Who does JAPAN LIGHTS shine on? - page 16

New comment
 
What do the numbers and colours on the candles mean?
 
:), candle code
 

Red - large

Blue-medium

Green-small

and the rest are doji,



want to know the distribution of the candles on the chart?


 
Kos >> :

want to know the distribution of the candles on the chart?

yes, maybe you can modify your indices a bit http://forextools.com.ua/analyse/barstat.html and http://forextools.com.ua/analyse/barstatline.html

 
Sure....Very interesting....
 

Sample size: 5000

instrument: GBPUSD

Period: H1


Average candle size: 229 (on 5 digits)


large candlesticks (>=337): 21.3%

middle candles (337> 229>115): 39.6%

small candlesticks(<115): 37.2%

doji: 1.8%

 

Now we need to make the second step: look at the frequency distributions of candlestick pairs (and maybe even triples). If the samples show something similar to the histogram of normal distribution - then we can check: at this time after the bar that is drawn at Time[0] we most often get such-and-such, so we have to do such-and-such...

 

Dug out my swatcher, which I exhibited at ATC'07. For analysis and trading I used only one previous daily bar (GBPUSD), 1-2 trades every day.

I decided to test it without reoptimization till now (i.e. I've left all parameters of September 2007 without changes, except for 5-digit switch).

And of course I removed reinvestment. Frankly speaking, I thought it would fail with the old parameters, the market has changed very much.

The test results are rather modest (The Ministry of Defense and modelling quality leaves much to be desired...) but they do make me think about using candlesticks in real trading.

Yesterday
Alpari-Classic (Build 225)


Symbol GBPUSD (Great Britain Pound vs US Dollar)
Period Day (D1) 2006.01.02 00:00 - 2009.10.09 00:00 (2006.01.01 - 2009.10.12)
Model All ticks (most accurate method based on all smallest available timeframes)
Parameters Expert_Properties="-------Expert-Properties-------"; Expert_Id=3773; StartLot=0.1; StartBalance=10000; BalanceStep=1000; LotStep=0; lot_max=100; maxstop=600; x=0; y=400; z=1100; z1=9999; a=150; b=100; c=150; N=300; j=300; s=50; g=0; tpw1=1100; w1=500; w12=500; w13=200; tpq2=1000; q2=300; q22=300; q23=100; tpw2=500; w2=400; w22=200; w23=200; tpq3=1500; q3=100; q32=100; q33=1000; tpw3=800; w3=400; w32=200; w33=400; tpq4=700; q4=100; q42=200; q43=500; tpw4=1000; w4=200; w42=200; w43=300; Trade_Properties="--------Trade-Properties-------"; Slippage=5; PauseBeforeTrade=5; MaxWaitingTime=300; OrderBuyColor=Lime; OrderSellColor=Red; Allow_Flags="-----------Allow-Flags---------"; Allow_Info=true; Allow_LogFile=true; Allow_TradeLogFile=true; Allow_ErrorMail=true; Allow_ErrorLogFile=true; Info_Properties="--------Info-Properties--------"; EnglishInfo=false; Font_Size_Variant=4; Standart_Color=White; Warning_Color=Magenta; Price_Up_Color=Lime; Price_Down_Color=Red;
Bars in history 1981 Modelled ticks 16607889 Modeling quality n/a
Chart mismatch errors 355
Initial deposit 10000.00
Net profit 13212.24 Total profit 42415.68 Total loss -29203.44
Profitability 1.45 Expectation of winning 8.45
Absolute drawdown 31.96 Maximum drawdown 973.86 (5.50%) Relative drawdown 5.50% (973.86)
Total trades 1563 Short positions (% win) 771 (58.88%) Long positions (% win) 792 (56.44%)
Profitable trades (% of all) 901 (57.65%) Loss trades (% of all) 662 (42.35%)
Largest profitable trade 150.00 losing transaction -60.00
Average profitable deal 47.08 losing trade -44.11
Maximum number continuous wins (profit) 11 (576.52) Continuous losses (loss) 7 (-364.60)
Maximum Continuous Profit (number of wins) 576.52 (11) Continuous loss (number of losses) -364.60 (7)
Average continuous winnings 2 Continuous loss 1
Graph

 
Kos >> :

Sample size: 5000

Symbol: GBPUSD

Period: H1


Average Candle Size: 229 (on 5 digits)


large candlesticks (>=337): 21.3%

middle candles (337> 229>115): 39.6%

small candlesticks(<115): 37.2%

doji: 1.8%

We are also working on a similar idea, we calculate the average difference between the low and the high of the last quarter on M30 and if the difference between the low and the high of the analyzed bar is more than 2 times, ......

But so far the results are worse... If you use a statically entered value, it doesn't matter if it is 35 points or 50 points.

Maybe there is a problem with the period for determining the average bar, maybe a quarter is too long and a couple of weeks is enough??? Or maybe on the contrary, we should take a year? Or maybe there is just another error in the code or algorithm, in short, we need to figure it out... We're just at the beginning of a long journey...

 
RomanS >> :

We are also working on a similar idea, we calculate the average difference between Khai and Low for the last quarter on M30 and if for example the difference between Khai and Low of the analyzed bar is more than 2 times, then ......

But so far the results are worse... If you use a statically entered value, it doesn't really matter if it is 35 points or 50 points.

Perhaps, there is a problem with the period of determining the average bar, maybe a quarter is too long and a couple of weeks are enough, or maybe we should take a year, or maybe we just have another error in the code or algorithm, so we must decide... we are just at the beginning of a long journey...

It is not just the body of the candle that needs to be analysed, but also the shadows

1...9101112131415161718
New comment