TradeProgrammer^TM Manual

Copyright © Henwood Software Corp.

_{Last Update 9-2-08}

Table of Contents

• System Requirements
• Main Menu Guide
• Window and Dialog Guide
• Backtest Dialog
• Backtest Results Window
• Chart Window
• Connection Settings Dialog
• Database Settings Dialog
• Message Windows 1 and 2
• Options Dialog
• Orders Status Window
• Order Submit Dialog
• Positions Window
• Quote Edit Dialog
• Scan Results Window
• Scanner Dialog
• System Status Window
• Trade Statistics Window
• Activation
• Backtesting
• Charting
• Charting Tutorial
• Creating Custom Indicators
• Creating Systems
• Frequency
• TA-Lib Indicators
• Module chart reference
• Module tp reference
• Optimization
• Running Your Systems
• Scanning
• System Walkthrough
• OpenSSL and SSLeay License
• Technical Notes
• Proper User of range() Within Indicators

• Python 2.5
• Windows XP. Windows Vista not supported.
• 384 MB RAM minimum. 1 GB recommended.

Main Menu

• File
• Open Chart - Opens a chart file into a new chart window.
• Save View As - Saves position of all open windows to file. If you close the application when windows are open you don't need to save them, will be reloaded upon application start.
• Load View - Loads saved view file.
• Close - Close application.
• Command
• Connect - Connect to selected trade and data providers. To select a provider, go to Options->Connection Settings.
• Disconnect - Disconnect from trade and data provider.
• Start Select Systems - Starts only systems you select from systems directory.
• Start all Systems - Start all systems in systems directory.
• Stop Scripts - Signals all running systems to stop. (next call to tp.timeToQuit() returns true)
• Update Database - Downloads all required data to bring symbol files in data directory up to date. To change the starting date of stored data, go to Options->Database Settings.
• Market Open - Triggers market open processing as if market open time reached when in unattended mode.
• Market Close - Triggers market close processing as if market close time reached when in unattended mode.
• Tools
• Backtest - Opens Backtest Dialog
• Order - Opens Order Submit Dialog
• Chart - Opens a new Chart Window
• Scanner - Opens Scanner Dialog
• Debug - This menu sections contains tools for troubleshooting and debugging your systems
• Quote Edit - Opens Quote Edit Dialog
• View Data - Saves all historical data in memory to a text file and opens file for viewing.
• Save Messages - Saves all messages in Message Window to a text file.
• Start IB TWS - Automatically start TWS program, enters password and username, and logs in. You must have IB username and username entered in Login Settings Dialog for this to work.
• Unload Python - Development use only. Do not use.
• Account - Subscription account-related actions
• Activate - Activates the product, allowing it to run on your computer.
• Clear Oldest Activation - Removes last activation, freeing up one slot for activation on another machine.
• Clear All Activation - Removes all activations from account.
• Change Login - Change account password.
• Options
• Options - Opens the general Options Dialog
• Connection Settings - Opens Connection Settings Dialog
• Database Settings - Opens Database Settings Dialog
• Window
• System - Opens/Closes System Status Window
• Orders - Opens/Closes Order Status Window
• Backtest Results - Opens/Closes Backtest Results Window
• Scan Results - Opens/Closes Scan Results Window
• Messages 1 - Opens/Closes Messages 1 Window
• Messages 2 - Opens/Closes Messages 2 Window
• Trade Statistics - Opens/Closes Trade Statistics Window

Backtest Dialog

Backtest Results Window

You can sort any of the above fields by clicking the column label.

Chart Window

Connection Settings Dialog

Database Settings Dialog

Messages Windows

Right clicking on a message window, then selecting "clear all" will delete all messages in the window.

Options Dialog

If you do however require this feature there is currently one limitation. If you close the TradeProgrammer application it will lose track of which script each open order and position belongs to. This will be fixed soon.

** This depends on when your trading day begins. Some may only trade between 9:30 am to 4 pm, the standard hours. Others might want to include the full possible extended period, 7 am to 6 pm or more. You should set the open time to a few minutes before your actual trading day open when using unattended mode to give your trading and data software time to initialize.

Order Status Window

You can sort any of the columns by clicking the column label.

Order Submit Dialog

Positions Window

Quote Edit Dialog

Scan Results Window

You can sort any of the columns by clicking the column label.

Scanner Dialog

System Status Window

Trade Statistics Window

Activation and Authentication

Activation
The first time you use TradeProgrammer on a computer you will need to activate it. Open the Messages window if not already opened. You should see a message - "Not activated..." To activate, first enter your TradeProgrammer account login information in the Logins Settings Dialog by clicking Options->Login Settings from main menu. Then click from main menu Account->Activate. In you message window you should see "Activated." Now restart TradeProgrammer and it will be ready to use.

If the product was not activated you might receive one of these responses:

Product previously activated and ready to use.

Activation failed. Email not found or password incorrect. Check and try again.

No more computers allowed for your account. Try clearing oldest activation or purchase an additional subscription.

Authentication
Authentication is a process that occurs automatically every time TradeProgrammer is started. Upon success you will receive a "Authentication passed" message and "Authentication failed" upon failure.

Authentication works by collecting hardware information from your computer - the motherboard serial number, hard drive volume number (which changes when you reformat your hard drive), and network card MAC address. This information is sent to TradeProgrammer servers and checked against the information collected when the application was activated.

Switching Computers
If you decide you would like to use TradeProgrammer on a different computer than which it is currently activated for click from main menu Account->Clear Oldest Activation. Then install the TradeProgrammer software to the new computer and follow the first-time activation process.

Backtesting

You can backtest against one or more specified symbols or your entire base. You should close the Order, Position, and Trade Statistics Window before backtesting with a large database. The sheer number of orders generated (possibly thousands) will slow performance when these window are open.

At this time, using multiple symbols within the same system is not supported for backtesting, although it will work in live trading.

To start a backtest, click Tools->Backtest from the main menu. See the Backtest Dialog reference for information on setting the backtest parameters. When you are finished setting the parameters, click Ok and the backtest will begin. To view the backtest results, open the Backtest Results Window, from the main menu by clicking Window->Backtest Results.

You should read section Creating Systems to learn about the changes you must make to a system to make it ready for backtesting. Here is a summary:

• Place the following code within your system main loop:

  if(!tp.testStep())
  	break
  

• use tp.sleep() instead of time.sleep()
• If your system depends on the time, use tp.testing() to bypass. Time is not simulated in backtests.

If you want to see where the orders for a backtest show up first run the back test then open a chart, add a price indicator such as candlestick, then add the "orders" indicator. Only the orders for the last backtest symbol can be viewed like this. If you already have such a chart open, you will need to refresh it, by resizing it, or closing and reopening it, to see the orders.

Batch Backtesting

If you want to peform multiple backtests and don't have time to start each test you can use Batch Backtesting feature . First make a script in the /tradeprogrammer/scripts directory. Use the tp.backtest() function for each test to be run. For example the script below will perform 5 tests total, one for each of the past 5 days, using tick data (frequency -4), symbol XYZ, and then print a message when done. How is this different from simply using a start and end of 0 and 5 using the backtest dialog? The below example will generate 5 sets of backtest results, one for each day, and the backtest dialog would create only one set. Creating a different set for each day allows you to compare the individual days to eachother.

Testing individual days is not the only use of batch builds. You can run any combination and number of tests you need.

for i in range(0, 5):
	tp.backtest("c:/tradeprogrammer/system/mySystem.py", 'XYZ', -4, i, i)

print 'finished all testing'

Charting

Scanning

Unlike backtests, which display scan results in terms of figures, such as profit/loss, the scan results are whatever you like. For example, if you are scanning for stocks who's volume greater than 1 million and less than 2 million, and you then want the volume displayed as the result, it would look as follows:

volume = tp.volume(tp.getSym(), tp.getFreq(), 1)
if(volume > 1000000 and volume < 2000000):
	tp.scanResult(str(volume))

Scans use all the same functions a systems but they are much simpler and have no need for a loop. Two scans are never run at the same time in sperate threads and you cannot start a scan when systems are running. This eliminates the problem of variable name conflicts so unlike in systems you can use same-name variables in many different scan files without any special steps.

Optimization

Optimization of trading systems is the process of making small adjustments to the variables of that system with the goal of maximizing the system's profits. Optimization is not automatic with TradeProgrammer. You must manually program which variables will be adjusted and what the adjusted values will be. Normally you will adjust just one or a few variables. For example if you are optimizing a SMA system you might try SMA 9, 10, 11 and 12 to see which produces the best results.

Optimization is achieved through the setOpt() function. This function stores the specified value for the script from which it was called. That value is then automatically passed along with all calls to order(), isPos(), and setPos(). Two calls to any of these functions with the same symbol, but with a different stored optimization value can be considered a call with two completely different symbols. There will be a different set of orders, positions, and profit loss for each symbol/optimization combination.

You can use optimization when backtesting or testing in real-time mode. The result will be displayed in the Backtest Results Window when backtesting or in the Trade Statistics Window when testing in real time. If you want to eventually use the same system your are testing for live trading you will need to write some conditional code in the system, using the tp.isTesting() function, to bypass the optimization code.

Below is a copy of the system presented in System Walkthrough only now it will trade sma 10, 11, and 12 in addition to sma 9.

1:	import ind
2:	import tp
3:	
4:	def sma9sys():
5:		sym = ‘XYZ’
6:		periods = 9
7:		size = 100
8:		freq=0
9:
10:		if(tp.testing())
11:			opts = [9, 10, 11, 12]
12:		else
13:			opts = [9]
14:	
15:		while not tp.timeToQuit():
16:			for opt in opts:				
17:				tp.setOpt(str(opt)):
18:				periods = opt
19:				
20:				if not tp.isOrder(sym, 0):
21:					last = tp.last()
22:					sma = ind.sma(sym, freq, periods, 0) 
23:					if last > sma and not tp.isPos(sym, 1):
24:						tp.order(sym, size, 0)
25:					if last < sma and not tp.isPos(sym, -1):
26:						tp.order(sym, size*-1, 0)			
27:				if(!tp.testStep())
28:					break				
29:				tp.sleep(1000)

Creating Systems

import tp

while not tp.timeToQuit():
	#do system logic

while not tp.timeToQuit():
	#do system logic
	if(!tp.testStep())
		break

while not tp.timeToQuit():
	if not tp.isOrder(symbol, 0):
		#do system logic
		if(!tp.testStep())
			break

If you run many systems at the same time you run the risk of conflicts between same-name variables in different scripts. Each script is run in a different thread but threads share global memory. In python all variables which are not declared in functions or in another module are global. Unlike other languages, such as C, declaring a variable within a block creates a global variable rather than a variable who's scope is limited to that block.

Here's how it works in C/C++, which might be what you are used to:

int i = 1;
cout << i;

if(condition=true)
{
	int i=4;
	cout << i;	 
}

There is no conflict with the above code, thread 1's output is always 1, and thread 2's output is always 4. However in python it is different:

i = 1
print i

if(contrition=True)
	i = 4
	print i

def mySysFunc():
	...do you system logic here
	
mySysFunc() #make a call to your function

tp.bid(symbol) 
tp.ask(symbol)
tp.open(symbol, frequency, offset)
tp.high(symbol, frequency, offset)
tp.low(symbol, frequency, offset)
tp.close(symbol, frequency, offset)
tp.volume(symbol, frequency, offset)

sma(sym, freq, offset, periods)		# simple moving average
bband(sym, freq, offset, periods)	# bollinger bands

System Walkthrough

1:	import ind
2:	import tp
3:	
4:	def sma9sys():
5:		sym = ‘XYZ’
6:		periods = 9
7:		size = 100
8:		freq=0
9:	
10:		while not tp.timeToQuit():
11:			if not tp.isOrder(sym, 0):
12:				last = tp.last()
13:				sma = ind.sma(sym, freq, periods, 0) 
14:				if last > sma and not tp.isPos(sym, 1):
15					tp.order(sym, size, 0)
16:				if last < sma and not tp.isPos(sym, -1):
17:					tp.order(sym, size*-1, 0)
18:			if(!tp.testStep())
19:				break
20:			tp.sleep(1000)

Running Your Systems

Place all systems you would like to run in the systems directory. You can change this directory in the Options Dialog. Before running systems you must connect to a data and or trading provider. Once connected, choose main menu Command->Start Systems to start running all systems in the systems directory.

It is possible to run systems which use only a data provider or only a trading provider. If you only need one, data or trading, choose SIMULATION as the other provider.

Unattended Mode

Unattended mode is designed to allow hands-free automated trading. You can set unattended mode in the main Options dialog, by selecting Options->Options from the main menu.

In unattended mode you must set the market open and market close fields in the Options dialog. The ATS will then start your systems at the market open and stop them at the market close. It will also handle the initialization of your trading or data software if necessary.

Creating Indicators

indFunc(str sym, int freq, int periods, int offset, list params)

sma('XYZ', 0, 9, 1, [])

indCustom(str name, object plotFunc, object infoFunc)

Module chart

This module contains functions for custom-draw indicators.

Many functions have the parameter itemId of type int. This is the id of the indicator on the chart. When the indicator draw function is called this id is passed to it. You must then pass this id back with the functions listed below so the application knows for which indicator to perform the requested draw action.

• box
• color
• info
• indDefault
• indCustom
• lineto
• moveto

box(int itemId, double top, double left, double btm, double right, double, bool fill)

color(int itemId, int r, int g, int b)

info(int itemId, str infoText)

indCustom(str indName, PyObject plotFunc, PyObject infoFunc)

indDefault(str indName, PyObject indFunc)

lineto(int itemId, double x, double y)

moveto(int itemId, double x, double y)

Module tp

• avgPosPrice
• barDate
• barPos
• closePos
• getDataApiName
• getPeriods
• getOrders
• getFreq
• getSym
• hist
• isHist
• isOrder
• isPos
• order
• ordersAtPlace
• quote
• rpl
• scanResult
• setFreq
• setOpt
• setSym
• setTestOrders
• sleep
• startHist
• startHist
• text
• testing
• testStep
• timeToQuit
• unloadHist
• upl
• volume
• waitHist

barDate(str sym, int freq, int offset

backtest(str systemPath, str symbol, int frequency, int start, int end)

float avgPosPrice(str sym)

int barPos(str sym, int freq, int year, int month, int day, int hour, int min)

int closePos(str symbol)

str getDataApiName()

int getPeriods(str sym, int frequency)

list getOrders(sym)

int getFreq()

str getSym()

isHist(str symbol, int frequency)

bool isOrder(str symbol, int type)

Returns 1 if there is an open order for symbol, 0 if not. Type can be 1 to signify a long position, -1 for short position, or 0 for any (long or short) position.

Checking whether or not there is an open order before proceeding with system logic is important. Imagine you have a system which has just given a signal to go long. An order for 100 shares is submitted, but only 25 is filled. The signal then gives a short signal. Assuming the previous order has filled, an order to sell 200 shares is sent, so that after filling you will have a short position of 100. However, since only 25 shares of the long order were filled, you end up with a short position of 175 shares. Now, if the short position produce a loss, it will be almost twice as expected.

bool isPos(str symbol, int type)

order(str symbol, int size, float price)

list ordersAtPlace(str symbol, int frequency, int offset, bool test)

scanResult(str resultText)

setFreq(int frequency)

setOpt(str optvalue)

setSym(str symbol)

setTestOrders()

sleep(int seconds)

startHist(str symbol, int frequency, int periods, int offset)

text(str s)

bool testing()

t = time.localtime()
hour = t[3]
if(hour >= 10 or tp.testing())
	...

testStep()

timeToQuit()

unloadHist(str symbol, int freq)

volume(str symbol, int frequency, int offset)

waitHist(str symbol, int frequency, int periods, int tries, int intervals)

Technical note: Proper use of range() within indicators.

def sma(sym, freq, periods, offset, params):
	rval = 0
	for i in range(offset+periods-1, offset-1, -1):
		rval = rval + tp.close(sym, freq, i)
	rval = rval / periods
	return rval

range(offset+periods-1, offset, -1)

Using TA-Lib Indicators

Using TA-Lib indicators requires some familiarity with TA-Lib, see ta-lib.org

A call to a TA-Lib indicator takes the following form:
tp.taind(indicator_name, symbol, destination/exchange, frequency, offset, params)

All indicator names available can be seen here.

destination/exchange is not required when trading stocks and an empty string can be used. If you trade futures with IB howerver you will need to enter the exchange.

frequency is bar frequency. See section explaining Frequency.

params is all parameters for the indicator, seperated by space if more than one. For example simple moving average, which has a ta-lib name of SMA, has only one parameter, the number of periods.

The return value of an indicator is a list. Most indicators, such as SMA, only have one return value so it is a single item list. Here is a call to a 9 day simple moving average:
sma = (tp.taind('SMA', 'XYZ', "", 0, 1, '9'))[0]

The same call but in two lines for easier reading:
rval = tp.taind('SMA', 'XYZ, "", 0, 1, '9')
sma = rval[0]


Here is a call to the bollinger bands indicator, which has 4 parameters (inputs) produces 3 outputs:
rval = tp.taind('BBAND', 'XYZ', "", 0, 1, '5 2 2 0')
upperBand = rval[0]
midBand = rval[1]
lowerBand = rval[2]


The best way to learn about what parameters (inputs) an indicator has is to open a chart and then attempt to add an item. The Add Item dialog contains a list of all TA-indicators and upon selecting one all input information is displayed. Here it is for BBAND:

To learn about the return value (outputs) for a TA-Lib indicator add the item to the chart. The information for each output will be displayed on a seperate line in he top left the chart, allowing you to see how many outputs there are as well as what each output is for, such as upper or lower band:

Frequency

Copyright © 2007 Henwood Software Corp.
TradeProgrammer^TM is a trademark of Henwood Software Corp.