|
Joined May 2008
|
Status: do i look confused?
|616 Posts
Hey Ozzie.
I'm running 2 grids, starting the second one when it reaches near b2 or s2.
the most number of grids which is comfortable to run is 3. but i've stopped doing that because sometimes the starting price for the grids end up getting pretty close to each other which removes the benefit of having a multigrid to start with. with 2 grids it takes a lot longer before the starting prices start bunching up. it would take a serious spike move.
Hi Ogie, Welcome to the group. I am always happy to have additional brains with additional ideas. Since it sounds like you have already had some experience with the grid, I am interested in your experience and changes you requested that actually helped.
Would you please describe the changes you asked your programmer to make ? Would you be willing to post your code. I understand if you are not since you alone paid for them. Many people here have volunteered many hours of hard work programming, testing, and working through new ideas.
Either way, welcome aboard and hope you can help to make ViperGrid even better.
Viper
P.S. I am a programmer and could have made your changes for free. :-)
it is first time I am posting here and I hope and believe I can contribute to this thread. I am new to the forum, but certainly not a novice for Forex - have been doing it for a bout 5 years full time.
I have used mGRID (the original version, to which I have mad many modifications) for quite some time on a live account until yesterday, when I lost about $5k on GBPCHF. I lost it because the EA does not work properly - in time of high volatility (and very low volatility too) when spreads are much higher, the orders are all over...
In back tests, at the end of the testing period, every position is liquidated. Since the grid often runs drawdowns before it eventually closes in profit, the last trade set, even though in drawdown is closed for a loss. This is simply a poorly designed backtesting utility.
During backtests for EURUSD H1 no matter what settings I apply (1/2/3 level, 25 increment/autoincrement, trend filter on/off, volatility filter on/off, atr filter on/off, low water mitigation) I always get a big loss which cuts my account by more than a half.
This is a really good idea for those with a limit to the number of open and pending orders. On my demo account, I have a limit of 200 orders and I think many others have similar limits. This would GREATLY reduce the number of open orders. I may implement this.... Of course it does not help with margin limits...
It seems the other ideas are simply entry filters. I have not spent much time on the entry filters. I have been working more on the risk mitigation (if you start getting in trouble) and making it work better in general.
Here is how I have changed mGrid(please consider this as a suggestions to improve the ViperGrid)
1. Consolidate the pending orders - no need to have 3,4,5 or more pending orders on one level - it probably bothers the broker and more importantly - it is difficult to track and monitor the work of the EA. As soon as a pending order has to be placed on a level that already has a pending order on it - delete the existing pending order and place the total of the existing and the new lots.
By the way, I bothered to read through all 61 pages of this thread and I never, not once saw the word optimization. Is anyone doing any optimization or is everyone suicidal ( only if you use it live).
One question to whoever can answer it:
Is the ViperGrid EA time frame dependent, ie, does it matter the time frame of the chart it is placed on. I would think it is because of all filter added, they must be dependent on the time frame of the chart, but I would like someone who knows the EA really well to confirm it
Hi plasma, I have been playing a little with multi grids and if you start at b3/s3 it is impossible to have two grids both in consolidation. When the first grid crosses the pivot and heads to the opposite side, the second grid will close. At least you can only have one grid caught in the death trade.
Continued thanks for all the help and testing.
Later Viper
I'm running 2 grids, starting the second one when it reaches near b2 or s2.
the most number of grids which is comfortable to run is 3. but i've stopped doing that because sometimes the starting price for the grids end up getting pretty close to each other which removes the benefit of having a multigrid to start with. with 2 grids it takes a lot longer before the starting prices start bunching up. it would take a serious spike move.
This is a really good idea for those with a limit to the number of open and pending orders. On my demo account, I have a limit of 200 orders and I think many others have similar limits. This would GREATLY reduce the number of open orders. I may implement this.... Of course it does not help with margin limits...
It seems the other ideas are simply entry filters. I have not spent much time on the entry filters. I have been working more on the risk mitigation (if you start getting in trouble) and making it work better in general.
Viper
Ignored
consolidating orders may help for backtesting purposes as I believe it is based on 200 max orders (with Alpari UK).
However, this wouldn't help on accounts where a limit is set on the amounts of lots you can trade. Alpari's micro account has a 2 lot maximum (which happens to be 200 x 0.01lots!). Perhaps the demo has the same limit, but I don't think so.
I also have attached my results (and report in the zip file to show my settings are identical) as I have had widely different results between vgrid 2.4 and 2.5 with all settings equal - can anyone point out why?
In back tests, at the end of the testing period, every position is liquidated. Since the grid often runs drawdowns before it eventually closes in profit, the last trade set, even though in drawdown is closed for a loss. This is simply a poorly designed backtesting utility.
Viper
Ignored
I know how the backtester works but I am talking about the trades occuring within the backtest. Seems as though reducing the profit factor can't stop you from getting into those high DD situations which lose a big portion of your account.
Entry filter would then try to eliminate these sets that cause the problem.
So your supposition is that when the price closes above or below this i-regr ( bolliger band type indicator) then you should take a trade?
I like the idea since either it would be a breakout or sharp retracement if it reaches the outer edges of the i-regr band.
I am testing it out on Viper Grid 2.5
Jag
Ignored
Not Exactly. You have to enter ONLY if price crosses the iRegr line AND it has had one or more closed bars outside the iRegr channel. Besides, you have to enter in the right direction. For example - If the iRegr is slopped up enter only at the price crossing the BOTTOM iRegr line in direction from outside to the inside of the channel after it has closed one or more bars outside the iRegr channel. If the iRegr channel is slopping down enter at the price crossing the TOP iRegr line in the direction from outside to the inside of the channel after it has closed one or more candles outside the iRegr channel.
Another idea which would make the EA even more selective in the entries - place the pending orders ONLY if you have a Moving average cross in the direction of the iRegr channel slope. I have used MA 5 (close) and MA8(open) very successfully to manually trigger the EA to place pending orders. For example - if iRegr channel is slopping up (up trend) ONLY place the pending orders if MA5 crosses OVER MA8 AND the crossing is OUTSIDE the iRegr channel. All those parameters should be of course optimizible.
It is a good idea to set a profit exit - the EA should close all open and pending orders when a certain profit is reached. I have had it so many times the price gets to 1-2 pips from the TP level and reverses to give a big drawdown (read - hearth attack). Include also a Trailing stop - not on the individual positions but on the total profit - that may add couple of more years to your life before hearth attack.
It is a good idea to set a profit exit - the EA should close all open and pending orders when a certain profit is reached. I have had it so many times the price gets to 1-2 pips from the TP level and reverses to give a big drawdown (read - hearth attack). Include also a Trailing stop - not on the individual positions but on the total profit - that may add couple of more years to your life before hearth attack.[/quote]
One more thought about the Take Profit Idea. If you set the TP (that would be s $ amount) to a very low amount - let's say $10 then you can use the EA almost as a scalper. So, it would be a scalper that always wins or a very safe Grid EA.
Not Exactly. You have to enter ONLY if price crosses the iRegr line AND it has had one or more closed bars outside the iRegr channel. Besides, you have to enter in the right direction. For example - If the iRegr is slopped up enter only at the price crossing the BOTTOM iRegr line in direction from outside to the inside of the channel after it has closed one or more bars outside the iRegr channel. If the iRegr channel is slopping down enter at the price...
Ignored
I tested it out and it gave a margin call on the second set lol. Problem is when a say a long trend is suddenly a short trend iRegr takes a while to adjust, and when price reaches within the boundaries again after a while an entry is placed which is bad. iRegr may work within the current major trend up or down but fails when it changes. The bad entry could also be avoided by maybe using your MA cross idea, or simply by saying dont enter if MORE THAN x BARS closed above or below the iRegr boundary (probably trend changing).
I tested it out and it gave a margin call on the second set lol. Problem is when a say a long trend is suddenly a short trend iRegr takes a while to adjust, and when price reaches within the boundaries again after a while an entry is placed which is bad. iRegr may work within the current major trend up or down but fails when it changes. The bad entry could also be avoided by maybe using your MA cross idea, or simply by saying dont enter if MORE THAN x BARS closed above or below the iRegr boundary (probably trend changing).
Jag
Ignored
Let me have it, I would like to test it too. Did you optimize it and what time frame did you use it? Since you have an iRegr it is time dependant. My experience with mGrid is that you have to test it on 1 hr or 4 hours, 4 hr is best but it gives you the least profit.
It is even better if you have the take profit in $ amount
For the lots, you would have to use the high/low water mark and set the value less than the broker allowed amount.
As for your trade results, why do the settings (input parameters) shown in your version 2.4 results contain parameters added only for version 2.5 ??? They did not even exist in v2.4...
consolidating orders may help for backtesting purposes as I believe it is based on 200 max orders (with Alpari UK).
However, this wouldn't help on accounts where a limit is set on the amounts of lots you can trade. Alpari's micro account has a 2 lot maximum (which happens to be 200 x 0.01lots!). Perhaps the demo has the same limit, but I don't think so.
I also have attached my results (and report in the zip file to show my settings are identical) as I have had widely different results between vgrid 2.4 and 2.5 with all settings equal - [b]can...
For the lots, you would have to use the high/low water mark and set the value less than the broker allowed amount.
As for your trade results, why do the settings (input parameters) shown in your version 2.4 results contain parameters added only for version 2.5 ??? They did not even exist in v2.4...
Viper
Ignored
Yes, that is weird as stealth didn't exist in v2.4. I didn't rename the file or anything. And even if 2.5 somehow morphed into 2.4 - you'd think it would give similar results....
I may have to just do a fresh install or something.
were you using the protective features in vgrid when you did the testing....
Ignored
Hi Plasma,
Sorry it took me while to get back to you. Yes , I was using the protective features - actually only one - the trend filter. The volatility filter does not seem to work. The high-low water filters were set to 0.
ViperGRID is an excellent EA but it still lacks some better entry and exit. My suggestion was to use the IRegr channel, but just in general i believe that a standard deviation channel or even regression channel will do the job, and may be even better. Tis for the entry side.
On the exit side - A simple TP for the entire set and a trailing stop in $amount would do the job. See my other postings, I just do not want to repeat myself.
Hi Viper, I did not get back to you because I was hopping you have seen my other postings and noticed that I posted the code of mGrid with iRegr long time ago. I took me a while to digest the turkey and come back to life.
Anyway - I have been testing the Viper Grid and I find it be an excellent EA. On the back tests, however, I noticed that it does not place all the orders all the time - this is a problem I had with mGrid EA too. For example - If the levels are set to 3, sometimes it places 3 long and 2 short pending orders or some times it places only the long orders.
I am still trying to figure out the logic on which it places the new orders when a level is hit. I would appreciate it if you save me the effort and just tell me the logic - is it a formula you are using to determine the amount of lots to place on the different levels or it is just based on empirical experience. I myself am working on a formula which supposedly has to lower the DD and I just did not want to reinvent the wheel.
I believe with a better entry ( iRegr channel, Standard Deviation Channel or even just a regular regression channel ) we can reduce the DD. I have not heard your opinion of using the iRegr channel yet.
Exit is the other end of it - Can you please just add a TP in $ amount for the whole set. If you have the TP then you can easily have a "rolling grid". Just for the example - imagine you have a 0.1 lot ($1/pip) and an increment of 10pips on let's say 3 levels. If you set the TP to $10 then the grid is going to close when the position bought on level 1 hits level 2 and takes profit. If the entry conditions are still in place a new grid will open with the zero line (the pivot) at the level of the previous level 2 of the old grid.
Here is a ? for you. If I have a number in the input field "Initial Pivot Price" do I still have to have all entry conditions in place to have the pending orders placed or they will be place when price hits that level regardless of the entry conditions.
Happy trading.
Ogie D.
PS. I am not sure but i believe i saw it in one of the other postings - te Volatility filter does not seem to work.
Hi Ogi, I wanted to apologize. I did read your postings but did not examine the source code you posted. I thought they were indicators, not the EA.
I have updated ViperGrid to include a new volatility filter, as Ogi has suggested (I don't have the iRegr indicator so I used standard BBands instead), changed the logic of the price action filters (trend and volatility) so that if ANY are a go we can start a new grid instead of needing ALL to give the go. Added "sweet spot" starting price along with specified starting price so if price goes either way you will catch the next psychological S/R. Fixed a few minor bugs.
I will need to test new features when the market opens but expect to release the next version next week, assuming no bugs are unearthed :-)
Hi Viper, I did not get back to you because I was hopping you have seen my other postings and noticed that I posted the code of mGrid with iRegr long time ago. I took me a while to digest the turkey and come back to life.
Anyway - I have been testing the Viper Grid and I find it be an excellent EA. On the back tests, however, I noticed that it does not place all the orders all the time - this is a problem I had with mGrid EA too. For example - If the levels are set to 3, sometimes it places 3 long and 2 short pending orders or some times it places...
Hi Ogi, I wanted to apologize. I did read your postings but did not examine the source code you posted. I thought they were indicators, not the EA.
I have updated ViperGrid to include a new volatility filter, as Ogi has suggested (I don't have the iRegr indicator so I used standard BBands instead), changed the logic of the price action filters (trend and volatility) so that if ANY are a go we can start a new grid instead of needing ALL to give the go. Added "sweet spot" starting price along with specified starting price so if price goes either...
Ignored
Thanks Viper,
I do not know if the BB will do the same job. Just in case you want to use the iRegr Indicator I am attaching it here.
Did you have a TP for the set inserted in there?
I am working on an idea of a rolling grid and will present you with it very soon. I just have to put it together in the right format
