hyperdimension

Don't be fooled by the tight spread. Commission is far too high, and they are not regulated.

It would be very impressive if their statement about small stop loss is true:

# Small stop losses mean lower risk (10-12 pip stops)

What an idiot. I found a while ago that he has a web site where he sells EAs that I assume he gets from here. Ban this guy.

 

Anyway, why would anyone want an MQ4 To EX4 "Decompiler"? If you want to convert an MQ4 file to EX4, MetaEditor does that for you.

Past a 1000$ account at 4500 in 6 month in takeit 0.01 at 0.29 position (0.29 just 1 time in 6 month), i think it's very very good.

Now start the backtest just before the largest drawdown that you see in those results (around trade 296), and see if it crashes.

 

A lot depends on the initial deposit, the available leverage, the stop-out level, and how soon you get into a big drawdown from the start - if there's not enough equity at the time of a big drawdown, it will crash. In those backtests, it didn't crash because there happened to be enough equity (from accumulated profits) during those times of large drawdowns.

There is a big jump in equity due to the following trades:

42638269	2010/05/21 21:11:00	Buy	 8.00	EURUSD	1.2458		1.257	2010/05/21 21:16:00	1.257	 0.00	 0.00	8 960.00	 112.0
42638276	2010/05/21 21:12:00	Buy	8.00	EURUSD	1.2458		1.257	2010/05/21 21:16:00	1.257	0.00	0.00	8 960.00	112.0
42638282	2010/05/21 21:12:00	Buy	8.00	EURUSD	1.2458		1.2558	2010/05/21 21:16:00	1.2558	0.00	0.00	8 000.00	100.0
42638350	2010/05/21 21:15:00	Buy	3.00	GBPUSD	1.4357		1.4454	2010/05/21 21:16:00	1.4448	0.00	0.00	2 730.00	91.0

 

Notice that they were all closed at 21:16:00. The first two EURUSD trades above closed at the price 1.257, however the third closed at 1.2558. It is a bit suspicious to see trades closed at the same time at different prices (12 pip difference).

There could be a possibility that times in the statement only have minute precision (and not seconds), and the price did drop from 1.257 to 1.2558 within the same minute.

If you want to check whether price did move that much within the minute, you would have to find out what broker he used so that you can check on the chart at the correct time in history.

Noticed that the risk reward ratio is 1:1.

Are you looking at a different EA?

 

Take profit distance = 10 pips

Stop loss distance = 150 pips

 

After a losing trade, it will enter the next trade with twice the lot size of the preceding losing trade. We saw 0.01, then 0.02, then 0.04 and then 0.08.... Once the trade is a winner, the lot size for the next trade will resume at 0.01. Guess this multiple lot sizing should be configurable.

I don't see this trade size progression. In the backtests it's 0.3 lots, becoming 0.4 lots as equity increased. Are you looking at the same EA?

My understanding is that you need to have a candle close first that has brought the RSI either above or below the 10 -10 level. If the candle hasn't closed then there is a chance the RSI could move back into the no trade zone. If the candle has close and all conditions are met then trade

But needing a candle close does not explain the "at least one" part, as it insinuates that you need to look at multiple RSI Histo bars and check that there is at least one that is above 10.

 

I scanned the document and found a possible answer on page 52:

Look at the RSI Histogram indicator. You can see that it helped to ensure that there were no bad trades taken since not one Histo bar closed above the +10 line throughout the day.

Notice the phrase "throughout the day". That suggests we need to look at all RSI Histo bars "throughout the day" to check that at least one of them is on or above 10. However this is the only sentence throughout the entire 89 page document that specifies the period to check at multiple RSI Histo bars.

 

So what exactly is meant by "throughout the day"? It is never explained, but I would assume it is the time period between previous and next 00:00 GMT.

 

Now that the "at least" part is explained, it is in complete contradiction to what he says on Page 22:

To confirm: only enter a trade at the start of the next 15 minute candle once the Histo bar of the current candle has closed on or over the +10 or -10 level.

Instead of saying "at least", he only points to the RSI Histo level of the current candle. Everywhere else, he uses the "at least one bar' phrase when referring to the RSI Histo bar condition.

I'm reading the manual and the rules don't seem perfectly clear. Here are the rules for a Buy entry:

 

Condition 1: The black line MA (“MA(7)”) needs to have crossed above the blue MA (ie, the “MA(21)”) indicating a change in the trend.

Condition 2: A price candle needs to have crossed and closed above the short-term trend line - red MA (ie, the “MA(84)” line) for a BUY trade. TIP: the closer the cross of the black and blue MAs to the red MA, the stronger the trade is likely to be.

Condition 3: The CCI(5) needs to be on or above the “0” line at the start of a 15M candle (not part way through the 15min time period) for there to be a valid BUY trade.

Condition 4: The RSI Histo (13) indicator must have at least one 15M bar closed above the +10 level line for a valid BUY trade.

 

My understanding of the rules are the following:

At the moment of a 15 minute bar open, open a buy trade if all of the following are true:

1. Black MA > Blue MA

2. Previous candle close > Red MA

3. CCI >= 0

4. RSI Histo >= 10

 

Notice particularly that my interpretation of the first and second conditions are different to the author's written rules. The author used the phrase "needs to have crossed above", which technically means an entry is only valid if a cross has just occurred, not 2 or 3 bars ago. But based on examples in the manual (page 34 bottom chart), the crosses do not necessarily need to have just occurred (on or after the previous candle); it's enough that one is above the other (which means a cross did happen some time ago). This distinction between a "cross" and "greater than" needs to be highlighted.

 

With condition 4, I'm trying to understand the "must have at least one 15M bar closed above the +10 level" part; why does the author need to say "at least one 15M bar"? It would only make sense if you have to look at multiple historical bars (e.g. the last 5 RSI Histo bars), but nowhere is such an action specified. Should the rule simply be "RSI Histo is on or above 10"?

Just wondering, have you been testing or using this author's method? Is it valid or profitable?

I haven't thought about the actual strategy much, so I don't have much to say about it.

 

I originally was interested in taking a look when you claimed that it was written like a university doctorate thesis, only to find that it was very informally and sloppily written in just 21 pages, and because of this I have some doubts to whether the author does actually have a PhD.

 

Anyway, it could still be worth testing.

I am particularly confused about the author's wagering and staking logic/plan.

Do you mean this section?:

 

The Betting Algorithm

$200.00 of your account is specified as your Reserve. $100.00 of your account is specified as your fractional betting equity, or FB for short.

 

Define the risk on your trade. Risk on the trade is the difference between your market entry price and your stop-loss protective order. If the difference between your entry price and stop-loss is 54 ticks/pips/points. Your risk is 54.

Whenever the FB is large enough we will bet fractionally:

 

Risk 15% of the FB on the trade.

 

If your FB is $100.00, and your risk is 54, and the pip value of 1 lot is .10 (10 cents), 54 x .10 = $5.40. $5.40 x 3 is $16.20, close to 15% of your FB. So, you would trade 3 lots.

 

If your FB drops to a level were 15% risk would be less than or equal to 1 lot, trade only 1 lot. Continue to trade only 1 lot until your FB increases to the point where you can once again trade multiple lots.

 

As your FB increases, bet 15% of the new FB.

 

It is simply trade sizing such that no more than 15% of the Fractional Bet (FB) amount will be lost if the stop loss is hit. His calculation is backwards however, and his definition of one "lot" is actually a micro lot (1000 units of base currency).

 

Using his example values, if FB = 100 USD then you would only want to lose 15 USD if the trade hits the stop loss.

The stop loss distance is 54 pips.

A 1 pip movement in the price causes a change of 0.1 USD to your profit if you had 1 micro lot open.

So the right trade size would be: 15 / (54 * 0.1) = 2.78 micro lots.

Based on some of the previous comments, I was expecting to see a lot of university level statistics in the book, but it actually turned out to be fairly simple, with no strange Greek symbols, and it's just 21 pages!

 

If you want to see some really deep mathematical stuff you should check out some Quantitative Finance / Econometrics books and academic research; I'm still trying to get into it, after downloading hundreds of books and research papers... I'm a wanna-be "Quant".

http://www.myfxbook.com/members/pc8multifx/xbarsx-1-c9x-1all/30527

Notice that he has changed the name to try to hide the fact that he is using XBars:

http://www.myfxbook.com/members/pc8multifx/x-1-c9x-1all/30527

Results look good so far but not long enough to be conclusive.

 

We also need to see backtest results.

 

This particular question and answer is quite worrying:

What's the minimum account size Pip Brains will work with?

 

Pip Brains will work on any account size, so don't be shy!

Since it is a martingale-grid strategy, initial equity, initial trade size, leverage and account stop out level are some of the most important things that can determine whether the account crashes or not. So it is foolish for the vendor to state that any initial account size is fine, as it may place the customer in great risk.

As for EA I would not expect anything as he shows only cherry picked trades with long candles and on backtests only.

Where are the backtests?

it could practically bomb out tomorrow, figuratively speaking.

not saying it will, but i guess caution is due.

/bratt

It bombed out just 12 days after your warning: http://www.myfxbook.com/members/MT4Trading/harvester-live/18004

here ... live results!

 

http://micutzu.mt4live.com/

 

It looks like you are way overleveraged. You started with around 1500 USD and you have 0.7 lots of GBPUSD open plus 0.1 USDCHF. You should be trading micro lots instead with such small starting equity.

I've found Fiddler to be very useful.

I have spoken and conducted in depth studies with some very brilliant math PhDs in economics and all of them without reservations concluded that a concept based on martindale, progression, labouchere is doomed to fail, the equation being just a matter of time unfolding itself.

I think Everything you wrote in that post is correct if we assume that the markets behave as a purely random walk, which I believe those mathematicians would have done. However if this assumption is correct, this means that trading itself cannot be a profitable endeavor over the long term, as every trade would be like tossing a coin. The additional loss of the Ask-Bid spread would run an account to the ground eventually.

 

But if we eradicate this random walk assumption and state that markets do have some statistically significant patterns of behavior that can be exploited, then trading can be profitable, and those mathematical models that are based on the random walk assumption can no longer be used. I think in this case martingale trade sizing can be profitable, and it would depend on whether the underlying strategy itself is profitable.

Trade sizes are constant, so it is not Martingale. Do you mean "Grid"?

Does it actually work? Or is it just voodoo?

Here are my Skyline backtest results using Dukascopy tick data and real Ask prices (therefore variable spread).

 

Commission = 3 USD per lot round turn.

 

All parameter values are default except:

LotSize = 0.1

 

Date range = 2008-04-17 to 2010-05-02

 

Uisng martingale trade sizing:

http://stashbox.org/884046/Skyline%20RealAsk_Data%3DDukascopy%20ticks_Symbol%3DUSDJPY_Commission%3D3USD_LotSize%3D0.1_2008-04-17to2010-05-02_ScreenCap.gif

 

Using constant trade size:

http://stashbox.org/884047/Skyline%20RealAsk_Data%3DDukascopy%20ticks_Symbol%3DUSDJPY_Commission%3D3USD_LotSize%3D0.1__ConstantTradeSize%3D1_2008-04-17to2010-05-02_ScreenCap.gif

If base trade size is 0.01 lots, and stop loss distance is the default 55 pips, then 9 consecutive losses would cause a total loss of almost 3K USD. The trade size series of the losing trades would be:

0.01

0.02

0.04

0.08

0.16

0.32

0.64

1.28

2.56

 

To be able to open the next trade (5.12 lots), assuming leverage = 500, you'd need a little over 1K USD left in the account. So around 4K USD of equity is enough to withstand a maximum of 9 consecutive losses, assuming base trade size is 0.01 lots.

 

To be able to withstand a maximum of 10 consecutive losses and be able to open the next trade, you'd need around 8K USD of equity.

 

It's the consecutive losses that kill you when using trading systems that use this kind of simple Martingale-based trade sizing. As long as you have high leverage, low account stopout level, enough equity (deposit + profits), and a good underlying trading strategy (trade entry and exit), it can work.

For ForexSkyline, what was the largest trade size you saw in backtests?

 

The code is very simple (less than 150 lines). The order modify code is poorly written, such that it has to modify the trade 3 times to add the Stop Loss and Take Profit levels.

I made changes to the code and am now backtesting this using Dukascopy tick data with real Ask prices, though it should look very similar to Stormin_Norman's results because the Take Profit distance is high and it's not spread- or time-sensitive.

See the recent posts in the thread titled "99% modeling quality when backtesting with MT4 build 225": http://indo-investasi.com/showthread.php/3294-99-modeling-quality-when-backtesting-with-MT4-build-225

Here are my backtest results using Dukascopy tick data and real Ask prices (variable spread).

Commission = 3 USD per round turn lot.

 

All parameter values are default except:

Scalper_UseMM=0

Scalper_UseAutoGMToffset=0

Scalper_ManualGMToffset=0

Scalper_StealthMode=0

Scalper_MaxSpread=3

WriteLog=0

PrintLogOnChart=0

 

Date range = 2007-04-02 to 2008-03-21:

http://stashbox.org/881204/FAPTurbo5%20RealAsk_Data%3DDukascopy%20ticks_Symbol%3DEURCHF_Commission%3D3USD_Scalper_UseMM%3D0_Scalper_UseAutoGMToffset%3D0_Scalper_ManualGMToffset%3D0_Scalper_StealthMode%3D0_Scalper_MaxSpread%3D3_2007-04-02to2008-03-21_ScreenCap.gif

 

Date range = 2008-03-25 to 2010-04-30:

http://stashbox.org/881206/FAPTurbo5%20RealAsk_Data%3DDukascopy%20ticks_Symbol%3DEURCHF_Commission%3D3USD_Scalper_UseMM%3D0_Scalper_UseAutoGMToffset%3D0_Scalper_ManualGMToffset%3D0_Scalper_StealthMode%3D0_Scalper_MaxSpread%3D3_2008-03-25to2010-04-30_ScreenCap.gif