in other which also fit this criteria.
Kind Regards
Riz
I have a few WORKING strategies for futures and forex.
I'd be interested in seeing other strategies which are also working.
My suggest if someone is interested, then we first exchange the system stats and graphs, and then if that is acceptable to both parties, then we exchange code.
Let me know your thoughts if anyone is interested.
My strategies are usually > 700% Return on account, so i would only be interested in other which also fit this criteria.
Kind Regards
Riz
Re: Anyone interested in sharing WORKING strategies
Hi rkhan,
As you are talking about strategies that work, I would be interested to know what broker you use to execute your forex strategies?
I assume of course that your strategies are executed on accounts, real or demo, at a broker connected with MultiCharts! ...
Kind Regards,
As you are talking about strategies that work, I would be interested to know what broker you use to execute your forex strategies?
I assume of course that your strategies are executed on accounts, real or demo, at a broker connected with MultiCharts! ...
Kind Regards,
Re: Anyone interested in sharing WORKING strategies
And what is the Max Strategy Drawdown (%) please ?My strategies are usually > 700% Return on account, so i would only be interested in other which also fit this criteria.
Re: Anyone interested in sharing WORKING strategies
Hi Strat Man,
We can exchange strategy results and you can see it all if you like.
But my draw downs are usually between 3K -15K over a 10 year back test on a single contract.
So some systems they are only 3K, other systems they are even up to 15K
We can exchange strategy results and you can see it all if you like.
But my draw downs are usually between 3K -15K over a 10 year back test on a single contract.
So some systems they are only 3K, other systems they are even up to 15K
Re: Anyone interested in sharing WORKING strategies
Thank you, but I just wanted to know the Max Strategy Drawdown (%) for the +700% strategy, I mean expressed in percentage and not in value... because I'm not able to appreciate a value when I don't know the initial capital... That's why I asked for the % of the Max DD which is on the Strategy Performance Report of MultiCharts.
Re: Anyone interested in sharing WORKING strategies
Dear Sir,
I can get you that when I am back on my computer
But the value i have given you has more meaning than what you are asking
Max DD as a % can be whatever i want. I can start with 100k and make it small or start with 10K and make it big
But you see return on account is based on maxDD it's the final profit as a % of maxDD, read it's description
So 700% means that I have made 7x my maxDD
I can get you that when I am back on my computer
But the value i have given you has more meaning than what you are asking
Max DD as a % can be whatever i want. I can start with 100k and make it small or start with 10K and make it big
But you see return on account is based on maxDD it's the final profit as a % of maxDD, read it's description
So 700% means that I have made 7x my maxDD
Re: Anyone interested in sharing WORKING strategies
Yes Riz, you're absolutely right, but to briefly analyze a strategy, I personally look at the annual rate of return %, max DD % and VAMI (and the number of trades ...). So I'm a little disappoint with Return on Account that I do not have a habit to analyze.
You can see some of my strategies here.
You can see some of my strategies here.
Re: Anyone interested in sharing WORKING strategies
StratMan,Yes Riz, you're absolutely right, but to briefly analyze a strategy, I personally look at the annual rate of return %, max DD % and VAMI (and the number of trades ...). So I'm a little disappoint with Return on Account that I do not have a habit to analyze.
You can see some of my strategies here.
Very impressive equity curves, wish I had something like it.
Re: Anyone interested in sharing WORKING strategies
Great job Stratman. I didn't expect less from you
Completely out of curiosity, do you have them together in a portfolio test? I like pretty pictures
Do you have more history in those back-tests?
What FX data do you use?
In the past, I spent 3 years on a portfolio of FX systems which were extremely impressive. Sadly, due to manipulation in the data by my brokers (which was proven), I threw them all out and chalked it up to a learning experience as I moved back to more reliable markets.
My data sources were TS fx (pretty good), MB Trading (manipulated the most), FXCM (best), Dukascopy (horrible), Histdata (worst), and TrueFX (2nd best). This was a couple years ago.... but I had put all data side by side and studied it extensively. VERY different from source to source and the catcher that changed things for me: Each month the realtime data that I collected from each broker source would not match its historical data of that same month. This was verified by 4 engineers from different teams.
Completely out of curiosity, do you have them together in a portfolio test? I like pretty pictures
Do you have more history in those back-tests?
What FX data do you use?
In the past, I spent 3 years on a portfolio of FX systems which were extremely impressive. Sadly, due to manipulation in the data by my brokers (which was proven), I threw them all out and chalked it up to a learning experience as I moved back to more reliable markets.
My data sources were TS fx (pretty good), MB Trading (manipulated the most), FXCM (best), Dukascopy (horrible), Histdata (worst), and TrueFX (2nd best). This was a couple years ago.... but I had put all data side by side and studied it extensively. VERY different from source to source and the catcher that changed things for me: Each month the realtime data that I collected from each broker source would not match its historical data of that same month. This was verified by 4 engineers from different teams.
Re: Anyone interested in sharing WORKING strategies
I am unable to use the portfolio MultiCharts with forex!Completely out of curiosity, do you have them together in a portfolio test? I like pretty pictures
(Or rather, the portfolio of MultiCharts is not usable for forex because of different currencies) and I'm too lazy to create my portfolio on excel
I had, I made, I limited to a full year, but I can get all periods thanks to MultiCharts!Do you have more history in those back-tests?
These reports are based on lmax data (real account) but I certainly can reach similar performance with other FX data... For example, the backtest with lmax demo account data is much better !What FX data do you use?
I am personally convinced that there is no single forex broker, which does not manipulates data ... It is an unregulated market ... where the brokers must pay commissions to all intermediariesIn the past, I spent 3 years on a portfolio of FX systems which were extremely impressive. Sadly, due to manipulation in the data by my brokers (which was proven), I threw them all out and chalked it up to a learning experience as I moved back to more reliable markets.
I know... and made the same kind of comparison... I checked for example FXCM data - iQfeed/fxcm - mcfx/fxcm etc...My data sources were TS fx (pretty good), MB Trading (manipulated the most), FXCM (best), Dukascopy (horrible), Histdata (worst), and TrueFX (2nd best). This was a couple years ago.... but I had put all data side by side and studied it extensively. VERY different from source to source and the catcher that changed things for me: Each month the realtime data that I collected from each broker source would not match its historical data of that same month. This was verified by 4 engineers from different teams.
and my conclusions were similar to yours ...
My latest tests with lmax allowed me to see that real-time data are not identical on two different computers, and are not identical to what I get in reload (historical data) ...
It is very close to the random lotto draw!
Well, I could write several volumes on the subject, but I've already wasted enough time and money to audit all these brokers ...
Apologies Riz for polluting your thread with this big comment off topic!
I won't do it again.
Re: Anyone interested in sharing WORKING strategies
Stratman, thank you for elaborating. You're doing a great job.
::Back on topic:: (sorry for that)
I like the direction that Rkhan is taking here. I know a few successful developers who all approach the market in a unique way. If a few decided to share their strategies and they were all built correctly, things could be very diverse and promising. I've only succeeded in doing this with one other person and I didn't care for their strategy so I've never used it. I handed them a Porsche and they handed me a used Honda... I'm still open to the idea. I think the pros far outweigh the cons.
I wish there was a tool out there that helped developers collaborate... I simply don't see it working well though. Trust is a rare thing in this industry.
::Back on topic:: (sorry for that)
I like the direction that Rkhan is taking here. I know a few successful developers who all approach the market in a unique way. If a few decided to share their strategies and they were all built correctly, things could be very diverse and promising. I've only succeeded in doing this with one other person and I didn't care for their strategy so I've never used it. I handed them a Porsche and they handed me a used Honda... I'm still open to the idea. I think the pros far outweigh the cons.
I wish there was a tool out there that helped developers collaborate... I simply don't see it working well though. Trust is a rare thing in this industry.
Re: Anyone interested in sharing WORKING strategies
I check out the discussion forum periodically to see if I can give back as I have received a lot of help over the past year here. A thread like this is a reason I'd likely visit everyday. Perhaps we can share some details to learn from versus the entire strategy. There are so many elements to system design and it is truly a work in progress, you are always learning. Especially the first time you go from backtesting to live and realize systems don't just print money when turned on. But that does not mean the strategy is bad it just means you likely have more work to do. Wanted to post something here as a thumbs up for this thread.
Re: Anyone interested in sharing WORKING strategies
I have a system developed for ZB to intraday scalp. I target roughly 6 ticks and average loss is just north of 4 ticks. But the way ZB has traded lately the system is not performing as well so I moved on to ES. I never liked ES but for an automated scalping product it has plenty of movement whether trading long and or short and is performing well. Ideally I want to have a portfolio of products across trading styles from scalping to swing, etc. I launched the ZB program with a remote client. It was quite the initial learning curve with variations in data from different providers (I use IQ, they were using TT) and the simple reality that PC clocks drift. So now we sync clocks daily and both use IQ.
I'm hesitant to approach the FX markets, at least spot. But perhaps futures. I'm rambling here but wanted to post a few thoughts.
I'm hesitant to approach the FX markets, at least spot. But perhaps futures. I'm rambling here but wanted to post a few thoughts.
Re: Anyone interested in sharing WORKING strategies
I've started a few scalping projects on ZB and ES as well. They fit the bill perfectly. I keep running into MC issues so I keep giving up until the next issue is fixed. No success so far...
All my successful strategies *that are trading live* are swing/trending systems.
All my successful strategies *that are trading live* are swing/trending systems.
Re: Anyone interested in sharing WORKING strategies
My two cents :
My very first step when I develop a strategy is to verify the data I will use for this strategy.
So, I check that I receive exactly the same data on 2 different computers based on different locations, and that the timestamp is correct with acceptable time of transmission.
(And for example, with LMAX, these two parameters are horrible that disable the possibility to use such a broker. Might as well play the lottery! )
My second step (when the first one is ok) is to check to have identical result between 3 identical charts with the same strategy.
One chart in auto trading mode, the second in real time data and the third in backtest mode.
And I modify my code until these 3 modes give identical results, accepting some differences only when justifiable by the delay of the signal transmission to the broker.
These are the first two stages of my long development process ...
My very first step when I develop a strategy is to verify the data I will use for this strategy.
So, I check that I receive exactly the same data on 2 different computers based on different locations, and that the timestamp is correct with acceptable time of transmission.
(And for example, with LMAX, these two parameters are horrible that disable the possibility to use such a broker. Might as well play the lottery! )
My second step (when the first one is ok) is to check to have identical result between 3 identical charts with the same strategy.
One chart in auto trading mode, the second in real time data and the third in backtest mode.
And I modify my code until these 3 modes give identical results, accepting some differences only when justifiable by the delay of the signal transmission to the broker.
These are the first two stages of my long development process ...
Re: Anyone interested in sharing WORKING strategies
My client runs IQ for data and I as well. As long as we each sync clocks every morning our results on a scalping system are to the tick about 95% of the time. When using different data providers though it was far greater. More so on a scalping system I would say data variations is a bigger problem. I would imagine less so on a longer trade timeframe.
MATricks what kind of MC issues have prevented you from running live, a scalping system? I'm just curious.
MATricks what kind of MC issues have prevented you from running live, a scalping system? I'm just curious.
Re: Anyone interested in sharing WORKING strategies
One question I would love to hear feedback on regards running an automated system non-stop or turning on and off during varying market conditions. A good trader knows when to trade and when not to trade. My view is an automated system is no different. No strategy works during all market conditions. The key is if you opt to turn it off is avoiding bias as much as possible and having some conditionals either within the signal (not easy to do) or a "mental conditional" where it's one of those "I've seen this market before" and "I don't want to trade it." I'm just curious what others do with an automated system. Do you (a) run it non-stop other than perhaps key data releases or (b) turn on and off based on what you are seeing at the time. I fully appreciate that an automated system should have zero bias and picking option A introduces bias, but at the same time I don't see how an automated system can navigate through all conditions. It's almost impossible to code everything in.
Re: Anyone interested in sharing WORKING strategies
Tony, if a good trader knows when to trade and when not to trade, so a good coder must code the if/then/else process that runs into the brain of the good trader
IMHO I prefer to code and let the system run H24 (when MultiCharts accept... )
For example, if I don't want to trade when the spread is to big, then my code will reduce the order size or disable this period of trade...
IMHO I prefer to code and let the system run H24 (when MultiCharts accept... )
For example, if I don't want to trade when the spread is to big, then my code will reduce the order size or disable this period of trade...
Re: Anyone interested in sharing WORKING strategies
Easier said than done but agreed in that ideally you can code all of what is in the trader's brain. I guess the reality is you will not code everything in until AFTER the fact. With my systems I study every day's trades in replay mode tick by tick after market close to see if anything stands out. If I find a change to make I then go back and backtest it to confirm that the benefits do not outweigh the costs (giving up winning trades). I am no where near complete with this task. Perhaps I never will be.
Re: Anyone interested in sharing WORKING strategies
Tony, my issue with automated scalping systems with MC is mostly order control. We have none. Also, analyzing the order book... we're weak there as well. All in time.. MC is unbelievably quick with their upgrades/updates.
There's always the debate of discretionary vs automated and the power of each. A program can do a good job at what a great trader does. It will never be better.. BUT a trader is stuck with 1-2 strategies. Once automated, you can have 100 strategies running Power of automated is in diversification and proving the edge is there.
There's always the debate of discretionary vs automated and the power of each. A program can do a good job at what a great trader does. It will never be better.. BUT a trader is stuck with 1-2 strategies. Once automated, you can have 100 strategies running Power of automated is in diversification and proving the edge is there.
- t-rader
- Posts: 139
- Joined: 02 Feb 2011
- Location: Australia
- Has thanked: 11 times
- Been thanked: 27 times
Re: Anyone interested in sharing WORKING strategies
I like this idea, but unfortunately back test results can be highly manipulated to show incredible returns and awesome equity curves! I remember my first system I created I had thought I created the holy grail! Then when I went live it just wasn't working as expected. I refer to this system now as the hindsight system.
I do something similar to what Stratman has said and once I've gone live and have a few trades I will load up the back test and see if these line up with the live trade results.
I think results also need to be compared year to year and over a number of years, not just applying your strategy across a number of years as this manipulates the results such as drawdown.
Also which is a better measure 'Return on Account' or 'Return on Initial Capital'? Personally I don't think 'Return on Account' is a good measure for what can be achieved realistically in the real world without really high risk? or am I incorrect in this thinking?
I do something similar to what Stratman has said and once I've gone live and have a few trades I will load up the back test and see if these line up with the live trade results.
I think results also need to be compared year to year and over a number of years, not just applying your strategy across a number of years as this manipulates the results such as drawdown.
Also which is a better measure 'Return on Account' or 'Return on Initial Capital'? Personally I don't think 'Return on Account' is a good measure for what can be achieved realistically in the real world without really high risk? or am I incorrect in this thinking?
Re: Anyone interested in sharing WORKING strategies
I don't think much for either. Return over max drawdown is what I look for.
And you nailed it. Most "developers" are similar to most "traders". But the flaws are easily spotted...
And you nailed it. Most "developers" are similar to most "traders". But the flaws are easily spotted...
Re: Anyone interested in sharing WORKING strategies
And that is why it seems impossible to ignore Bid & Ask for a forex strategy on small time frames such as 1, 5 or 15 minutes. Otherwise, the backtest will always be different from real time trading and if the simulator does not comply with the rocket, we are unlikely to reach the moon!I do something similar to what Stratman has said and once I've gone live and have a few trades I will load up the back test and see if these line up with the live trade results.
Re: Anyone interested in sharing WORKING strategies
I agree with you !I don't think much for either. Return over max drawdown is what I look for.
And you nailed it. Most "developers" are similar to most "traders". But the flaws are easily spotted...
I can show Annual rate of return of 500 % or more, but with drawdown I personally cannot accept
I always code my strategies in order to manage the risk per trade and set the parameters to see on the Strategy Performance Report a max drawdown between 3% and 5%...
- t-rader
- Posts: 139
- Joined: 02 Feb 2011
- Location: Australia
- Has thanked: 11 times
- Been thanked: 27 times
Re: Anyone interested in sharing WORKING strategies
Agree, an awesome return can be un-tradable if the drawdown is too big!I agree with you !
I can show Annual rate of return of 500 % or more, but with drawdown I personally cannot accept
I always code my strategies in order to manage the risk per trade and set the parameters to see on the Strategy Performance Report a max drawdown between 3% and 5%...
I'm curious to know when others discard a system as un-tradable, say for example you backtest your system over 5 years and all years look good except for one year where lets say that year the return is very low or the drawdown is very high...
Re: Anyone interested in sharing WORKING strategies
It's case by case, but I look at what was different during the irregular drawdown and assess whether I can filter these areas out. If not, I discard or work further on any/all strategies that do not show great performance through the entire duration of the max back-test I can put it through.I'm curious to know when others discard a system as un-tradable, say for example you backtest your system over 5 years and all years look good except for one year where lets say that year the return is very low or the drawdown is very high...
Re: Anyone interested in sharing WORKING strategies
Just a remark for periods of backtest :
The fractal nature of the markets shows that a test is in fact conducted on a number of bars and not only over a period. (you can read "The Misbehavior of Markets: A Fractal View of Financial Turbulence" by Benoit Mandelbrot)
As a reminder, test a strategy on 14400 bars (just as example) correspond:
on a Daily chart, 55 years of history!
on H1 chart, 600 days of history, that is more than 2 years!
on a 1 minute resolution,... ... just 10 days!
But in every case, the strategy has always worked on 14400 occurrences of OHLC objects.
So, I personally prefer to consider a number of bars rather than the test period.
When I do a backtest on 1 year, based on a 1 minute resolution graph, it is actually a backtest performed over approximately 370,000 bars of information ...
Calculate yourself how many years of history it represents daily bar
And please tell me if you prefer a 1 year backtest of a strategy applied to a 1 minute bar chart or 10 years backtest of a strategy applied to a daily chart
The fractal nature of the markets shows that a test is in fact conducted on a number of bars and not only over a period. (you can read "The Misbehavior of Markets: A Fractal View of Financial Turbulence" by Benoit Mandelbrot)
As a reminder, test a strategy on 14400 bars (just as example) correspond:
on a Daily chart, 55 years of history!
on H1 chart, 600 days of history, that is more than 2 years!
on a 1 minute resolution,... ... just 10 days!
But in every case, the strategy has always worked on 14400 occurrences of OHLC objects.
So, I personally prefer to consider a number of bars rather than the test period.
When I do a backtest on 1 year, based on a 1 minute resolution graph, it is actually a backtest performed over approximately 370,000 bars of information ...
Calculate yourself how many years of history it represents daily bar
And please tell me if you prefer a 1 year backtest of a strategy applied to a 1 minute bar chart or 10 years backtest of a strategy applied to a daily chart
Re: Anyone interested in sharing WORKING strategies
I agree with your view Stratman in regards to how many bars are tested versus simply the period. I guess it also depends on the type of strategy and the product being traded. If some strategies are more psychology related (for example Bollinger Bands) which focus on extremes in trader psychology (greed and fear, right?) they would likely tend to work far longer. Whereas some sort of arbitrage or pair trade will likely not work as long.
I'm "newer" to automated systems so the one question that haunts me for lack of better words, is if you develop a good strategy, how long will it truly work?
Sidetracking a bit here. I love trading off multiple data series using shorter time frames to indicate longer term direction and or momentum. Though I love MC for countless reasons, the limitations in backtesting across multiple data series is a concern. It makes development far more difficult. But I like multiple data series such that I am not going to forego them. If you are not aware, in backtesting even with one tick resolution, data1 will calculate on every tick, whereas data2 and higher (dataN) will calculate only on OHLC. At least how I understand it and what I see in backtesting.
Glad I'm not the only one spending Memorial Day weekend studying. I was working late last night when I saw MAT's post, now it's Sunday and Stratman is posting. I often spend Friday and Saturday nights studying and programming. I absolutely love this work. Though admittedly I have an addictive personality so walking away is extremely difficult at times. End of very long rant. I appreciate learning from others here. Hope I can give back as well.
I'm "newer" to automated systems so the one question that haunts me for lack of better words, is if you develop a good strategy, how long will it truly work?
Sidetracking a bit here. I love trading off multiple data series using shorter time frames to indicate longer term direction and or momentum. Though I love MC for countless reasons, the limitations in backtesting across multiple data series is a concern. It makes development far more difficult. But I like multiple data series such that I am not going to forego them. If you are not aware, in backtesting even with one tick resolution, data1 will calculate on every tick, whereas data2 and higher (dataN) will calculate only on OHLC. At least how I understand it and what I see in backtesting.
Glad I'm not the only one spending Memorial Day weekend studying. I was working late last night when I saw MAT's post, now it's Sunday and Stratman is posting. I often spend Friday and Saturday nights studying and programming. I absolutely love this work. Though admittedly I have an addictive personality so walking away is extremely difficult at times. End of very long rant. I appreciate learning from others here. Hope I can give back as well.
Re: Anyone interested in sharing WORKING strategies
10 year back-test on 1 minute barsplease tell me if you prefer a 1 year backtest of a strategy applied to a 1 minute bar chart or 10 years backtest of a strategy applied to a daily chart
Re: Anyone interested in sharing WORKING strategies
Interesting article and I completely agree with the conclusion. No ONE trading system will ever create what we all desire in a trading system. My version of a holy grail is a portfolio of 50+ uncorrelated systems across several markets and time frames.
-
- Posts: 6
- Joined: 24 May 2014
- Has thanked: 4 times
- Been thanked: 3 times
Re: Anyone interested in sharing WORKING strategies
I can hand you a Ferrari, but what would I do with a Porsche?I handed them a Porsche and they handed me a used Honda... I'm still open to the idea. I think the pros far outweigh the cons.
-
- Posts: 6
- Joined: 24 May 2014
- Has thanked: 4 times
- Been thanked: 3 times