Hi Guys

I love the idea of combining statistics and probability with my trading strategies so came up with the following idea;

Considering a single timeframe and looking at how each individual candle is formed we will see that each candle starts with the opening price (previous closing price) then make both a high and a low and eventually settle with the closing price.

This means that the majority of candles have some leg room between the opening price and it’s high and low. These wicks are usually referred to as the candle ‘shadow’

So the basic idea here is to try and leverage this phenomenon by entering a trade at the opening price of a candle and scalping some points for a profit.

To do this we need to first establish the statistical advantage or probability of the trade to succeed.

This means we need to evaluate the current market condition by looking at some of the previous candles to see which direction to trade as well as how many points to target.

We also need to take into consideration the spread and whether statistically speaking there are enough points to make the trade worthwhile.

So here is what I have so far:

Defparam cumulateorders = False

//Timeframe: I am thinking that timeframes in the range of 5min to 1Day would work best
//Market Hour Filter

If (hour < 9 or (hour = 9 and minute <= 0)) or (hour > 17 or (hour = 17 and minute >= 0)) Then
possize = 1 //Use 0 to enable filter
If longonmarket Then
Sell at market
ElsIf shortonmarket Then
Exitshort at market
EndIf
Else
possize = 1
EndIf

//variables

Spread = 1 //Enter spread here (For pips use format 0.0001)
SpreadVar = 4 // Spread variance factor is used to establish whether there is enough volatility taking into account the market spread (optimize for values between 3-7)
ShadowVar = 0.9 // Shadow variance factor is used to account for error in shadow length calculation (optimize for values between 0.8-0.9)
RiskReward = 3 // Risk to reward ratio used to calculate stop loss (optimize for values between 1-3)
StdDev = 1.2 // Standard deviation used to filter outliers (optimize for values between 1-2)
EvalPeriod = 6 // Evaluation period that will be used to make estimations (optimize for values between 3-9)

//Initialize Variables

FollowHighGreenA = 0
FollowLowGreenA = 0
FollowHighRedA = 0
FollowLowRedA = 0
FGA = 0
FRA = 0

//Establish the length of upper and lower shadows for candles following both red and green candles in the evaluation period

For i = 1 to EvalPeriod Do
If close[i+1] > open[i+1] Then
FollowHighGreenA = high[i] - open[i]
FollowLowGreenA = open[i] - low[i]
FGA = FGA + 1
ElsIf close[i+1] < open[i+1] Then
FollowHighRedA = high[i] - open[i]
FollowLowRedA = open[i] - low[i]
FRA = FRA + 1
EndIf
Next

//Calculate the raw average values of the shadows

FollowHighGreenB = FollowHighGreenA/FGA
FollowLowGreenB = FollowLowGreenA/FGA
FollowHighRedB = FollowHighRedA/FRA
FollowLowRedB = FollowLowRedA/FRA

//Test each shadow again similar to above but filter(cut) outliers against STD devaition

For ii = 1 to EvalPeriod Do
If close[ii+1] > open[ii+1] Then
If high[ii] - open[ii] <= (StdDev*STD(FollowHighGreenB)) Then
FollowHighGreenC = high[ii] - open[ii]
Else
FollowHighGreenC = (StdDev*STD(FollowHighGreenB))
EndIf
If open[ii] - low[ii] <= (StdDev*STD(FollowLowGreenB)) Then
FollowLowGreenC = open[ii] - low[ii]
Else
FollowLowGreenC = (StdDev*STD(FollowLowGreenB))
EndIf
ElsIf close[ii+1] < open[ii+1] Then
If high[ii] - open[ii] <= (StdDev*STD(FollowHighRedB)) Then
FollowHighRedC = high[ii] - open[ii]
Else
FollowHighRedC = (StdDev*STD(FollowHighRedB))
EndIf
If open[ii] - low[ii] <= (StdDev*STD(FollowLowRedB)) Then
FollowLowRedC = open[ii] - low[ii]
Else
FollowLowRedC = (StdDev*STD(FollowLowRedB))
EndIf
EndIf
Next

//Re-calculate the average values of the shadows

FollowHighGreen = FollowHighGreenC/FGA
FollowLowGreen = FollowLowGreenC/FGA
FollowHighRed = FollowHighRedC/FRA
FollowLowRed = FollowLowRedC/FRA

//Validate entry conditions and set targets and stops

If close > open and FollowHighGreen > FollowLowGreen and ((high - open)*ShadowVar) > (spread*SpreadVar) and ((FollowHighGreen*ShadowVar)/RiskReward) > (spread * SpreadVar) Then
Buy possize contract at market
Set target pprofit FollowHighGreen*ShadowVar
Set stop ploss (FollowHighGreen*ShadowVar)/RiskReward
ElsIf close > open and FollowHighGreen < FollowLowGreen and ((open - low)*ShadowVar) > (spread*SpreadVar) and ((FollowLowGreen*ShadowVar)/RiskReward) > (spread * SpreadVar) Then
Sellshort possize contract at market
Set target pprofit FollowLowGreen*ShadowVar
Set stop ploss (FollowLowGreen*ShadowVar)/RiskReward
ElsIf close < open and FollowHighRed > FollowLowRed and ((high - open)*ShadowVar) > (spread*SpreadVar) and ((FollowHighRed*ShadowVar)/RiskReward) > (spread * Spreadvar) Then
Buy possize contract at market
Set target pprofit FollowHighRed*ShadowVar
Set stop ploss (FollowHighRed*ShadowVar)/RiskReward
ElsIf close < open and FollowHighRed < FollowLowRed and ((open - low)*ShadowVar) > (spread*SpreadVar) and ((FollowLowRed*ShadowVar)/RiskReward) > (spread * spreadvar) Then
Sellshort possize contract at market
Set target pprofit FollowLowRed*ShadowVar
Set stop ploss (FollowLowRed*ShadowVar)/RiskReward
EndIf

Improvements would include:

Better ways to sample past candles and arrive at a more accurate ‘average’

Additional filters to enhance probability i.e. volatility filter