Killer 'B' System Q & A

Thanks Thom, One swing of the hammer & the nail is set.

I 100% agree that a good Backtest on "Session Start" would add to this particular Strategy. {I have only ran it Forward and manually tried a few variations of the Inputs.}

Mike, Great idea to see if this Model would function as well on a MWR.

Mods, I'm kinda thinking I would like to leave the thread here for the time being as it will probably stay more focused. Do you mind?

http://www.forexfactory.com/showthre...#post152610 2

Hi guys, I try to use Bemac's method. But i can't seems to calculate the MDR. Can anybody explain how to ?

Interesting … intuitively this concept appears to be solid and logical.

The calculations are straight forward (although well beyond OandA’s platform – ouch) and easy to conceptualize.

The $3.50 question would be … can this concept be transposed to a different trading session. Unfortunately, I can only get screen time during the Asian session for about 2-3 hours at most (10p-1a EST).

The more I mull this … the more I feel unless this can be automated … I’m S.O.L.

Any thoughts?

Good approach. 2 comments:
1)If have have this right, you are estimating total pip movement (in math terms the total variation) of a longer time set (like dailies) using shorter time sets (like minutes), by approximating the actual tick movements using the shorter time sets. It's a really useful number; for the short time sets the best approximation to the total pip movement is:
2*(High-Low) - abs_value(Open-Close) (I don't know what language you are working in, so I just left it generic). You can sumn those up (say) for every minute to get an approximation of the total pips travelled over an hour (or day or week). Note that in the original post the red "-" sign was a plus.. Sorry. This formula gives the minimum expected travel taking into accound different open & closes)

2)In something I worked on in the past (actually based on Point & Figure charting and statistically inverting them to get time expectations), I also found it useful not only to consider just the previous larger time frame, but to consider it in terms of it's position in the day (or week or month). The math for that is pretty hairy, so I won't go into it; but just that you may get better results if you consider in the long run treating certain time periods - like around regular releases, the London & New York opens, NY etc. as special cases, and hence tending to have their own distributions.

Don't know if I'm confusing the issue or not. I hope not! I'll keep reading, I like your approach.

OK, i did some searching . . . if one does a search over at Forex-TSD for "Dolly" there is an indy by cja that shows the daily/session open and then go out X pips to a suggested entry, then beyond that X pips to a suggested TP level. Maybe some of the colaborators over there might want to chime in here if there are any? I think the indy from CJA might be able to be modified for this purpose so i'll give it a go this week.

: ) Thom

I have 268 days of data for USD/JPY

I calculated MDR based on ...
min range + ((max range - min range)/2) = 171 pips

Based on this 268 days of daily range data (H-L) ... a 171 pip range only occurs ~3% of the time. Also, only 11% of the days have TOTAL range to hip TP level (171 * 80% = 136).

I must be calculating MDR wrong ... or ... I'm grasping the concept incorrectly.

FYI ... the pure median of the 268 (H-L) ranges = 75 pips.

Is this a more realistic figure ... or should I calculate MDR as stated but only concern myself with the previous day or two ? Thanks

Hi Rustbelt,
I understand what you are saying & have noticed this before.

I haven't found a way to code it in VTT {lacking required function} but how I would prefer to calculate MDR would be something like this in XL.

Instead of this;
min range + ((max range - min range)/2)

I would prefer;
Small(Range,2) + ((Large(Range,2) - Small(Range,2)) / 2)

This would remove any Large/Tiny Spike anomaly from the equation.

Hey Thom {Accrete} Are you getting anywhere?

Just about ready to request the thread be moved to the 'Sytems' Forum.

Hi Bemac,
Should I be able to see the graphic arrows?
I do see the indicator@ the bottom of the chart.
I did enable graphics, restarted but they still do not appear.
Thanks, Matt

Hey Matt,
Yes you should see the Graphics in full including the Signals.

Remember you will need a minimum of 3 days of data on the chart for it to calculate.

I use 2,500 bars on a 5Min chart.

Options Tab > Properties > General > General Properties >
Communication Tab > (# of bars)

..

In another life time I created a whole system trading statistical ranges with the setups being price at the extreme ends and then I'd trade back to the median. More often than not, the median would be consolidation, and I could hold on until price reach the other extreme. Usually. The nature of that move is nothing new. 1-2-3, Elliott Wave something or the other, flags, pennants, etc. However, using statistical ranges and a mean put hard line probable entry and exit areas on the chart. That is why I love this concept.

Figuring out the range of the bar is another crucial step to that system I used. I usuallly kept that indicator on the daily chart so that I could see how much of the Average Daily Range was getting filled. That is a concept I borrowed from a guy named Aayat who had a killer system heavily based on that concept alone.

Anywho, Bemac, I think I see what you're doing and I like how you're doing it. BTW, I made an indicator for VT back in 2003 or so to measure the average range. This is not the same as the ATR. What this indicator does is highlights the average high, average low, and median of x amount of bars and plot those as a sort of benchmark. Then there are oscillator bars that that rise as time fills up. Red if short, Green if long. Then you can gauge if the period is done yet. IE if the daily bar is green and only at the median line, there's a good chance there's still room left. I used to base my trailing stops off this indicator. You can find it here: http://www.visualtradingcharts.com/f...ht=daily+range

For you VT and MT4 users, it's pretty easy to put that concept into Excel via the DDE and have numbers instead of gauging off of an indicator. It's hard for me to believe that the concept never really caught on. For day traders, I think knowing what the average pip number was and how much of that was filled would be crucial.

I don't have the system anymore. I accidently deleted it all. Nice, eh? But I did find some screenshots. Here is the Average Range and fill thing I was talking about that is done in Excel. It tracks x daily periods and from there, I came up with Conservative, Moderate, and Aggressive pip range targets. Then as price filled up for the day, I could easily track if/when one of those targets was hit. If the conservative was hit, it was time to raise stops to at least lock in the conservative amount of pips.

Screenshot here: http://img69.imageshack.us/my.php?image=udx0ad.gif


I don't recall what that split stuff meant. I think it was used to determine direction. Anywho, I used to trade looking at that thing alone. No charts. Another part that isn't shown there showed pivot points, fibs, etc. All numbers. No charts. So when I knew when to buy/sell and based off of the range, an general idea on when to get out. Later on I added charts to show visually what the numbers were saying.

Here's a screenshot of the system with charts. For the most part, I mainly paid attention to the numbers. The key numbers were the deviations from the mean. When price was near one of those lines, it was time to enter. By using multiple time frames, I was able to get some pretty precise entries and exits. The premise being when price was far from the mean on multiple frames, price will return to at least the mean. Many times the mean on the daily chart was the other end of the range on the 1h or 4h chart. So knowing that, it was easy to sit on a trade until the daily mean was hit. Wait out some retrace or add on, and wait for price to hit the other side of the range on the daily using the smaller time frames as precise timing.

Ok, I'm hoping this is concept you're doing with this Killer B system. If not, ignore these posts completely.

Screenshot of the stastical mean numbers and deviations:
http://img440.imageshack.us/img440/4...s0104a17hf.gif

It's a lot of BS to look at, but the focus is the mean and the +-1 and +-2 deviations.

Those channel lines are impossible in VT, but using what you have is a good proxy and the concept is still the same.

btw. here's a small insight into who's talkin'
JR97's old and longtime avatar

Thanks Bill,
I did not have the correct # of bars.
Matt

Hi Bill, and gang. I'm no further in my understanding/doodling than i was last week , ( but what JR97 shared was a real whack to the side of the head. I have yet to find any VT/MQ4 coder that can take your idea over from VT into MT4.

I must say, and i know this is a bit off topic, that JR's mention of trading with pure numbers is very interesting!

Still pushing upward.
Thom

The best trader I have ever met was this hippy looking guy who came down from the mountains or something to this local 4x meeting I happened to attend about 5 years ago. He turned 4k into 60k in 3 months or something doing nothing but trading the bid/ask. He kept a note book and would keep track of what numbers kept showing up as the high/low of every time frame from 5 minutes up to the weeks and months. Doing that, he scalped every nook and cranny and basically beat the crap out of FXCM to the point where they called him a cheater and put him on manual execution. I don't get the calling him a cheater, but I can understand putting him on manual. Anywho... I was so completely blown away what this guy did just watching the bid/ask and writing down numbers in his little notebook. That was one of the major inspirations for going chartless and focusing on just the numbers.

As of the last few years, I've pulled a complete 180° turn. Now everything is about the chart and "time". I don't even care about price anymore. I just watch for time and enter and wait for time to run out and then exit.

Absolutely fascinating stuff! Is that realy true? Just a notebook? It must have taken him ages! I guess those reccurring numbers are pretty accurate levels of support resistance and the idea would be to scalp around those levels with some big leverage. Id best get some more paper.....

He didn't use big leverage. Just a ton of trades when all was said and done. The one that really stuck out to me that I think got lost on everybody else in the room was that he was also a very good craps and blackjack player. Good enough to almost make a living at it. The part that stuck out was that this guy was not afraid to lose a trade or a hand of blackjack. He said keep'em small and get on with the next trade. Stay with the game plan because you'll come out ahead in the end if you stick with it. Take care of risk blah blah blah. Everybody else in the room was too busy asking about the money they thought they could make doing it his way. These were almost all 4x Made Easy users btw.

Coming back to the topic in hand, I have a question regarding the MDR.

The calculation is stated as follows, ive inserted the questions:

Code Logic.

Step 1. At the end of Each Session, Subtract the Low from the High. eg. {H-L} [Todays Range]

This bits pretty obvious

Step 2. Add to the Lowest result of Step1, the (Highest result of Step 1 Minus the Lowest Result of Step 1) Divided by 2. eg. {Min + ((Max-Min)/2)} [Median Day Range]

If we've just calculated the range from yesterdays H/L , then there is only one result. When you say the lowest and highest result of step 1, do you mean the highest and lowest daily range observed previously? This is what I think you mean. Please correct me if im wrong. If this is correct, how far back do we look for the extremes of the range, a year? Would it be better to plot a normal distribution of the daily range and use the top and bottom 15% of the distribution as the high and low to remove even the effect of outliers from the MDR calculation?

Hope I'm not being retarded and have missed something obvious to everyone else.