100% Win Math Grid Ea

My results,

Increment 10 in 2008-2009


Increment 20 in 2008-2009


Increment 40 in 2008-2009


Increment 10 in 2009-2010


Increment 20 in 2009-2010


Increment 40 in 2009-2010

From now on, I will post results only if implementing the concept I mentioned earlier improve on at least one of these results.

Would you mind backtesting in 2008-2009 ?
also, What happen if you start with 2500$, 0.01lot to begin with, as I did ? I will try to perform a backtest with 100.000$ and 10lots like you as well.

++

Dear all, Here are the results of my work:

2008-2009


and 2009-2010


so far, I have a fixed-sized grid of 40pips increment, with a 80pips deviation in between sell-lvl and buy-lvl, I use my own basic lot-calculation (nothing fancy, just something that make sense from a math point of view, but I did not work on optimizing lot-calculation yet).

Thus the only things that change are that I start a new grid if it agrees with the concepts I mentioned earlier, and I use the partial-grid-closing as we discussed earlier with Wael117, and the one I illustrated few post ago.

Now, I re-prog'd my code so as to be able to have an adaptative grid (based on whatever I will come up with). I will now work on this.

PS: I found a nice death trade in mid-2010 that blows my account. This will give me the opportunity to work on some death-trade mitigation "for-real" before starting the adaptative grid. Good, goz I have none from 2007.01->2010.01 .... of course my grid size is 40pips ... but I backtest with what I would use in live-trading.

PPS: Wael, any news ?

PPPS: Max lot-size open is about 1.3lot I believe ... so this is reasonable.

XmPh

this is the golden idea, but regretfully ignored (damn, how *(*&()*^&^???)

1. double lots are totally UNNECESSARY, martingale only kills. in fact, you can use smaller lots for higher levels, damn, ppl just don't get it.

2. distance among different levels must be variable or desigatable, instead of dumb predefined.

Ozzie Taz's golden idea was just deserted like this。。。。。。

I was thinking the exact same thing when I tested this ea... most profitable on 1 level and no need to times 10 the opposite side.

testing the latest version on odl gives one buy and if the price drops to sell it's opening 10 lots in the opposite side... very odd logic

let the user controll the lot sizing, distance between each level, and how much to increase the lots by if price reverses

Dear Charvo and cyberevil

... referring to point "1: Double lots are totally Unnecessary" ... "ppl just don't get it" ...... do you mean that ppl don't understand that a grid strategy implies creating a hedge, that is overloading one side of the grid to have a breakeven level before reaching the TP level.

..... now I will even add to this: Mathematically, doubling lots might not even be sufficient to provide a hedge at the TP level for a 3level grid .... so before making strong statements like this ... sit back, relax, take a paper, do the math, realize that taking into account for spread makes it even harder, consider some random time delays in your mathematically correct lot calculation, realize (if you can make it up to there) that it's more complex than you thought, eventually give up on the math, come back here and say sorry ! .... I know it's hard to realize this coz it needs effort but hey, this is not noddy's world (http://www.noddy.com/index_us.php), there's money at stake.

.. and you know what, this is why I decided to continuously draw the BE level on my grid .... because you can realize graphically that the math are correct .... you do need to hedge sometimes more than you would think so to end up with a profitable sequence. (that is more than doubling) ......

well of course, my comments assume that you have a real lot-calculation strategy that takes into account that depending on the case it might be less risky to overload a little more the last level compared to the first one that may oscillate, and that you took this in you stats .... with the spread that add to this ..... and let's say one order delayed by 3 seconds for whatever which end up opening a position with a defined slippage of 4pips .... and .... well..... ok I'm overloading your brain now, I'll stop here.

..... referring to point "2: distance among different levels must be variable" .... well my clever friend, on what basis? how do you make sure that the math are preserved within an automatic variable grid (that is assuming you understand that a grid is more than 1 level to provide real grid-hedge on which you can build a real lot-strategy, and that you need to overload at least one of the level) ?

Now referring to "Ozzie Taz's golden idea was just deserted like this" ..... to have the MANUAL input for determining the grid dynamically while the EA is executing ... simply too simple ... you want it, be a nice guy, open the code, add the few required lines and you'll get it.

Considering the "most profitable on 1 level", and referring to "martingale only kills" .... honestly taking these two comments, and just thinking that you two agree ... I'm laughing my ass off ! a 1 level grid IS A PURE STUPID MARTINGALE. (these two comments put together had me laugh so much that it was the reason I decided replying to your post .. actually that and the fact that you throw strong false statement with only little respect for ppl who worked on that thread probably a little more than you did)

What every one suggest but no one actually provides evidence for is "how do you automatically have a dynamic grid increment that make sense from a MM math point of view (that is the point of the grid, holding the math together right)" ? That is not part of Ozzie's comments (although these were suggesting the proper approach, it does not provide the how, but the what if).

on a more positive note, the concepts I've thrown previously, with the few screenshots are meant to start providing some ideas for a dynamic grid. Personally, I have coded a dynamic grid EA and you know what .... defining the proper strategy for "lot calculation" coupled with "dynamic level increment" coupled to "death trade management" ... well it's not easy and that's where I am right now.

Now, sorry for the contributors of the thread that will think I shouldn't have answer such posts but I used this opportunity to clearly state some strong concepts that must not be forgotten and on which we need to further work to reach our goals from my own understanding of the problem.

XmPh.

Well here we go again. I'm glad you had a good laugh at the 1 level idea because I have been having nothing but profit thus far. Yes it is marigale but the travel distance of the price to be in profit is much less and safe from ranging market where if it wen up 3 lvls and down 3 and up 3 you would be done. with 1 level you know it'll go up or down sooner or later but its not eating up ur $$$ at 3 times the rate and/or get out even if it doesn't after 2 or 3 attempts and start again. I have other ideas to make it much more safer and profitable which i'm working on with someone...

thouh i do agree with your ideas too... a true grid is multi-level. I have played with them and they're great but not everyone has the balls to see the dd it can create...

I should have said for me and my logic it's profitable.

oh and since we're being aggressive. Here is a thought for you. No need to say sorry to the contributors because they've moved on to bigger and better things. Who wants to make all this profit over a year just to see it gone with one trade? What I sense from you is that you want to promote your own personal agenda by controlling the topic to your Ideas and shooting others down? IDEAS... where would this ea be without them?

take a moment think about it...

My friend, the agenda has been set by FF himself when he posted the following two posts:

http://www.forexfactory.com/showpost...&postcount=864
and most importantly here:
Nov 14, 2009 9:54am - http://www.forexfactory.com/showpost...&postcount=837

specifically when he said: and while posting his results demonstrating a full backtested year 2009-2010 with InitialDeposit=10000$, TotalProfit=17645.43, Benefit = x1.76
compared to mine: 2009-2010 InitialDeposit=2500$, Benefit = x1.82 ... and Benefit of x2.13 in 2008-2009.

Indeed, I'm following my own agenda which is close to only considering ideas that are sound mathematically (not saying the rest is worth nothing, but to me these do not apply to a grid strategy). Specifically, I used the comments/concepts of plasmaplants and hannover which pay attention to the math - I'm also reviewing Viper's code to get a better feel of his concepts.

If you want to use a one level strategy ....results will be similar to the well known "London Breakout-like" ones ... so far, it may be OK but no guarantee that a traditional martingale will not kill you (even at 1 level) .... which is why I moved to Grid, and why I stick to the math ... and so far I managed to have a full 2 yrs of backtest without death trade ... yes I do have some understanding of 1Level, breakout-like strategies:

http://www.forexfactory.com/showpost...&postcount=644 (that is also a good post when referring to lot-calculation and "perceptual multiplier" versus "mathematically sound multiplier")

For that specific strategy, I coded an EA and backtested it ... yes it worked well in 2009, well in 2010 after April ... but March was a killer .... the whole point being ... 1 level can be a killer, so why improving a 1level system knowing that there is far less mathematically sound improvement for a 1level system than for a 3level one (3 providing the required odds that a 2level doesn't have).

You will see that I dropped the "aggressiveness" in this post as I do agree with you it was unnecessary ... I have a quick starter when I read words like "dumb" which were not yours as a matter of fact.

As I said earlier, I will not share any code, but I will share the concepts I use in my EA providing these concepts improved/helped passing full-year backtest with this strategy and/or mitigate death-trades. This is my way to say thank you to the contributors of the thread, and at any moment I may stop posting if the current readers think this is a waste of thread space. I'm not looking for recognition, nor coding help, nor waiting for anyone to code my ideas. I do all that and provide everyone with the results I found sound.

Hope this clarifies a little my interventions.

PS: If you have food for thoughts on entries for your 1level strategy I'm all ears ;-) (although I'm working on the mitigation part right now as the entry strategy I have coded so far allows for a 2yrs backtest without failure).

mitigation part is very important for this type of ea because there is no avoiding the death trade. Death trade is defined by a trade that eats all of your margin. Now if we try to mitigate and allow say 50% margin to get used. We have just given away half of our accouts away not knowing if that trade was infact a death trade or could it have recovered? That's the tricky part and anything over 50% of margin is too much I think.

I think my thought pattern changed to 1 level mainly due to this reason because I couldn't think of anything else. But I know for a fact that this ea does not trade well using 1 level. I'll try to explain.

typical level 1 setup:

------------------ buy price

------------------ current price

------------------ sell price


My test:
I use bigger tp (ie 150 pips if its in a long term trend or $50 if it keeps failing) and adx 40 to calculate increments. as of now what its doing is going up to buy lets say and it hits buy 0.1 lots it's placing 1 lot of sell ( i don't know why) and sometimes it fails to put any pending sells and these two months its been luckly profitable. i don't think this will work out in long term but if it can make 100% i can take it out and leave it running for year or two on vps. if not learn and live...

Does anyone know how to set the leverage of 100:1 to 500:1 while using the MT4 FXT loader "metatrader.4.build.226.fxt.multiloader"

Although my mitigation does not work on 100:1 .... with a ratio of 5 (500:1) it would make it a little easier for the math (no magic at some point when you start with 2500$)

I will still try to make it work on a 100:1 but I don't know how feasible that is with such a small account size ..... otherwise, I'll start with a larger amount at the beginning (x5).

Thanks for any help.

For others:
You will have to edit experts/include/FXTHeader.mqh and change:
FileWriteInteger(handle, i_leverage, LONG_VALUE);
to:
FileWriteInteger(handle, AccountLeverage(), LONG_VALUE);

OR manually modify the i_leverage VALUE to whatever leverage value you want to backtest with.

Once this is done, recompile the script to regenerate the fxt file and re-generate it.

Done.

This is a wonderful tool to test EAs with different leverage settings.

X.

Dear all,

mitigation is on its way.
So far, I was stuck on year 2010, no matter the leverage (100:1 was stuck in a "not enough moeny" and 500:1 go right to the margin-call/stop-out account blowing).

Right now, I have implemented a buggy/sketchy mitigation version of what I'm working on ..... and results are great.

Check-out 2010 - with leverage 500:1 (causing the account blow-up before on january 28 ... the death day)


Same with Leverage 100:1 (causing a loss of 1500$ in January 28 previously) .... almost equivalent as previously ... (I will check this as this appears a little strange, although the magnitude of open position is 0.366lots Max compared to 1.14 before .. which hurts a lot at 100:1)


..... this is a grid strategy mitigated for death trade reaching ..... a 10% Draw Down I mean come on ...... not bad ;-)

and same code applied to 2009-2010 - with leverage 100:1 (as I did not have the time to recalculate the time frame with a different leverage so far).:


So indeed, one can see there is mitigation compared to what I posted before has .... sometimes it looses money .... well part of it is because it is still buggy and I did not do all the math yet ;-)

For the courageous ones, short explanation of what I mean by mitigation:
- Still, listen to what every one said previously: Pivot, Resistance, Support .... whatch the grid in backtest and you'll see for yourself .... yes the grid has to be dynamic - so I've implemented a ... sub-dynamic grid .... Mmmm sub says enough, think about it.
- there are two reasons for account blow-ups (mostly): (1) not enough free-margin ... and (2) equity=0 (happen a lot before with L=500:1) resulting in a margin call.

for (1), there are ways to free some cash from the grid during a sequence, without compromising the hedge.
for (2), I monitor (equity-margin)/margin >0.5 with respect to worst case scenario the "W" whiplash (to think 7 step ahead) and validate that I have enough $$ in bank ... otherwise start making some cash, or start .... a basket-management to breakeven.

To sum up: I only use Pivot/Resistance/Support .. the rest is only math, no indicator at all and I believe I will finish my EA like that ... except for the trailing stop of the 1-2-3 as discussed with Wael which I believe can trail any trend-following indi but I'm not interested in this as of now.

BTW .... any one here ?
Viper, are you still interested in this ?


ForexFlash, I would actually be interested in knowing whether or not your new grid strategy is performing well or not ? What is the result of the "voucher's" analysis ?

PS: what's next .... decreasing drawdown by using two grid on hedged pairs ..... to me, once my mitigation is in place, this will be one more step to the great temple of grid strategies.

Cheers
XmPh

Did anyone consider using a grid, split in between two currencies:
Like the bull side on EUR/USD and the Bear side on USD/JPY, managing everything from EUR/JPY ?

Viper ?

Hi XmPh,

Your mitigation work is really looking good. Many months ago I did a lot of testing/mods trying to work out something that would work in most circumstances, but didn't get anywhere.

Great to see your progress - thanks for your contributions!

Cheers, Andrew.

P.S. I like the idea of grids on multiple pairs.

hey
i just read your post on blessing thread
where is the last version of your grid ea so i can test it ?
thanks

How doe this system handle danglers?

IMO, as such, in present form, by opening (probably ill-spaced in fast moving markets) more and orders culminating into DEATH Trade (or a margin call);
(Especially with significant speed, with small account and large trade sizes, in a perfectly expanding triangle price pattern)
Not sure if you have any other specific DEFINITION of dangler than so called extreme price movements, to and fro

Hope that helps. !


P.S. If a newbie (solely relying on ur no. of posts), i advise considering some price based systems floating around FF for testing/even real trading (which most itch to; no offence, its an obvious n commonly seen phenomena)

Dear All,
I've been out, concentrating on another project for the last few weeks.

Sad to realize no one keeps on posting in this thread as there is great potential.

I hope to come back here in about a week or two. I will keep demonstrating how a dynamic grid can profitably support price action in the sense I explained in my previous posts.

X.

Hi there,

Sitting about one evening I thought I'd do a google search just to see if anybody else had something going along the lines I have.....having spent several months almost full time on a Gridding EA concept rather similar to the one here, and as a Trading Director of my own FX based company having had 2 other Grid EA's coded up - you could say I am rather interested

Firstly a big thanks for the time folks have generously donated, sharing their ideas here.

I'm not sure just how similar our EA's are, but the buy above and sell below a gap, rehedging at each failed swing is certainly very much along the lines of my EA. Mine's entirely mechanical, ie no indicators (though its on the list, I thought I best get the mechanical version 100% first). The backtests here look suspiciously similar to mine. Very much easier said than done the system itself despite how simple it seems - the math involved in just a basic version is quite something....start adding in logic that reacts to numbers of failed swings, spread, carry and expansion rates for TP and SL...and boy it gets complex exponentially. I am not entirely sure....but I dont think the EA system in this thread does continue to follow the trend (which was an objective of mine) as mine does but alot of the things said are equally applicable to my EA.

The one thing that absolutely stands out is the risk of the martingale - and thats what I would be most interested in finding out more on how to mitigate. One thought of mine was to have a max used margin drawdown limit...and set it just above the max drawdown you want to see before it closes the trades for a loss....rather basic I confess. Admittedly the tact I took was that the a martingale was inevitable and the goal is to make as much hay while the sun shines, running fixed lots (playing on high probability at the outset, then increased balance for increased probability of surving a bad marty), before it takes out acct....withdraw profit fast and regualr. Also, to spread risk across multiple pairs, allocating a portion of acct (in theory) to each pair...but playing on the fact that a bad marty will very rarely occur on more than one pair at a time. In this manner you can trade very small lots per pair on a bigger balance, having lots more margin to float through the occasional bad marty when a pair occasionally runs into trouble (trouble I exploit when it does occur mind you - refer below*). Basic stuff...but it all factors and in short my efforts have been more about ROI than risk mitigation (to date).

Towards risk mitigation my efforts have mainly been directed at finding settings that work consistently over time....and Ive a set (plus or minus) that work on pretty much on any pair for around 100% annually fixed lots. I found that trading starting a trade zone at 10:00 and going through to 18:00 works well....aiming to close flat for friday/weekend. My test survive 5 years...with the occasional big drawdown (but no losses).......demos have thus far made money and not blown up....its good but I am certainly on the lookout for ways to make it better.

Another idea I have - but have not coded yet - was to have additional grid layers in the gap in such a way that only one grid set is ever rehedged. Not sure if that makes sense or not...and the how to achieve will take up more space than is reasonable here and now. In this manner you can back off on initial trade size and spread the risk out via multi grids over essentially the same price action. In this manner, the martingale will only ever be on a smaller sized grid while still having a similar degree of exposure to capture the ROI.

If I was to donate one valueable peice of logic * (aside from my idea above) to the person behind the EA in this thread that has worked very well for me (assuming they havent done it yet and its something similar to mine).....is that after it rehegdes itself the net difference between the long and the short presents quite an opportunity to hold the weighted trade set open a bit longer for a much bigger gain with just an extra pip or two (or more)....so after each marty your backtest/equity will take a nice jump up each time. I trigger this single functionaility via a number of rules....certainly its been the biggest contributor to added ROI (it more than doubled it in fact). Very simple, too...but then the best ideas often are.

Anyhow, looking forward to being a part of this thread....I think I've timed it about right ...LOL

Cheers, Adam

ps....now I am going to have to study this thread to see if I can nail down the system and it difference to mine....this multi level sounnds intriguing (mine is just two levels) - anybody got a manual or a guide round here !!??

In the current (recent) EA - there is undervalued tool - set profit for a basket! And adding some pips in "our" way will triple (vary) or more the result (I've exploiting that few months ago but not in that ea).

edit: "to hold the weighted trade set open a bit longer for a much bigger gain with just an extra pip or two (or more)....so after each marty your backtest/equity will take a nice jump up each time. I trigger this single functionaility via a number of rules....certainly its been the biggest contributor to added ROI (it more than doubled it in fact). Very simple, too...but then the best ideas often are."
I am curious about these rules...?