and I will be very glad if xmph can/will share his versions
(but as I remember he did not will)...?
(but as I remember he did not will)...?
Gday Muskateer,
Thanks for your reply - although I'm not quite sure what you mean unfortunately.
I've given this thread a read, but it so long its hard to figure out where we are at here.
q - have you used the EA in question? does it work on any pair? whats the max dradown for what returns year to year? how many years does it go back without blowing up @ what ROI?
I'm pretty sure there's things we could both learn from each other but the issue is I don't know quite what I'm looking at here. So I put it to whom ever is reading this thread is that Ive done several full time months on a project like the EA in this thread and I'm offering to share some of my insights but I kinda need someone to give me a bit of a heads up here if I am to be able to share much of what I have learned.
I've done 6 years of full time hard slog in FX FWIW, IBC pools, managed FX (myself and via other traders) use of multiple EA's and have had 4 EA systems coded myself/thousands of hrs of backtesting experience, run a VPS full time with multiple client accounts...I run a FX company with an in house specialist trader. I am a serious full time participant in FX. Yes I'm trying to sell myself here so I can get some folks seriously interested in conversing here/with me or else I may have to pass on further particpation in this thread. Time for me is just too limited and valuable to not get a good return on its investment.
On a related subject - the coder I use I am able to put to work fulltime on my EA - but he is that over committed that I find myself waiting too long for simple things to be completed - so towards that end I put this out there:
I am looking for an experienced+ethical coder who's good with Grid EA's to either JV with or to pay/hr/job to further develope my EA. I have some significant improvements in mind................maybe even whoever coded the EA in this thread, surely you'd be interested in the many hundreds of hours of throught and developement that's gone into mine and perhaps some gems of ideas that have not crossed your mind - I am looking for the same and more.
Take good care.
Cheers, Adam
Thanks for your reply - although I'm not quite sure what you mean unfortunately.
I've given this thread a read, but it so long its hard to figure out where we are at here.
q - have you used the EA in question? does it work on any pair? whats the max dradown for what returns year to year? how many years does it go back without blowing up @ what ROI?
I'm pretty sure there's things we could both learn from each other but the issue is I don't know quite what I'm looking at here. So I put it to whom ever is reading this thread is that Ive done several full time months on a project like the EA in this thread and I'm offering to share some of my insights but I kinda need someone to give me a bit of a heads up here if I am to be able to share much of what I have learned.
I've done 6 years of full time hard slog in FX FWIW, IBC pools, managed FX (myself and via other traders) use of multiple EA's and have had 4 EA systems coded myself/thousands of hrs of backtesting experience, run a VPS full time with multiple client accounts...I run a FX company with an in house specialist trader. I am a serious full time participant in FX. Yes I'm trying to sell myself here so I can get some folks seriously interested in conversing here/with me or else I may have to pass on further particpation in this thread. Time for me is just too limited and valuable to not get a good return on its investment.
On a related subject - the coder I use I am able to put to work fulltime on my EA - but he is that over committed that I find myself waiting too long for simple things to be completed - so towards that end I put this out there:
I am looking for an experienced+ethical coder who's good with Grid EA's to either JV with or to pay/hr/job to further develope my EA. I have some significant improvements in mind................maybe even whoever coded the EA in this thread, surely you'd be interested in the many hundreds of hours of throught and developement that's gone into mine and perhaps some gems of ideas that have not crossed your mind - I am looking for the same and more.
Take good care.
Cheers, Adam
Hi there,
Its nice to come across a kindred spirit/s here...not too many folks in everyday life that share this line of thought. Apologies for my delayed repsonse (my excuse is that I am in deepest darkest EA testing and development bat cave...LOL, you know what I mean)
I think FOREXFlash summed it up it pretty well when he (I could mistakenly assume) said "Mathematicaly, this can not be beaten.....there will always be a point in time/price where you need to get out @ loss....mathematics on its own can not give you an edge, ...you need to know exactly how, when and where...
this is where statistics comes handy..."
Try as we might, there's no escaping the enivabitility of a bad martingale by whatever variant it arrives by. I believe I read that your system is non indicator based and relies on the price action as a guide for its direction? (as mine does so entirely). Versus the Blessing system which picks a direction and price averages/fades move and looks to get out of dodge quick, then goes about the usual practice of gridding in levels until the price moves against the system again. Simply said they've arived at an outcome that produce X # of 50% drawdowns (or whatever the equity hard loss is set at) for a higher amount of return over a given period - or as xmph said I think x amount of consecutive losses concept.
When I was looking at the equity graph of Blessing form on the backtester it was a thing of (decieving) beauty - in that it does NOT show the true volatility via that graph - only when you look at actual report. It goes fantastic...until that bad enevitable martingale event.
I've found the same with mine - so my main focus was to "make hay while the sun shines" , ie looking at way to make as much as possible while using statistics of its timed daily application to at least lay the bed of previous history.
Its now in the FWD test stage, debugging etc. Satisfying that my system I refer to as the "Pendulum" system (due to its nature of swinging back and fourth/rehedging until price breaks out) did actually produce a higher ROI for lower relative+max drawdown - taking nothing away from the Blessing system, there's some nice features they have and its a well sorted EA by the looks but certainly different in that its directionally based/relies on picking a direction. I guess you could say, their's is more a trend picking system, using grid methodology to execute the trade where as mine is a pure grid algorithm overlaid onto price - the core forumula is actually on a xcel spreadsheet.
One day I was thinking to introduce a trend input into mine but have purposefully kept strictly to the mechanical system - until just very recently (as in the last day!) testing a version with a volatility indicator. My system is by design headed in the right direction when the market gets going, so naturally it follows when the BB's open up we can hold or add X pips to the TP of the basket (of longs and shorts in my case, biased one way or the other). That mod was worth about 5-10% additional ROI for same DD....and it provides the EA with a system of trend following but entirely as a by product of the price action and how it affect the volatility inputs.
My lot progression are complex, involve a relationship with the TP/SL and where Ive found alot of extra profit. Assuming a crossover of 40 pips, above which is long for 20 pips TP and short below 20 pips for TP, lots progressions in my system can be adjusted by:
1) the TP can be fixed, no matter how often it cross it requires same TP and lot progression will fall into line.
2) the TP for same net profit as 1) can be increased, requiring smaller successive lot progression (but decreased probability of hitting)
3) the lot size for fixed TP can be held for longer when;
a) carry/roll is applied (cover carry plus some)
b) x number of failed swings (biggest winner)
c) volatility input screams busy, hold ur position and when adding new grid, add some extra TP to them.
(and my volatility input will tell the EA to keep hunting for new grid as well, if the market is still busy at tradezone end time)
A not extreme mix of the above, set to start around the UK open and stop hunting (looking for more grids) around US open does yield some good ROI for extended periods and the required set file doesnt differnt too much from pair to pair. Its often flat from day to day on some if not all 6 pairs on my demo and is designed to close flat for the weekend - getting it back to flat is important because of the market gap risk on open. In fact the whole operation thinking is always get back to flat, ever time a TP is hit - without exception thats what it does (or dies trying, its a martingaler ok..hehe)
Your comments on the hedge/exposure reducing do sound complex - a coders nightmare. I'm not certain if its the way to go because if you reduce the weighting of a hedge it has to be made up for in some other way. And if you retain the same weighting, you'd have to close some of it a a loss then? Eh I am open to feedback, too....this is absolute cracker with these systems and as i said a bove, my efforts have thius far been channeled more at the ROI side of the equation - for if we improve the ROI then that has by itself improved the ROI
D ratio/reduced the relative dradown. I do so think its the way to go though...playing to its strengths - we know its weakness and stats assist in determing how to avoid those market periods best avoided.
XMPH - that Blessing site was a superb referral, quite a team and developed EA, thanks for that. I can see ur a deep thinker and a good stats man, humbly I confess I dont quite get you on the comments you made about looking at it in a static manner (even more so because I think it applies especially to my non direction/random but timed entry model) - but I am interested in knowing! If you would expand on that I'd appreciate that and any other feedback you might have about anything I said that made some sense to you.
For another day I would like to share more with my concept of gridding inside the gap between the long and the short (for those who comprehende) - using my algorithim for when any trades requirement maintainence outside the grid gap. In this manner, by my logic - the system can run the same total amount of exposure but splayed in such a way that only 1/3rd of it could ever martingale up (based on a STD 20 pip TP and 40 pip gap). This functionality at present is only conceptual, and involves a much higher degree of complexity than the present "single layer" verison I have now - but its the next stage for me development wise. This is where I believe the biggest risk reduction can be achieved.....in the meantime I do subscribe to dividing a set risk up into multiple currencies so that the odds are better that a martingale event will only absorb a relatively small % of acct capital AND if a pair starts to martingale up do not add extra instances of EA on other pairs until its gone flat (remember my system often has pairs gridded to flat everyday, then it commence again next day/trade zone).
Hoping maybe this rant has given fellow gridders some food for thought, certainly this thread has been an interesting one for me and I look forward to reading more.
Cheers, Adam
ps...when I post enough I guess I recieve posting privilages? nothing commercial, just screenshot/back tests etc.
pps - FOREXFlash - just how far did you get with you system - care to post a latest backtest run of it?
Its nice to come across a kindred spirit/s here...not too many folks in everyday life that share this line of thought. Apologies for my delayed repsonse (my excuse is that I am in deepest darkest EA testing and development bat cave...LOL, you know what I mean)
I think FOREXFlash summed it up it pretty well when he (I could mistakenly assume) said "Mathematicaly, this can not be beaten.....there will always be a point in time/price where you need to get out @ loss....mathematics on its own can not give you an edge, ...you need to know exactly how, when and where...
this is where statistics comes handy..."
Try as we might, there's no escaping the enivabitility of a bad martingale by whatever variant it arrives by. I believe I read that your system is non indicator based and relies on the price action as a guide for its direction? (as mine does so entirely). Versus the Blessing system which picks a direction and price averages/fades move and looks to get out of dodge quick, then goes about the usual practice of gridding in levels until the price moves against the system again. Simply said they've arived at an outcome that produce X # of 50% drawdowns (or whatever the equity hard loss is set at) for a higher amount of return over a given period - or as xmph said I think x amount of consecutive losses concept.
When I was looking at the equity graph of Blessing form on the backtester it was a thing of (decieving) beauty - in that it does NOT show the true volatility via that graph - only when you look at actual report. It goes fantastic...until that bad enevitable martingale event.
I've found the same with mine - so my main focus was to "make hay while the sun shines" , ie looking at way to make as much as possible while using statistics of its timed daily application to at least lay the bed of previous history.
Its now in the FWD test stage, debugging etc. Satisfying that my system I refer to as the "Pendulum" system (due to its nature of swinging back and fourth/rehedging until price breaks out) did actually produce a higher ROI for lower relative+max drawdown - taking nothing away from the Blessing system, there's some nice features they have and its a well sorted EA by the looks but certainly different in that its directionally based/relies on picking a direction. I guess you could say, their's is more a trend picking system, using grid methodology to execute the trade where as mine is a pure grid algorithm overlaid onto price - the core forumula is actually on a xcel spreadsheet.
One day I was thinking to introduce a trend input into mine but have purposefully kept strictly to the mechanical system - until just very recently (as in the last day!) testing a version with a volatility indicator. My system is by design headed in the right direction when the market gets going, so naturally it follows when the BB's open up we can hold or add X pips to the TP of the basket (of longs and shorts in my case, biased one way or the other). That mod was worth about 5-10% additional ROI for same DD....and it provides the EA with a system of trend following but entirely as a by product of the price action and how it affect the volatility inputs.
My lot progression are complex, involve a relationship with the TP/SL and where Ive found alot of extra profit. Assuming a crossover of 40 pips, above which is long for 20 pips TP and short below 20 pips for TP, lots progressions in my system can be adjusted by:
1) the TP can be fixed, no matter how often it cross it requires same TP and lot progression will fall into line.
2) the TP for same net profit as 1) can be increased, requiring smaller successive lot progression (but decreased probability of hitting)
3) the lot size for fixed TP can be held for longer when;
a) carry/roll is applied (cover carry plus some)
b) x number of failed swings (biggest winner)
c) volatility input screams busy, hold ur position and when adding new grid, add some extra TP to them.
(and my volatility input will tell the EA to keep hunting for new grid as well, if the market is still busy at tradezone end time)
A not extreme mix of the above, set to start around the UK open and stop hunting (looking for more grids) around US open does yield some good ROI for extended periods and the required set file doesnt differnt too much from pair to pair. Its often flat from day to day on some if not all 6 pairs on my demo and is designed to close flat for the weekend - getting it back to flat is important because of the market gap risk on open. In fact the whole operation thinking is always get back to flat, ever time a TP is hit - without exception thats what it does (or dies trying, its a martingaler ok..hehe)
Your comments on the hedge/exposure reducing do sound complex - a coders nightmare. I'm not certain if its the way to go because if you reduce the weighting of a hedge it has to be made up for in some other way. And if you retain the same weighting, you'd have to close some of it a a loss then? Eh I am open to feedback, too....this is absolute cracker with these systems and as i said a bove, my efforts have thius far been channeled more at the ROI side of the equation - for if we improve the ROI then that has by itself improved the ROI
D ratio/reduced the relative dradown. I do so think its the way to go though...playing to its strengths - we know its weakness and stats assist in determing how to avoid those market periods best avoided.XMPH - that Blessing site was a superb referral, quite a team and developed EA, thanks for that. I can see ur a deep thinker and a good stats man, humbly I confess I dont quite get you on the comments you made about looking at it in a static manner (even more so because I think it applies especially to my non direction/random but timed entry model) - but I am interested in knowing! If you would expand on that I'd appreciate that and any other feedback you might have about anything I said that made some sense to you.
For another day I would like to share more with my concept of gridding inside the gap between the long and the short (for those who comprehende) - using my algorithim for when any trades requirement maintainence outside the grid gap. In this manner, by my logic - the system can run the same total amount of exposure but splayed in such a way that only 1/3rd of it could ever martingale up (based on a STD 20 pip TP and 40 pip gap). This functionality at present is only conceptual, and involves a much higher degree of complexity than the present "single layer" verison I have now - but its the next stage for me development wise. This is where I believe the biggest risk reduction can be achieved.....in the meantime I do subscribe to dividing a set risk up into multiple currencies so that the odds are better that a martingale event will only absorb a relatively small % of acct capital AND if a pair starts to martingale up do not add extra instances of EA on other pairs until its gone flat (remember my system often has pairs gridded to flat everyday, then it commence again next day/trade zone).
Hoping maybe this rant has given fellow gridders some food for thought, certainly this thread has been an interesting one for me and I look forward to reading more.
Cheers, Adam
ps...when I post enough I guess I recieve posting privilages? nothing commercial, just screenshot/back tests etc.
pps - FOREXFlash - just how far did you get with you system - care to post a latest backtest run of it?
- #1,237
- Oct 29, 2010 2:00am Oct 29, 2010 2:00am
hello everyone, i knows its been a long time i haven't posted here, but was busy on other projects..
the developments on my end is that i've been working on other systems, with previous experience in programming i learnt MQL and developed quiet well in the past couple of months...
Now something that came certain to my understanding and many people have advised before that holy grail doesn't exist.. and that is true, looking back if there was a holy grail no one would have had time to even check on forums or learn form each other and so on. and i would like to thank all the people here, for their contribution.
My myself i am developing a system which is quiet good with the expectations i have in mind..
But here comes the question that i wanted to ask many people like FF and xmph and all other developers who are trying to develop martingale systems..
What are your expectations??
to be honest i think about this as an investment because the phrase' 34481... % in 45 days do not exist..
To be honest all the commercial EAs work, I will not deny that they do not.. but its just a matter of expectations.. after you see a commercial EA releases a statement that it makes 300% in 45 days and after you hit your first SL after 1 week, you lose hope and you see that you lost 20% of your account.. you start going for the next one.
Now here i am trying to be realistic and in the same time not too greedy..
This is what i am developing
A - martingale system that will run on multiple pairs, there is no input, all inputs such as grid, tp and "SL" are all variable and they are all automated by calculation based on the market movement and outcomes of many calculations and yes it has a stop loss, denying the fact that there is no stop loss just doesn't complete the equation of an investment, it is soooo unebelivable that people are looking to invest without even putting into consideration that yes you might lose, its a fact, reality and people have to put this as a very true possibility. creating a system with no stop loss is just like tossing a coin. now martingale can save up the SL with maximizing your exposure and increase your DD, but there has to be a limit to this. and this is when your SL comes into play, we have to be again realistic, martingale systems are high risk low return systems, but still SL has to be there. no SL is just a pure gamble, i would have more fun in vegas than sitting in front of my computer trying develop a system for the past couple of months.
B - My return that i expect per year per pair is 5-10% with a very low DD looking not to exceed 10% DD on the account per pair.
C - I have created a manager EA which supervises the SUB martingale EAs.
THe reason to this i think that martingale are very profitable, though very dangerous when left unattended, on the other hand, if you gonna stay up to attend it 24/5 then just do it manually.. so i created this Manager EA which over looks every pair trading and calcualtes the exposure of the capital and the DD and based on some calculations it has the ability to disable other pairs or enable them, or even hedge all positions or stop them out and limit the DD to a certain extent, in which this would be the only situation that the owner of the EA needs to intervene to restart the Master EA therefore bringing all SUB EAs into play once again.
Now if i am running a min of 4-5 pairs, and i gain 40% to 50% annual thats almost 10x - 20x of what i will gain with what i would do if i deposit this money into the bank, or put my money in any non-leveraged investment scheme.
I just want to summarize, that maybe martingale is not the problem, maybe our expectations that we would like to run a $10k account, turn them into a $200k by the end of the year, and doing more than 1000 lots on a 10k account is just quiet unrealistic and doesn't make sense.
Once again, if someone proved the above calculation to be true, banks would either hire or kill to get his services
in the end of the day just my question, how realistic do you get? or do you have certain targets before creating the EA..
many people started with getting 5-10-15% on their strategy annually, and then they kept adding mods in order to make it reach 200 - 300% but then complain about a blow out.. but i just think if it reaches 5-10% annual per pair.. thats it, thats more than what you would ever get with any other investment scheme. so if its not broken and its reliable DO NOT FIX IT!!..
i hope this is going somewhere, and maybe it can brush up new ideas.. for the guys down here.
now how my idea started is to create a simple EA, with entries and stop losses and to go from there and to see how does it perform, and once i see this is stable enough i then start implementing the martingale to it in order to minimize or take advantage of the retracements that did not fit into my logic.
But according to my analysis to many martingale EAs that has been created, the martingale was the basis and then the logic came into play. which according to my very little knowledge should be the opposite. once again i am trying to think outside the box, and maybe thinking the opposite of what everyone thinks.. 90% could be wrong.. and 10% might open up a new idea for people to test and develop something better.
Sorry if this was a waste of time, i just tried to be constructive.
please comment constructively.. i'll be more than happy to ready your comments.
the developments on my end is that i've been working on other systems, with previous experience in programming i learnt MQL and developed quiet well in the past couple of months...
Now something that came certain to my understanding and many people have advised before that holy grail doesn't exist.. and that is true, looking back if there was a holy grail no one would have had time to even check on forums or learn form each other and so on. and i would like to thank all the people here, for their contribution.
My myself i am developing a system which is quiet good with the expectations i have in mind..
But here comes the question that i wanted to ask many people like FF and xmph and all other developers who are trying to develop martingale systems..
What are your expectations??
to be honest i think about this as an investment because the phrase' 34481... % in 45 days do not exist..
To be honest all the commercial EAs work, I will not deny that they do not.. but its just a matter of expectations.. after you see a commercial EA releases a statement that it makes 300% in 45 days and after you hit your first SL after 1 week, you lose hope and you see that you lost 20% of your account.. you start going for the next one.
Now here i am trying to be realistic and in the same time not too greedy..
This is what i am developing
A - martingale system that will run on multiple pairs, there is no input, all inputs such as grid, tp and "SL" are all variable and they are all automated by calculation based on the market movement and outcomes of many calculations and yes it has a stop loss, denying the fact that there is no stop loss just doesn't complete the equation of an investment, it is soooo unebelivable that people are looking to invest without even putting into consideration that yes you might lose, its a fact, reality and people have to put this as a very true possibility. creating a system with no stop loss is just like tossing a coin. now martingale can save up the SL with maximizing your exposure and increase your DD, but there has to be a limit to this. and this is when your SL comes into play, we have to be again realistic, martingale systems are high risk low return systems, but still SL has to be there. no SL is just a pure gamble, i would have more fun in vegas than sitting in front of my computer trying develop a system for the past couple of months.
B - My return that i expect per year per pair is 5-10% with a very low DD looking not to exceed 10% DD on the account per pair.
C - I have created a manager EA which supervises the SUB martingale EAs.
THe reason to this i think that martingale are very profitable, though very dangerous when left unattended, on the other hand, if you gonna stay up to attend it 24/5 then just do it manually.. so i created this Manager EA which over looks every pair trading and calcualtes the exposure of the capital and the DD and based on some calculations it has the ability to disable other pairs or enable them, or even hedge all positions or stop them out and limit the DD to a certain extent, in which this would be the only situation that the owner of the EA needs to intervene to restart the Master EA therefore bringing all SUB EAs into play once again.
Now if i am running a min of 4-5 pairs, and i gain 40% to 50% annual thats almost 10x - 20x of what i will gain with what i would do if i deposit this money into the bank, or put my money in any non-leveraged investment scheme.
I just want to summarize, that maybe martingale is not the problem, maybe our expectations that we would like to run a $10k account, turn them into a $200k by the end of the year, and doing more than 1000 lots on a 10k account is just quiet unrealistic and doesn't make sense.
Once again, if someone proved the above calculation to be true, banks would either hire or kill to get his services
in the end of the day just my question, how realistic do you get? or do you have certain targets before creating the EA..
many people started with getting 5-10-15% on their strategy annually, and then they kept adding mods in order to make it reach 200 - 300% but then complain about a blow out.. but i just think if it reaches 5-10% annual per pair.. thats it, thats more than what you would ever get with any other investment scheme. so if its not broken and its reliable DO NOT FIX IT!!..
i hope this is going somewhere, and maybe it can brush up new ideas.. for the guys down here.
now how my idea started is to create a simple EA, with entries and stop losses and to go from there and to see how does it perform, and once i see this is stable enough i then start implementing the martingale to it in order to minimize or take advantage of the retracements that did not fit into my logic.
But according to my analysis to many martingale EAs that has been created, the martingale was the basis and then the logic came into play. which according to my very little knowledge should be the opposite. once again i am trying to think outside the box, and maybe thinking the opposite of what everyone thinks.. 90% could be wrong.. and 10% might open up a new idea for people to test and develop something better.
Sorry if this was a waste of time, i just tried to be constructive.
please comment constructively.. i'll be more than happy to ready your comments.
- #1,239
- Edited Nov 1, 2010 1:27am Oct 31, 2010 10:10am | Edited Nov 1, 2010 1:27am
- | Joined Sep 2010 | Status: Trader | 27 Posts
Hi XMPH,
Your responses are always full of detail and I spend a good amount of time digesting them and "converting" them into the necessary paralells with my system.
Thinking about all of this I think a great deal of the conversation is about avoidance of and the dealing with, the death trade, via two ways (aside from the use only multi pairs/multi demensional approach which deserves a post in itself -for later);
1) firstly avoidance by way of using statistics to run it at the best times
2) developing a robust setup that fits into the price action regulary enough that it yield consistent returns commensurate with the risk taken - using stats again.
AND
3) Money management, ie using a small enough lot and allowing for statically rare bad case scenarios (also based stats) AND
4) Martingales based mitigation, such as taking a loss/or part loss and the reduction of exposure - whch is system based firstly, then stats tell us if its useful or not.
Probably I think 4) is where the tastiest fruit lie (for me here and now) because 1+2 are more about stats than anything else, backtesting, optimization and the like, unque to ur system and 3) is very simple common sense.
So to sum it up, it seems to be either be about avoiding the death marty trades, and the actual dealing with that death marty trade scenario when it occurs, right?
Ive racked my brains, for the life of me, when I overlay your comments onto my system I still cant figure your process of exposure reduction,how it applies to a marty/death trade and I think its because we are using same terms that means different things to us?
So lets use an example.
If my system is 10.0 lots long, and 4.0 short and I need 20 pip higher to hit a full take profit, then this makes me net 6.0 lots long?
Now, if I dump say 3 short lots and 3 long lots, leaving me 7 lots long and 1 lot short (net 6 lots long) then I still require 20 pips to get me my take profit.
I'd have to wait until the 10 lots long and 4 lots short were at break even before i could dump them as well, so a move X pips into my favour is required.
There doesnt seem anything gained here except margin reduction and given my system rarely hold a position longer than a day, interest is hardly a concern.
Now, if we dumped some, and the net affect was a change in the net long bias value then if the net long was higher, then a lower take profit in pips would be required BUT at the same time if the market were to reverse, then it becomes more adversely exposed in the other direction. Vice versa also applies.
So, xmph what am I missing here?
========================
Aside from avoiding the marty trades via stats, timed entry, good robust setup it seems taking the odd big hit/more regular smaller ones is the order the day to mitigate the marty/death trade.
Now, Marty based MM always have quite big hits/losers, compared the incremental gains. My simple idea (that I have not had coded in or tested yet) is to take a part/smaller loss in the series of any single martingale.
So in affect we're at the casino,we've bet $20, lost, we've bet $40 on surer thing, lost, bet $100 on even surer thing, lost, we bet $200, we loose again............NOW, with my system we still actually have the $100 bet running, if the market continues that way then we win $20 and if it doesnt go that way and returns then we are back the $200 risk bet again - which we can win on if it goes in the right direction (opposite direction to the $100 bet) OR we can lose again (opening up the still running $100 bet).
The issue with the above is that it the market could range sideways and keep hitting that $200 losing bet - need a rule to cut that at some point.
(also, with my system as the we "loose" successively - actually we are just rehedging, and the net short long bias becomes more extreme, thus just a small increase in the take profit value in pips yields several time the usual $20 return, so we get some extra mileage on the martingale - I would guess its equivilent to the Blessing holding onto its heavy position for a bit longer for a bigger payoff)
Again, I will have to use statistics to see if taking a smaller loss more regularly as a matter of course yeilds more consistently lower dradown for still consistent and worthy gains. I believe this same approach could be used as it applies to Blessing/direction approach and possibly your 100% win mather?
Keep in mind xmph I still dont know your 100% win math grid system, and think the reverse applies too, so I hope I'm making some sense
One thing is for sure, its hard to talk in absolutes when we have limited understanding.
===============================================
To the fellow who asked what are the objectives of a marty system?
From a system designers perspective I dont think there should be any ceiling on how much ROI is acceptable for the DD (drawdown - edit), the more ROI you can get for the same or less DD is about the best absolute single meaure all else aside.
I used to think 30% for 100% annualized with compounding was decent. However, with my system in testing the year to date figure is just under 30% for around 400-500% on the GBPUSD, 300-400% for the EURUSD and not much less on the USDJPY. Similar sets work on other pairs, its quite a universal system but other pairs have not had the required optmization/testing really. Ive a fellow doing some optimizations with some heavy duty hardware and says he's he seen 1000% YTD for 40% DD - yet to run sets myself so cant really comment further. Still some code bugs/gremlims as well which makes me doubly excited of its potential.
Once you have base line ROI for the drawdown - then as xmph has done just gear accounts to your risk profiles and employ money management as you see fit. You could gear the system right back, have very minimal dradown and high probability of long term success....or you can crank the thing into oblivion, aiming for 100%+ monthly, withdrawing profits every week.
We all fit in there somewhere ...hehe
I'd really love to touch on the use on multi pairs using master EA's boy thats interesting but will leave that for another post.
Cheers, Adam
Your responses are always full of detail and I spend a good amount of time digesting them and "converting" them into the necessary paralells with my system.
Thinking about all of this I think a great deal of the conversation is about avoidance of and the dealing with, the death trade, via two ways (aside from the use only multi pairs/multi demensional approach which deserves a post in itself -for later);
1) firstly avoidance by way of using statistics to run it at the best times
2) developing a robust setup that fits into the price action regulary enough that it yield consistent returns commensurate with the risk taken - using stats again.
AND
3) Money management, ie using a small enough lot and allowing for statically rare bad case scenarios (also based stats) AND
4) Martingales based mitigation, such as taking a loss/or part loss and the reduction of exposure - whch is system based firstly, then stats tell us if its useful or not.
Probably I think 4) is where the tastiest fruit lie (for me here and now) because 1+2 are more about stats than anything else, backtesting, optimization and the like, unque to ur system and 3) is very simple common sense.
So to sum it up, it seems to be either be about avoiding the death marty trades, and the actual dealing with that death marty trade scenario when it occurs, right?
Ive racked my brains, for the life of me, when I overlay your comments onto my system I still cant figure your process of exposure reduction,how it applies to a marty/death trade and I think its because we are using same terms that means different things to us?
So lets use an example.
If my system is 10.0 lots long, and 4.0 short and I need 20 pip higher to hit a full take profit, then this makes me net 6.0 lots long?
Now, if I dump say 3 short lots and 3 long lots, leaving me 7 lots long and 1 lot short (net 6 lots long) then I still require 20 pips to get me my take profit.
I'd have to wait until the 10 lots long and 4 lots short were at break even before i could dump them as well, so a move X pips into my favour is required.
There doesnt seem anything gained here except margin reduction and given my system rarely hold a position longer than a day, interest is hardly a concern.
Now, if we dumped some, and the net affect was a change in the net long bias value then if the net long was higher, then a lower take profit in pips would be required BUT at the same time if the market were to reverse, then it becomes more adversely exposed in the other direction. Vice versa also applies.
So, xmph what am I missing here?
========================
Aside from avoiding the marty trades via stats, timed entry, good robust setup it seems taking the odd big hit/more regular smaller ones is the order the day to mitigate the marty/death trade.
Now, Marty based MM always have quite big hits/losers, compared the incremental gains. My simple idea (that I have not had coded in or tested yet) is to take a part/smaller loss in the series of any single martingale.
So in affect we're at the casino,we've bet $20, lost, we've bet $40 on surer thing, lost, bet $100 on even surer thing, lost, we bet $200, we loose again............NOW, with my system we still actually have the $100 bet running, if the market continues that way then we win $20 and if it doesnt go that way and returns then we are back the $200 risk bet again - which we can win on if it goes in the right direction (opposite direction to the $100 bet) OR we can lose again (opening up the still running $100 bet).
The issue with the above is that it the market could range sideways and keep hitting that $200 losing bet - need a rule to cut that at some point.
(also, with my system as the we "loose" successively - actually we are just rehedging, and the net short long bias becomes more extreme, thus just a small increase in the take profit value in pips yields several time the usual $20 return, so we get some extra mileage on the martingale - I would guess its equivilent to the Blessing holding onto its heavy position for a bit longer for a bigger payoff)
Again, I will have to use statistics to see if taking a smaller loss more regularly as a matter of course yeilds more consistently lower dradown for still consistent and worthy gains. I believe this same approach could be used as it applies to Blessing/direction approach and possibly your 100% win mather?
Keep in mind xmph I still dont know your 100% win math grid system, and think the reverse applies too, so I hope I'm making some sense
One thing is for sure, its hard to talk in absolutes when we have limited understanding.
===============================================
To the fellow who asked what are the objectives of a marty system?
From a system designers perspective I dont think there should be any ceiling on how much ROI is acceptable for the DD (drawdown - edit), the more ROI you can get for the same or less DD is about the best absolute single meaure all else aside.
I used to think 30% for 100% annualized with compounding was decent. However, with my system in testing the year to date figure is just under 30% for around 400-500% on the GBPUSD, 300-400% for the EURUSD and not much less on the USDJPY. Similar sets work on other pairs, its quite a universal system but other pairs have not had the required optmization/testing really. Ive a fellow doing some optimizations with some heavy duty hardware and says he's he seen 1000% YTD for 40% DD - yet to run sets myself so cant really comment further. Still some code bugs/gremlims as well which makes me doubly excited of its potential.
Once you have base line ROI for the drawdown - then as xmph has done just gear accounts to your risk profiles and employ money management as you see fit. You could gear the system right back, have very minimal dradown and high probability of long term success....or you can crank the thing into oblivion, aiming for 100%+ monthly, withdrawing profits every week.
We all fit in there somewhere ...hehe
I'd really love to touch on the use on multi pairs using master EA's boy thats interesting but will leave that for another post.
Cheers, Adam
Hi XMPH,
.......
The issue with the above is that it the market could range sideways and keep hitting that $200 losing bet - need a rule to cut that at some point.
Cheers, Adam
From what I can read, I think you understood what I meant but let me rephrase what you said here:
I'd have to wait until the 10 lots long and 4 lots short were at break even before i could dump them as well, so a move X pips into my favour is required.
(1) you know in advance how far you can go in terms of orders and lot size based on your grid and on your deposit / leverage / BrokerStopOut.
(2) you have the knowledge of Sup/Res which is not the case of the law of series so in that regards, it's a an edge you have on your statistics.
... (3) ... you may add a little lot here and there based on Sup/Res or any other hedge (I use the London open as a good hedge on break-out) to increase your hedge .... yes that may increase you risk as well - so these, if not in your favor, should be taken away asap during the exposure reduction process.
This lead to the following: determine the BE-level based on (1) and (2), not on the amount of lots you are willing to "kill".
I'm not sure I fully understand your "betting" strategy. Could you please provide me another explanation, I think I'm a bit slow today. Can you submit an actual screenshot of a visual backtest please. WE can do this by PM - send me a message and I will provide you my email, or post it here as I did previously when I explained my S/R strategy.
However, you have to consider the previous understanding with that of these posts http://www.forexfactory.com/showpost...postcount=1193
A quick backtest results would lead to the following: http://www.forexfactory.com/showpost...postcount=1197, that is 300% gain for a 30% DD. Not too bad.
In anything, any part of strategy or whatever you wanna call it any comments I make, I always have in the back of my head that no matter how good are stats and how much I love math, these have to make sense from a market point of view.
Thus reducing exposure while maintaining the hedge with purpose to reduce margin is the best way I found so far to "make room" to plan how to get out-of the unexpected death trade ..... that is bringing TP closer to price action.
X.