"about the final result" - TWR(Terminal Wealth Relative).

For example, if the income was 12%, theTWR = 1.12.

Or, if the loss was 18%, theTWR = 0.82.

The probability of winning will be indicated byp.

The probability of a loss q = 1 - p

The total number of transactions -N, the number of winning deals -V, and Losing -L.

The size of the winning deals in absolute terms will be indicated by the symbola, Losing transactions - b, while the amount of gain to loss - a k = a / b.

If it will be the size of the winning \ Losing transactions in shares on the equity, the symbol will be, respectively, a%, b%. The ratio of a% b% will bek. I note that the margin rate on capital will be denoted asf.

The probability of events that we will receive on time payments, will be referred to as Prob.

p = 0.45,q= 0.55,a% = 0.08, b%= 0.05, N = 20, the results will be - MO = 0.0085, TWR = 1.170.

TWR =N * MO + 1

we introduce two additional formulas:

for the calculation of income at a known number of winning bids (3)

and calculate the probability of occurrence in the series a certain number of winning bids (4):

The results of our calculations according to formulas (3) and (4) are represented by green dots.

For example, the probability of occurrence of a series of transactions whose TWR = 1.04, is 0.162.

Maximum winnings TWR = 2.60, - Prob = 1.2E-07.

Blue dots indicate the same data, but in the form of cumulative probability.

Capital, which is involved in trade, is called aDeposit. We have the right, using the leverage Leverage, buy and sell contract sizeLotSize. The contract may be fractional, so we need the notion of Lots, a characteristic that points to the ways in which the size of contracts we operate. Accordingly, the actual size of the lot operated in the base currency will call LotPrice. For the right to buy or sell, we must make a Margin. If you express the Margin as a percentage of the Deposit, we'll get the option Margin%. Together with these we will use the notion of StopOut, which indicates a minimum of Margin, the achievement of which leads to the closing of this transaction, and forced the completion of trade. Thus, there are two different situations where continued trading impossible (with the desired amount of LotPrice), ie a situation of collapse. Additionally, was introduced another parameter, which is conventionally called Sigma. In essence, this is - the ratio used in commercial transactions with a view to a loan size of their own funds. The ratio of the operated facilities for the actual capital, ie analogue of leverage, just assigned to the entire Deposit, and not toLotPrice.

One of the basic concepts relating to the trade, it is Quote - current exchange rate. Minimal change of course is called Tick. The size of the minimum exchange tool - TickPrice, seen as part of Deposit - Tick%.

In addition, we have a few more parameters associated with the change of course: the so-called TP (TakeProfit) and SL (StopLoss) - change of course in which fixation occurs profit or loss. If you express these parameters in the currency, then we will call them TPprice and SLprice. And, accordingly, expressed as part of Deposit, and is called a TP% SL%. Also there are different kinds of dealer commissions, such as Spread, Swap, etc. We will not address all the diversity of these options, stop only at the Spread, which is traditionally expressed in paragraphs change course. Data Swap, if it is expressed in points, it is very easy to accommodate, if necessary. If the Swap is presented as a real bank interest in the use of leverage through borrowings, the situation is more complicated.

Let's try to compare the parameters obtained by us with those we used in the problem of "absorbing wall". It is clear thatSL% corresponds tob%. Like, TP% =a%. The initial capital we have been presented as TWR = 1. In this case, the same as our calculations are based on parameters, expressed in fractions of units. Used boundary value z% + b% can be considered Margin%. Between the two there is some difference, but it is not a principle, if you do not take into account the existence of StopOut. In general, it turned out that the challenges are similar to the first approximation. Later we check this assertion.

A careful look at the formula (9, 11, 13, 15), then all of a parameter Sigma. As already mentioned, this is the analogue of leverage, and it is used in all the important formulas. Directly from Leverage depends only Margin%.

The correctness of the calculations can be verified in the following example. We know that when Leverage = 100, LotSize = 100000, Lots = 0.1, Deposit = 10000 size of the collateral is 100. If we look at a timetable (brown), we find that the Margin% = 0.01, in the Deposit = 10000 it is 100.

Crude Oil Inventories (Forex Factory)

and the oil price

Files:Crude Oil Inventories

Files:Crude Oil Inventories

and

Crude Oil COT report

Files:COT Crude Oil

Files:A.) SVM input example

Train examples: i=1..100

Class: C[0+i]-O[6+i]

Attribute1: C[1+i]-O[6+i]

Attribute2: C[2+i]-O[6+i]

Attribute3: C[3+i]-O[6+i]

Attribute4: C[4+i]-O[6+i]

Attribute5: C[5+i]-O[6+i]

Test examples: i=0

Class: C[0+i]-O[6+i]

Attribute1: C[1+i]-O[6+i]

Attribute2: C[2+i]-O[6+i]

Attribute3: C[3+i]-O[6+i]

Attribute4: C[4+i]-O[6+i]

Attribute5: C[5+i]-O[6+i]

B.) SVM input example

Train examples: i=1..100

Class: C[0+i]-O[0+i]

Attribute1: C[1+i]-O[0+i]

Attribute2: C[2+i]-O[0+i]

Attribute3: C[3+i]-O[0+i]

Attribute4: C[4+i]-O[0+i]

Attribute5: C[5+i]-O[0+i]

Test examples: i=0

Class: C[0+i]-O[0+i]

Attribute1: C[1+i]-O[0+i]

Attribute2: C[2+i]-O[0+i]

Attribute3: C[3+i]-O[0+i]

Attribute4: C[4+i]-O[0+i]

Attribute5: C[5+i]-O[0+i]

Files:Offset=6

B.) SVM input example

Train examples: i=1..100

Class: C[0+i]-O[0+i]

Attribute1: C[1+i]-O[0+i]

Attribute2: C[2+i]-O[0+i]

Attribute3: C[3+i]-O[0+i]

Attribute4: C[4+i]-O[0+i]

Attribute5: C[5+i]-O[0+i]

Test examples: i=0

Class: C[0+i]-O[0+i]

Attribute1: C[1+i]-O[0+i]

Attribute2: C[2+i]-O[0+i]

Attribute3: C[3+i]-O[0+i]

Attribute4: C[4+i]-O[0+i]

Attribute5: C[5+i]-O[0+i]

Files:Simple Trading System

Source Checkout - wtx - Google Code

class Strategy:

def __init__(self, deposit=400000, price_per_point=200, commission=500):

self.bars = None

self.current_index = -1

self.points_balance = 0

self.entry_price = 0

self.current_contracts = 0

self.price_per_point = price_per_point

self.commission = commission

self.initial_deposit = deposit

self.balance = deposit

self.win = 0

self.lose = 0

def Buy(self, price):

self.ExitShort(price)

self.entry_price = price

self.current_contracts = 1

#print "Buy @%d"%(price)

def Sell(self, price):

self.ExitLong(price)

self.entry_price = price

self.current_contracts = -1

#print "Sell @%d"%(price)

def ExitLong(self, price):

if self.current_contracts > 0:

points_balance = (price - self.entry_price)*self.current_contracts

if points_balance>=0:

self.win += 1

else:

self.lose += 1

print "ExitLong: %d points (%d -> %d)"%(points_balance, self.entry_price, price)

self.points_balance += points_balance

self.balance += points_balance * self.price_per_point - self.commission

self.current_contracts = 0

def ExitShort(self, price):

if self.current_contracts < 0:

points_balance = (price - self.entry_price)*self.current_contracts

if points_balance<=0:

self.win += 1

else:

self.lose += 1

print "ExitShort: %d points (%d -> %d)"%(points_balance, self.entry_price, price)

self.points_balance += points_balance

self.balance += points_balance * self.price_per_point - self.commission

self.current_contracts = 0

def ShowResults(self):

self.ExitLong(self.Close(0))

self.ExitShort(self.Close(0))

transaction = self.lose+self.win

print "--------------- Performance Report ------------------"

print "%d Transaction, Win=%d (%.1f%%), Lose=%d(%.1f%%)"%(\

transaction, self.win, self.win*100/transaction, self.lose, self.lose*100/transaction)

print "Initial deposit=%d"%(self.initial_deposit)

print "points_balance is %d points, equals to %d (NTD)"%\

(self.points_balance, self.points_balance*self.price_per_point)

print "commission is %d"%(self.balance-self.initial_deposit-self.points_balance*self.price_per_point)

print "Final balance is %d"%(self.balance)

print "Actual earning is %d"%(self.balance-self.initial_deposit)

print "--------------- Performance Report ------------------"

def CrossOver(self, a, b):

#prev a current b

print "[%s] Close=%d, Fast[-1](%d)=Slow(%d) ? %d"%(self.bars[self.current_index].time, \

self.bars[self.current_index].close, a(1),b(1),a(0),b(0), a(1)

=b(0))return a(1)

=b(0)def Close(self, offset=0):

if self.current_index < offset:

return 0

else:

return self.bars[self.current_index - offset].close

def Open(self, offset=0):

if self.current_index < offset:

return 0

else:

return self.bars[self.current_index - offset].open

def High(self, offset=0):

if self.current_index < offset:

return 0

else:

return self.bars[self.current_index - offset].high

def Low(self, offset=0):

if self.current_index < offset:

return 0

else:

return self.bars[self.current_index - offset].low

def Sum(self, f, count, offset=0):

sum = 0

if self.current_index < offset:

return sum

elif self.current_index < offset + count:

count = self.current_index-offset

for e in map(f, range(offset, count+offset)):

#print "sum(%d) += %d"%(sum, e)

sum += e

#print "[%d] Sum(count=%d, offset=%d)=%d"%(self.current_index, count, offset, sum)

return sum

def Average(self, f, count, offset=0):

avg = 0

if self.current_index < offset:

return 0

elif self.current_index < offset + count:

count = self.current_index-offset

if count == 0:

return 0

avg = self.Sum(f, count, offset)/count

#print "[%d] Average(count=%d, offset=%d)=%d"%(self.current_index, count, offset, avg)

return avg

def Run(self, Open, High, Low, Close):

pass

def RunWithBars(self, bars):

self.bars = bars

self.count = len(bars)

for i in range(0, self.count):

self.current_index = i

#print "[%d][%s] %d %d %d %d"%(i, self.bars.time, self.bars.open, \

#self.bars.high, self.bars.low, self.bars.close)

self.Run(lambda x: self.Open(x), \

lambda x: self.High(x), \

lambda x: self.Low(x), \

lambda x: self.Close(x))

self.ShowResults()

Files:An attempt to beat the market by generating trading strategies using genetic programming

pipster - Google Code

Capital Management

On the management of capital

"about the final result" -

TWR(Terminal Wealth Relative).For example, if the income was 12%, the

TWR= 1.12.Or, if the loss was 18%, the

TWR= 0.82.The probability of winning will be indicated by

p.The probability of a loss

q = 1 - pThe total number of transactions -

N, the number of winning deals -V, and Losing -L.The size of the winning deals in absolute terms will be indicated by the symbol

a, Losing transactions -b, while the amount of gain to loss - ak = a / b.If it will be the size of the winning \ Losing transactions in shares on the equity, the symbol will be, respectively,

a%, b%. The ratio ofa% b%will bek. I note that the margin rate on capital will be denoted asf.The probability of events that we will receive on time payments, will be referred to as

Prob.p= 0.45,q= 0.55,a%= 0.08,b%= 0.05,N= 20, the results will be -MO= 0.0085,TWR= 1.170.TWR=N*MO+1we introduce two additional formulas:

for the calculation of income at a known number of winning bids (3)

and calculate the probability of occurrence in the series a certain number of winning bids (4):

The results of our calculations according to formulas (3) and (4) are represented by green dots.

For example, the probability of occurrence of a series of transactions whose TWR = 1.04, is 0.162.

Maximum winnings TWR = 2.60, - Prob = 1.2E-07.

Blue dots indicate the same data, but in the form of cumulative probability.

Files:Capital, which is involved in trade, is called a

Deposit. We have the right, using the leverageLeverage, buy and sell contract sizeLotSize. The contract may be fractional, so we need the notion ofLots, a characteristic that points to the ways in which the size of contracts we operate. Accordingly, the actual size of the lot operated in the base currency will callLotPrice. For the right to buy or sell, we must make aMargin. If you express theMarginas a percentage of theDeposit, we'll get the optionMargin%. Together with these we will use the notion ofStopOut, which indicates a minimum ofMargin, the achievement of which leads to the closing of this transaction, and forced the completion of trade. Thus, there are two different situations where continued trading impossible (with the desired amount ofLotPrice), ie a situation of collapse. Additionally, was introduced another parameter, which is conventionally calledSigma. In essence, this is - the ratio used in commercial transactions with a view to a loan size of their own funds. The ratio of the operated facilities for the actual capital, ie analogue of leverage, just assigned to the entireDeposit, and not toLotPrice.One of the basic concepts relating to the trade, it is Quote - current exchange rate. Minimal change of course is called

Tick. The size of the minimum exchange tool -TickPrice, seen as part of Deposit -Tick%.In addition, we have a few more parameters associated with the change of course: the so-called TP (TakeProfit) and SL (StopLoss) - change of course in which fixation occurs profit or loss. If you express these parameters in the currency, then we will call them

TPpriceandSLprice. And, accordingly, expressed as part ofDeposit, and is called aTP% SL%. Also there are different kinds of dealer commissions, such asSpread, Swap, etc. We will not address all the diversity of these options, stop only at theSpread, which is traditionally expressed in paragraphs change course. Data Swap, if it is expressed in points, it is very easy to accommodate, if necessary. If the Swap is presented as a real bank interest in the use of leverage through borrowings, the situation is more complicated.Let's try to compare the parameters obtained by us with those we used in the problem of "absorbing wall". It is clear that

SL%corresponds tob%. Like,TP%=a%. The initial capital we have been presented asTWR= 1. In this case, the same as our calculations are based on parameters, expressed in fractions of units. Used boundary valuez% + b%can be consideredMargin%. Between the two there is some difference, but it is not a principle, if you do not take into account the existence ofStopOut. In general, it turned out that the challenges are similar to the first approximation. Later we check this assertion.A careful look at the formula (9, 11, 13, 15), then all of a parameter

Sigma. As already mentioned, this is the analogue of leverage, and it is used in all the important formulas. Directly fromLeveragedepends onlyMargin%.The correctness of the calculations can be verified in the following example. We know that when Leverage = 100, LotSize = 100000, Lots = 0.1, Deposit = 10000 size of the collateral is 100. If we look at a timetable (brown), we find that the Margin% = 0.01, in the Deposit = 10000 it is 100.

Files: