Snowballs and the anti-grid | Page 8 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Grid - New indicator that replaces chart grid 60 replies

Grid (not grid system) or horizontal lines indicator 6 replies

Trading the Sixths with Snowballs 68 replies

Grid, Burst Grid, Grid Risk and Grid Profit 1 reply

Double Time Grid and Pip Grid 0 replies

Attachments: Snowballs and the anti-grid
Exit Attachments

Snowballs and the anti-grid

  • #141
  • Aug 10, 2010 10:08am Aug 10, 2010 10:08am
I have some questions about your EA if you could respond please. I have it on my charts placing trades and orders correctly with common functions in my includes folder. I have a few problems however. I have no green slash line showing BE. I can't change anything in the code without getting errors. I'm also looking for a line in the code to change so that after TP occurs, new buy and sell orders would automatically reopen without a manual restart. Yes I realize you told us this was not a set and forget EA but I'd like the option for an auto. restart for some occasions. thank you
  • #142
  • Aug 13, 2010 11:28pm Aug 13, 2010 11:28pm
Thanks for a great EA. Can I offer a couple of usage ideas to help perhaps.

1) Entry of time to start running the EA after mode is selected.
2) Allow the stop to be set by width of the bollinger bands -dynamic
3) Auto close trades on condition. Example: Long, made profits, close all
when high is less than upper bollinger. Disable expert until next session.

The idea here is to autostart the expert. Autostop the expert. Maximize profit. This way we can get some sleep too and not have to start/stop manually.

Before london open, range is usually narrow --- our stop on ranging prices.
But as volitility increases rapidly new buystops should have wider stop as the width of the bollingers. Then when volitility drops off the first highest bar price inside the bollinger will close out all trades and disable trading until next london session. Visa versa for sellstops positions.

These are my ideas. Thank you.
  • #143
  • Aug 23, 2010 5:20am Aug 23, 2010 5:20am
Hi Gio,
Are you still using the anti-grid approach? Any updates? Thanks.
Quoting gio.kele
I am starting snowball on the london open on gu,eu and ej with bidirectional mode.
  • #144
  • Aug 23, 2010 7:41pm Aug 23, 2010 7:41pm
Hi 7bit,
Are you still trading this idea? Any new insights or updates? Thank you.
  • #145
  • Aug 24, 2010 1:26pm Aug 24, 2010 1:26pm
Quoting yesbee1
Hi 7bit,
Are you still trading this idea? Any new insights or updates? Thank you.
I still trade it but I don't think I have any significant new insights. The only thing I can tell you is be careful with your lot size, never ever (NEVER) use more than leverage 1 for the initial level (0.01 lot if you have $1000 in the account, 0.1 lot if you have $10000) (all snowballs on the same account combined)! Trade only one snowball at a time, this is better for the psychology. Don't be greedy, take profit at one level above BE or make intelligent usage of the pause/resume function. Prefer EURUSD, this seems to have very snowball friendly characteristics.

No software updates yet, but still a few minor bugs. Not much time at the moment.
  • #146
  • Aug 24, 2010 2:20pm Aug 24, 2010 2:20pm
Hi 7bit
Thank you for your response.
  • #147
  • Nov 23, 2010 1:50am Nov 23, 2010 1:50am
Hello 7bit,

I have a newbie question that I hope you'll entertain. I'm on Oanda, so I can't make use of your EA. I read your website, but I did couldn't fully understand the various modes and how each mode differs. I'm trying to replicate your logic manually in Oanda, and I was hoping that you might expand the description of the logic a bit.

Aside from the 4 modes, I'm trying to figure out some basics:

1. When placing buy and sell stop orders above and below the current price, at what spacing? If the spacing is 30 pips, then is the buy at 30 above the current price and the sell 40 below the current price? Or is it 15 pips above and below? Or do you place buy and sell at the same price?

2. For each order, it seems there are stops at twice the spacing? So, at 30 pip spacing, the stop for each order is 60?

3. Since we cannot have long and short positions at the same time, you're essentially closing a position and opening the opposite position when your stop is hit? (Sorry, Oanda doesn't have buy stop orders, so I don't quite understand how this works...) How can I replicate this on Oanda? Rather than using stops, place opposing limit orders with 2x the volume?

4. I think I read that you take profit at 2x the spacing? Wouldn't that simply be canceled out by the stops?

5. If the market ranges forever within a range that hits your stops over and over again without ever breaking out, your account will be fully drained?

Thank you very much for any help you might give
  • #148
  • Edited 7:32am Nov 23, 2010 6:50am | Edited 7:32am
Quoting snipescout
1. When placing buy and sell stop orders above and below the current price, at what spacing? If the spacing is 30 pips, then is the buy at 30 above the current price and the sell 40 below the current price? Or is it 15 pips above and below? Or do you place buy and sell at the same price?
if the spacing is 30 pips then every order happens in this fixed 30 pip intervals. You buy every 30 pips or you sell every 30 pips, nothing happens outside this fixed grid.

Quote
2. For each order, it seems there are stops at twice the spacing? So, at 30 pip spacing, the stop for each order is 60?
No. If the grid spacing is 30 pips then each 30 pips is a buy order and each buy order has a stoploss (which is the same a sell order) 30 pips below it. In other words: for each buy order there is a corresponding sell order (that is implemented as a stoploss in MT4) 30 pips below it (or if you view it from the other side: each sell order has its buy order 30 pips above it).

A stoploss of a buy is the same as a sell order (in oanda you would implement it as a limit sell, but you can of course also use stoplosses instead, it makes no difference, it is the same in every aspect). You must fully understand why they are equal until you can understand the concept as a whole. This will also make you understand why there is no use in nedging (same pair hedging) and *everything* can be reduced to buy and sell and every close of a buy is a sell (and a sell can and will act as a closing of a buy and vice versa and all that remains is the net difference). This is essential.

In Mt4 it is implemented with stoplosses which are also just buys or sells into the other direction. I cannot use stop orders to scale out in mt4 because the stupid MT4 would try to open additional trades (which would have the same effect on profit or loss but would slowly eat up margin and i would have to close out these nedges from time to time. Therefore I use stoplosses.

Quote

3. Since we cannot have long and short positions at the same time, you're essentially closing a position and opening the opposite position when your stop is hit? (Sorry, Oanda doesn't have buy stop orders, so I don't quite understand how this works...) How can I replicate this on Oanda? Rather than using stops, place opposing limit orders with 2x the volume?
There are no opposite positions open at any time. You are either long or short. In oanda there are limit orders which is the same as buy/sell-limits or buy/sell-stops. They are pending orders, how they are called is not so important for our purpose.

In oanda you would place limit buys above price and limit sells below price. the buys would scale in when it goes up and the sell would scale out again (and finally reverse) when it goes down. For each triggered buy you would place a new sell 30 pips below it and for each triggered sell you would place a new buy 30 pips above it. They could be viewed like stoplosses for each other, it is completely symmetrical. All orders have the same volume.

Quote

4. I think I read that you take profit at 2x the spacing? Wouldn't that simply be canceled out by the stops?
I take profit anywhere above breakeven. 2*spacing is just an arbitrary number. sometimes I leave it running longer, sometimes i also close with a loss and wait for better times.

The stops have nothing to do with it. The stops behind price are to scale out if it goes back to reduce the loss and the stops in front of price are used to scale in and add even more to the winner.

Quote

5. If the market ranges forever within a range that hits your stops over and over again without ever breaking out, your account will be fully drained?
yes.

You need nerves of steel and great patience, there is a slow constant bleeding that can go on for weeks until it finally makes a directional move and shoots through a bunch of entry orders and the total floating profit makes a positive spike to reward you for all the losses. The time window for taking profit (immediately closing all trades) is very short. You will have to stay in front of the screen 24/5 if you want to do this manually.

It is like being on the other side of a martingale game (you are the casino, the market is the gambler), the market makes constant *small* gains but you know the day will come when the bets are large, the table limit is reached (you decide the limit) and the gambler will make a big loss when you pull the plug at the right moment.

You can apply the same logic to any other form of martingale or martingale-like systems, just reverse it in such a way that the market is martingaling and you are the casino. The market (the player) has an infinite supply of capital to survive any martingale but you (the casino) have the power to be as unfair as you want and stop the game at any moment (preferably exactly when it would be maximal uncomfortable for the martingale player).
  • #149
  • Nov 25, 2010 5:58am Nov 25, 2010 5:58am
Thank you 7bit for going through and responding to my questions, despite a couple of typos and some questions being redundant.

I've been admiring you from afar from reading your posts elsewhere of your Oanda non-api autotrading, so I downloaded AutoIt, and so far, all I figured out is how to auto login. It's going to be long road, I can see. I have no programming training.

Since all programming is new to me, do you think I'm better off just getting the Oanda API (assuming I've got 12M volume)? If so, I get the sense that Java is easier to program than C++, but Java is much slower. Do you have any opinions on this?

Thank you again!
  • #150
  • Dec 30, 2010 8:15am Dec 30, 2010 8:15am
Quoting snipescout
Hello 7bit,

I have a newbie question that I hope you'll entertain. I'm on Oanda, so I can't make use of your EA. I read your website, but I did couldn't fully understand the various modes and how each mode differs. I'm trying to replicate your logic manually in Oanda, and I was hoping that you might expand the description of the logic a bit.

Aside from the 4 modes, I'm trying to figure out some basics:

1. When placing buy and sell stop orders above and below the current price, at what spacing? If the spacing is 30 pips, then is the buy at...
i am also using oanda. and i have to do it manually. i use a excel tracker to track the entries and exits, its laborious , but no other option. thats why i use a higher pipstep 50. this gives some breathing space
  • #151
  • Edited 5:31am Jan 5, 2011 5:15am | Edited 5:31am
Quoting snipescout
If so, I get the sense that Java is easier to program than C++, but Java is much slower.
This is a myth. Java is just as fast as C++, some operations can be made even faster (optimized better) in Java than in C++ because the JIT compiler has more knowledge about the data types. There are various benchmarks that prove this. If you hear somebody saying "Java is slow" without specifying exactly *what* it is that is supposed to be slow, then it is most likely just some uninformed FUD based on knowledge (if any) from the year 1980.

Also if somebody tells you the only other alternative to Java or C# is C++ (or the other way round: the only alternative to C++ is Java (or C#)) then you instantly know that the one you are talking to has no clue about programming. The same is true for people telling you that C++ is a good programming language (or they like C++) => no programmer.

Trust only people who are very critical about C++ and know (and use) more than only 2 languages (not including Java, C#, php¹).

____
¹) I explicitely mention this odd language that doesn't really fit into the row because it is the ultimate wannabe-indicator. Every 12 year old claims to know this language (but none of them can master it and those who can will carefully avoid ever mentioning this dirty word). If you find this on someone's résumé => recycle bin.
  • #152
  • Jan 5, 2011 6:31am Jan 5, 2011 6:31am
Quoting 7bit
This is a myth. Java is just as fast as C++, some operations can be made even faster (optimized better) in Java than in C++ because the JIT compiler has more knowledge about the data types. There are various benchmarks that prove this. If you hear somebody saying "Java is slow" without specifying exactly *what* it is that is supposed to be slow, then it is most likely just some uninformed FUD based on knowledge (if any) from the year 1980.
The myth is on you. Yes, in theory, a JIT compiler can do more magic than a static compiler, but this is just IN THEORY. In practice, Java is much slower because no JVM yet implemented these THEORETICAL optimizations. And C++ compilers are not standing still either. WPO and especially PGO optimizations (to use Microsoft terminology) do some of the stuff which supposedly is not possible to do in an ahead-of-time compiler (like inlining virtual methods, yes VIRTUAL).

And about benchmarks: http://shootout.alioth.debian.org/u3...gpp&lang2=java

Java wins only one benchmark from this set. Also look at memory consumption while you're at it. Where are those advanced JIT optimizations? Oh, I forgot, NOBODY implemented them yet. Guess next year....

BTW, the Facebook front-end is written in PHP. And then it's automatically translated to C++ (FOR SPEED). But what do they know... probably just a bunch of wannabees.
  • #153
  • Jan 5, 2011 11:09am Jan 5, 2011 11:09am
Quoting Adal
The myth is on you. Yes, in theory
Lol, I knew that such an answer HAD to come. The speed differences between Java and C++ are *irrelevant* in all cases and RAM is much cheaper than manpower.

The much cited language shootout is *irrelevant* since the different implementations of the same algorithms are all written by entirely different people, differently organizing their data structures and doing things not always the way it is meant to be done in the specific language.

Quote
BTW, the Facebook front-end is written in PHP. And then it's automatically translated to C++ (FOR SPEED). But what do they know... probably just a bunch of wannabees.

If this is true then this is 100% proof for what I have said about php. If the usage of php leads to such bizarre constructs as a last resort to avoid the ruin of a business then one should think twice the next time before hiring the next wannabe php script kid. And yes, they obviously know nothing or they would have done it better.
  • #154
  • Jan 5, 2011 11:38am Jan 5, 2011 11:38am
Quoting 7bit
The speed differences between Java and C++ are *irrelevant* in all cases and RAM is much cheaper than manpower.
But you said that Java is can be much faster. How come in the real world is much slower? And why aren't the Java programmers offended by such bad benchmark implementations which give theirs language a bad name? And I guess you didn't heard of cache-pressure and working-set size - RAM is not as cheap as you think it is.

Quoting 7bit
The much cited language shootout is *irrelevant* since the different implementations of the same algorithms are all written by entirely different people, differently organizing their data structures and doing things not always the way it is meant to be done in the specific language.
Please, point to the relevant benchmarks then. Or you just know the truth and don't need some irrelevant micro-benchmarks.

Quoting 7bit
If this is true then this is 100% proof for what I have said about php. If the usage of php leads to such bizarre constructs as a last resort to avoid the ruin of a business then one should think twice the next time before hiring the next wannabe php script kid. And yes, they obviously know nothing or they would have done it better.
Yes, they are true failures. Building a 50 billion company starting from 0 in 5 years. What a bunch of losers. If they started with a "real" language like Java, they would be worth 1000 billion today. Or maybe they wouldn't exist, because by the time they implemented their site in Java....
  • #155
  • Jan 6, 2011 7:41am Jan 6, 2011 7:41am
Hi 7bit,
Just finished reading this thread and your website, and trying out your EA. I think linearly increasing loss against geometrically increasing profit is a great concept. I had a similar idea that I worked on, but this is a much more polished version. Kudos to you.

Thanks !
  • #156
  • Edited 6:43pm Jan 6, 2011 6:32pm | Edited 6:43pm
Quoting Adal
But you said that Java is can be much faster.
I said that it is as fast as C++ (or not notably slower) (implying that nobody will force you to do things in Java that are known to waste CPU)

Quote
How come in the real world is much slower? And why aren't the Java programmers offended by such bad benchmark implementations which give theirs language a bad name?
because they usually just don't care about it. And many of them simply don't know what they are doing.

Quote
Please, point to the relevant benchmarks then.
You can try it yourself. Just recently I did exactly this, while trying to track down some odd double precision strangeness when using the x87 FPU I wrote a little program (5 lines) to approximate the value of e in Java, in C and in Pascal. The immediate result was that the Java program was twice as fast as my C and Pascal implementations and only after using -O3 in gcc I got it as fast as the Java version (while triggering the above mentioned deviation in the end result that also was present in the Java version) and only some hand written inline assembler (after disassembling the gcc output to see what clever trick it was using) in my Pascal program could make this also equally fast. The optimization resulted in completely leaving all variables always on the FPU stack (thus suddenly using 80bit precision instead of 64 bit) and only a handful of carefully chosen assembly instructions surrounding it to do the looping.

All these micro benchmarks uncover an amazing execution speed that is not worse than gcc -O3. Only when you start instantiating all these bloated classes from the standard libs for *every* little thing the performance will drop massively.

Quote
Yes, they are true failures.
Yes they are. Somebody who compiles old php code written by incompetent php kiddies 10 years ago to C++ (only to see that it won't fix the problem in any way because these scripts are IO bound and spend all their time waiting for IO) instead of hiring a few experienced programmers is truly incompetent. The fact that other incompetent people are throwing huge amounts of money at them does not make them more competent. But this is not my problem.

And I personally would not use Java either, there are more appropriate languages and frameworks and tools to write server applications that can handle such massive parallel IO requirements without blocking and unneeded CPU and memory consumption.
  • #157
  • Jan 7, 2011 7:46am Jan 7, 2011 7:46am
Hi

I like to use snowball to accumulate lots as the price go up and maybe use MPTM to close with a basket profit.

Any chance of having snowball mod with no stop loss?

Regards
Ian
  • #158
  • Jan 12, 2011 1:35pm Jan 12, 2011 1:35pm
Quoting mallim
Hi

I like to use snowball to accumulate lots as the price go up and maybe use MPTM to close with a basket profit.

Any chance of having snowball mod with no stop loss?

Regards
Ian

I too do not understand the need for negative stoploss which can bleed the account to death.
The single losing trade will be outnumbererd by winning trades once the trend starts, close as a basket. If distance is 30pips, at +25 or just before the next level is opened have the current trade set a SL of +10. Then if market retraces before basket closure we get +10 on the positive trades and only one of the positive trades can reverse to become a negative, which in turn becomes a positive if the trend continues the other way. So we get a few +10 profit trades and no losses while ranging. Huge profit when the trend takes off!

Please, Please make a version without negative SL for experimentation.

Thanks
  • #159
  • Jan 13, 2011 12:45pm Jan 13, 2011 12:45pm
Quoting Penny
I too do not understand the need for negative stoploss which can bleed the account to death.
Thanks
The whole point in using stoplosses is to scale out if price goes into the wrong direction. There is no other way to avoid huge losses than immediately closing the trades when price reverses. And also the stoploss should not trigger before price is below the entry level and has a chance to trigger the entry again or you would lose levels.
  • #160
  • Jan 13, 2011 5:18pm Jan 13, 2011 5:18pm
Quoting Penny
I too do not understand the need for negative stoploss which can bleed the account to death.
The single losing trade will be outnumbererd by winning trades once the trend starts, close as a basket. If distance is 30pips, at +25 or just before the next level is opened have the current trade set a SL of +10. Then if market retraces before basket closure we get +10 on the positive trades and only one of the positive trades can reverse to become a negative, which in turn becomes a positive if the trend continues the other way. So we get a few +10 profit...
Then just use the MakeGrid193 ea...

The Snowball ea is a great piece of work- one of the few ea's avail here on the forum that works 'bug free'. It's great as a 'trade manager' if you just randomly place it on a pair and pick long or short without proper analysis- then yes you will be stuck with huge DD.
The comforting piece of this EA-- what IF your analysis is wrong-(Sometimes I am) or the trend turns against you for a day or two-- the EA minimizes losses by not re-engaging in the trade until price as moved back to the trend.

Luiboy has a great thread started- and it looks like possibly some good results combining with another analysis tool.

http://www.forexfactory.com/showthread.php?t=267952
0 traders viewing now