Estimated reading time: 14 min

PrescienTrader adds the following AFL functions to AmiBroker. Most users will only need to use the PrescienTrader() , PTPlotFLDs() and PTBacktest() functions.

If you just want to use PrescienTrader as a standalone tool to analyze charts, you can simply double-click the PrescienTrader AFL formula, to create a new PrescienTrader chart pane. You don’t need to read the rest of this article.

The rest of the functions are for advanced users who want to call our algorithms from custom AFL scriPTs, for trading system development.


PrescienTrader

PrescienTrader ()
  • This is the main function that displays a PrescienTrader chart pane , displays all the plots and generates reports in the Interpretation window.
  • The function accepts no parameters, because all parameters are set in the AmiBroker Parameters window.
  • Although it’s possible to plot PrescienTrader on an existing price chart pane, it contains numerous indicators and graphs that may cause too much clutter and may conflict with your existing indicators. Therefore, we recommend creating a separate pane, exclusively for PrescienTrader.
  • You must set the API Key in the Parameters window. If you don’t set the API Key, PrescienTrader will display an Invalid API Key message.
  • To learn how to use each of these parameters, please read the article, PrescienTrader Parameters.

PTPlotFLDs

PTPlotFLDs()
  • This function plots the Future Line of Demarcation (FLD) for each cycle peak identified by PrescienTrader.
  • The function accepts no parameters, because all parameters are set in the AmiBroker Parameters window.
  • Since the FLD is closely related to price and uses the same scale as price, we recommend plotting the FLDs on your price chart pane, NOT on the PrescienTrader pane.
  • This function uses PrescienTrader as its data source to generate the FLD plots. Therefore, it REQUIRES the PrescienTrader() function to be running in a separate chart pane.
  • You may set the following parameters in the Parameters window:
    • FLDs (1-10) – This allows you to select which FLD plots to display on the chart. Each cycle peak frequency has an associated FLD plot, but this parameter allows you to avoid chart clutter, by limiting the number of plots.
    • Sync with PrescienTrader – When this is enabled, the function will automatically run once per second, to fetch the latest data from PrescienTrader, ensuring the FLD plots stay synchronized with the cycles analysis generated by PrescienTrader. If you disable sync, you may have to manually refresh the chart each time.
  • The function plots past, present and future FLDs. However, to display the past FLD plots, you must enable the static indicators in PrescienTrader. This is because the cycle frequencies change with each new bar. Therefore to plot a past FLD, the function needs access to past cycle analyses, which require the static indicators. If the static indicators are disabled, you’ll only see FLD plots for the current bar and future bars. Keep in mind, the future FLD plots are only an estimate and may change as new bars are added to the chart.

PTBarAnalysis

PTBarAnalysis (Data Series, Polarity, PL Basis, Lookback Range, Min Frequency, Max Frequency, Harmonic Filter, Min Fitness, Best X Cycles, Analysis Bar, Plot Start Bar, Plot End Bar, Cache Index) 

PTBarAnalysis performs a point-in-time cycles analysis on a single bar. It then generates various plots within the specified plot range, going backward and optionally, forward in time. The backward plot can be used to visually verify the accuracy of the cycles analysis, by confirming its shape follows the corresponding price bars. The forward plot is used to predict future prices or to view the accuracy of its predictions on past bars. This is the function you’d use for live trading.

This function does not return any values directly. Instead, it populates a set of AFL variables, including scalars, arrays and matrices. These arrays can be plotted like any AFL array.

Parameters

  • Data Series (array)
    • This parameter expects an AmiBroker array, containing the actual data series you want to analyze. For example, if you want to analyze the closing prices, you’d pass the built-in AmiBroker Close (C) variable, which is an array containing the closing prices from your data series.
  • Polarity (integer)
    • 0 = Positive
    • 1 = Negative
  • PL Basis (integer)
    • 0 = Amplitude
    • 1 = Strength
  • Lookback Range (integer)
    • 1 – 15
  • Min Frequency (integer)
    • Typically 10 – 100
  • Max Frequency (integer)
    • Typically 150 – 300
  • Harmonic Filter (integer)
    • 0 – 100
  • Min Fitness (integer)
    • 0 – 100
  • Best X Cycles (integer)
    • 1 – 10
  • Analysis Bar (integer)
    • This is the bar index for which the analysis will be performed.
    • When performing an analysis for live trading, you’d set the analysis bar equal to the bar index of the last bar on the chart.
    • AmiBroker displays the first bar on the chart as bar number 1. However, the internal index used by AFL starts at bar 0. Therefore, if you want to analyze a specific bar, you need to subtract 1 from the bar number displayed in AmiBroker to get the correct bar number. If there are 5,000 bars on a chart, the last bar number would be 4,999, not 5,000.
    • You can specify a past bar index to perform a point-in-time analysis for that bar. For example, if you specified a bar index of 1,000, you could then compare the Prescient Line from bars 1,001 to the last bar on the chart, with the actual price bars for that bar range to see how accurate the forecast was at that point in time. When generating the cycle plots, the function ignores any price data to the right of the specified bar number.
  • Plot Start Bar (integer)
    • Specifies the bar index where the plot should start.
    • It’s usually not necessary to calculate plots for the entire series. For example, if your chart view displays 100 bars, you only need plots for those 100 bars, not the entire series.
  • Plot End Bar (integer)
    • Specifies the bar index where the plot should end.
    • You can project the plots into the future by specifying a Plot End Bar greater than the number of bars on the chart.
  • Cache Index (integer)
    • The function contains an internal caching mechanism intended to prevent sending multiple consecutive API requests for the same data. The cache stores the data series, parameters and results for each API request. If it receives the exact same data series and parameters twice in a row, it skips the API request and immediately returns the cached result.
    • If you’re just analyzing a single data series on a chart, you should set the Cache Index to 0.
    • If you’re analyzing multiple data series on a chart, you should specify a unique cache index for each data series. For example, if you’re analyzing the base period, plus two higher time periods, you’d specify Cache Index 0 for the base period, Cache Index 1 for HTP 1 and Cache Index 2 for HTP 2. This would create a separate cache for each time period.

Response

AFL Arrays:

Each element in these arrays represents a bar, starting with bar zero.

If you set Plot End Bar to a number greater than LastValue(BarIndex()), it will automatically shift the arrays backward by the difference. You can then project the plots into the future using the Plot function’s XShift parameter.

For example, suppose your chart contains 5,000 bars, meaning the last bar index is 4,999. If you set Plot End Bar to 5,009, the array would be shifted to the left by 10 bars. You’d then set XShift to 10, to plot 10 bars into the future. Since the array was shifted to the left by 10 bars, the XShifted plot will be correctly aligned with the chart.

  • PTPL – the Prescient Line plot
  • PTPLSlope – the slope of the Prescient Line. A positive value means it’s sloping upward and a negative value means it’s sloping downward.
  • PTTrendBar – the relative bar number within the current Prescient Line trend. The first bar number in each trend is bar 0 and all subsequent bar numbers are relative to the first bar of the trend.
  • PTTrendBars – the total number of bars in the current Prescient Line trend.
  • PTTrendPct – the percentage completed of the current Prescient Line trend. The percentage will never reach 100, because that would mean a new trend has started with a zero percent completion. So, 100% of the old trend is equivalent to 0% of the new trend.

AFL Matrices:

  • PTCycles – contains one row for each individual cycle frequency, in the range specified by the minFrequency and maxFrequency parameters. Each row contains eight columns, with the column numbers identified by AFL variables. When reading the values from the matrix, you should always reference the column numbers using their associated AFL variables, rather than hard coding the column numbers. That way, if the column numbers change in the future, your code will still work,
    • Frequency (PTColFrequency) – the cycle frequency.
    • Amplitude (PTColAmplitude) – the cycle amplitude.
    • Strength (PTColStrength) – the cycle strength, which is amplitude divided by frequency.
    • Start Bar (PTColStartBar) – the bar index where the cycle plot starts. This is used to set the correct phase for the cycle plot, but it’s generally not necessary, as the function provides the plots for you in the PTPlots matrix discussed below.
    • Fitness (PTColFitness) – a statistical measure of how reliable the cycle has been, from 0% to 100%.
    • Peak (PTColPeak) – indicates whether or not the cycle is a peak and if so, whether or not it was filtered by one of the filters specified in the parameter settings:
      • 0 = Not a peak
      • 1 = Valid peak
      • -1 = Peak filtered by Min Fitness filter
      • -2 = Peak filtered by Harmonic Filter
      • -3 = Peak filtered by Best X Cycles filter
    • Slope (PTColSlope) – indicates whether the cycle is sloping up or down. Slope will only be calculated on cycles that are valid peaks (Peak = 1):
      • 1 = Sloping up
      • -1 = Sloping down
    • FLD Price (PTColFLDPrice) – indicates the price of the cycle’s FLD. This will only be calculated on cycles that are valid peaks (Peak = 1).
  • PTPeaks – the PTPeaks matrix contains the same information in the same format as the PTCycles matrix, except it only includes cycles that are valid peaks (Peak = 1). The cycles will be sorted by frequency.
  • PTPlots – the PTPlots matrix contains the plots for each cycle that’s a valid peak (Peak = 1), sorted by cycle frequency. Each row represents a peak cycle frequency and each column represents a bar. As with the arrays, if you specify a Plot End Bar beyond the last bar on the chart, the matrix rows will be shifted to the left by the difference. You may use the MxGetBlock function to convert the matrix rows to AFL arrays.
  • PTFLDPrices – The PTFLDPrices matrix contains the FLD prices for each cycle that’s a valid peak (Peak = 1), sorted by cycle frequency. Each row represents a peak cycle frequency and each column represents a bar. This matrix will only contains FLD prices for the current bar and future bars. Use the PTRangeAnalysis function to obtain past FLD data. As with the arrays, if you specify a Plot End Bar beyond the last bar on the chart, the matrix rows will be shifted to the left by the difference. You may use the MxGetBlock function to convert the matrix rows to AFL arrays.

AFL Scalars

  • PTFLDScore – The FLD Score is a powerful indicator that can help to confirm or reject trades proposed by the Prescient Line and the other cycle indicators.
    • The FLD Score is calculated by multiplying the strength or amplitude (depending on the PL Basis setting) of each peak cycle frequency by either 1 or -1, depending on whether the series price is above or below the FLD price, and summing them all together.
    • If the FLD Score is positive, it indicates the market is in a cyclical uptrend.
    • If the FLD Score is negative, it indicates the market is in a cyclical downtrend.

PTRangeAnalysis

PTRangeAnalysis (Data Series, Polarity, PL Basis, Lookback Range, Min Frequency, Max Frequency, Harmonic Filter, Min Fitness, Best X Cycles, Start Bar, End Bar, Process Arrays)

PTRangeAnalysis analyzes an entire range of bars, returning the point-in-time values for each bar in the analyzed range. Calling PTRangeAnalysis for a range of 1,000 bars is equivalent to calling PTBarAnalysis 1,000 times, once for each bar in the range. This is the function to use for backtesting and optimization, NOT for live trading.

This function does not return any values directly. Instead, it populates a set of AFL arrays and matrices. These arrays can be plotted like any AFL array.

Parameters

  • Data Series (array)
    • This parameter expects an AmiBroker array, containing the actual data series you want to analyze. For example, if you want to analyze the closing prices, you’d pass the built-in AmiBroker Close (C) variable, which is an array containing the closing prices from your data series.
  • Polarity (integer or array
    • 0 = Positive
    • 1 = Negative
  • PL Basis (integer or array)
    • 0 = Amplitude
    • 1 = Strength
  • Lookback Range (integer or array)
    • 1 – 15
  • Min Frequency (integer or array)
    • Typically 10 – 100
  • Max Frequency (integer or array)
    • Typically 150 – 300
  • Harmonic Filter (integer or array)
    • 10 – 100
  • Min Fitness (integer or array)
    • 0 – 95
  • Best X Cycles (integer or array)
    • 1 – 10
  • Start / End Bar (integer)
    • The Start Bar and End Bar parameters specify the bar range to analyze. Unlike PTBarAnalysis, End Bar may NOT exceed the bar index of the last value on the chart. This function is used for backtesting, not live trading, thus it would make no sense to project the plots into the future.
  • Process Arrays (integer)
    • This parameter determines whether the function processes the parameters as integers or arrays. This only applies to parameters that can be either integers or arrays, which include Polarity, PL Basis, Lookback Range, Min Frequency, Max Frequency, Harmonic Filter, Min Fitness and Best X Cycles.
    • Normally, these parameters are integers, which means the same parameter setting would be applied to the entire range of analyzed bars. This is the default setting and most users should leave this alone. However, if you’re an advanced user, this gives you the incredibly powerful capability to vary the parameter settings on a bar-by-bar basis by passing the values as arrays.
    • If you do use arrays, you must set the Process Arrays parameter to 1, so PrescienTrader knows to process them as arrays. Enabling this feature greatly increases the file size of each API request, which consumes bandwidth and slows processing time, so only enable it if you know what you’re doing and have a good reason for doing it.
      • 0 = Off
      • 1 = On

Response

AFL Arrays

Each element in these arrays represents a bar, starting with bar zero.

  • PTStaticPL – the static Prescient Line. This contains the Prescient Line plot values calculated at the point-in-time of each bar in the specified range. This is very different from the standard Prescient Line analysis returned by PTBarAnalysis, which calculates the Prescient Line at a single point-in-time, then projects the plot backward and forward in time.
  • PTStaticPLSlope – the slope of the static Prescient Line. A positive value means it’s sloping upward and a negative value means it’s sloping downward.
  • PTStaticPLSlopeCum – the cumulative slope of the static Prescient Line. This is what you’d use to plot the static Prescient Line.
  • PTStaticTrendBarthe relative bar number within the current static Prescient Line trend. The first bar number in each trend is bar 0 and all subsequent bar numbers are relative to the first bar of the trend.
  • PTStaticTrendBars – the total number of bars in the current static Prescient Line trend.
  • PTStaticTrendPct – the percentage completed of the current static Prescient Line trend. The percentage will never reach 100, because that would mean a new trend has started with a zero percent completion. So, 100% of the old trend is equivalent to 0% of the new trend.
  • PTStaticFLDScore – see the description in the PTBarAnalysis section for an explanation of FLD Score.

AFL Matrices

These matrices return a list of values pertaining to the peak cycle frequencies of each bar. Each matrix represents a different indicator, but they’re all sorted by frequency. So, for example, row 3, column 999 corresponds to the same peak cycle frequency at bar 1,000, in all the matrices. The matrices contain the number of rows needed to list all the peak cycle frequencies for all bars.

  • PTStaticFrequencies – each row in this matrix corresponds to a peak cycle frequency.
  • PTStaticSlopes – each row in this matrix corresponds to the slope of the frequency found in the PTStaticFrequencies matrix, at the same row and column.
  • PTStaticFLDPrices – each row in this matrix corresponds to the FLD price of the frequency found in the PTStaticFrequencies matrix, at the same row and column.

PTBacktest

PTBacktest ()

This function is used to run a single-thread analysis for an exploration, backtest or optimization. It takes no parameters because you configure all the parameters in the Parameters window. When it runs, it calls PTRangeAnalysis and generates the AFL arrays and matrices described in the PTRangeAnalysis function.

Unlike the PrescienTrader function, which supports two higher time periods, the PTBacktest function supports unlimited higher time periods.

Read the Exploration, Backtesting and OPTimization article, for a detailed explanation of how to use this function.


PTBacktestMultiPrepare

PTBacktestMultiPrepare ()

This function is used to prepare a multi-threaded analysis for an exploration, backtest or optimization. It uses the parameters defined in the Parameters window to generate the data in the proper format for an API request.

Read the Exploration, Backtesting and OPTimization article, for a detailed explanation of how to use this function.


PTBacktestMultiExecute

PrescientAnalysisExecute ()

This function is used to execute a multi-threaded analysis for an exploration, backtest or optimization. It parses the data, outputs the results to the exploration columns and generates AFL arrays containing the results.

Read the Exploration, Backtesting and OPTimization article, for a detailed explanation of how to use this function.


PTLogToFile

PTLogToFile (Message)

You can use PTLogToFile to write messages to the log file defined in the backtest Parameters window.


PTWait

PTWait (Seconds)

You can use PTWait to suspend execution for the specified number of seconds. Normally, this is used when making API requests. If an API request fails due to a down Internet connection, you can use PTWait to wait a certain number of seconds before retrying.

PTPositionSize

PTPositionSize (Points Risked, Percent of Equity Risked, Account Base Currency)

This function automates Van Tharp-style, volatility-based position sizing in AmiBroker. We think this is the best approach to position sizing, because it automatically manages risk based on volatility and portfolio equity.

AmiBroker’s built-in SetPositionSize function includes an option to specify position size as a percent of equity, but it considers position size to be the margin amount required to open the position. The problem is, margin is only loosely associated with risk. While its true that riskier instruments generally have higher margin requirements, we think it’s best to calculate position size based on the exact risk of each trade, not by relying on your broker to manage your risk for you. Therefore, we need to use the cash value of the position, not the margin value. For instance, if you buy a futures contract and its price drops by 2%, you’ve lost 2% of the cash value of that contract, not 2% of your margin deposit!

This function sets the position size AmiBroker uses to execute trades; it does NOT return a value. It’s intended to replace the built-in SetPositionSize function.

  • The Points Risked argument should be an array, specifying the number of points you’re willing to risk on each trade. Typically, this would be the same number of points you’re using for your stop. If you’re not using stops, estimate the maximum number of points your trading system typically loses before it exits from or reverses a position. We recommend using a formula that considers the typical price range of the current instrument. For example, ATR(20) * 2 would specify a maximum risk of twice the Average True Range. It’s particularly important to use a formula, not a static value, when backtesting multiple instruments (portfolio backtesting), so Points Risked automatically adapts to the price range and volatility of each instrument.
  • Percent of Equity Risked is the percentage of your current equity you’re willing to risk on each trade. We normally use 2%, but you may use a higher or lower value, depending on your risk tolerance.
  • Account Base Currency is a string containing the base currency of your trading account. For example, “USD”.

The function uses the following fields from the AmiBroker Information window:

  • Margin Deposit
  • Point Value
  • Currency

Additionally, the function supports multi-currency position sizing, which is necessary when the base currency of an instrument differs from the base currency of your account. Most commonly, this happens with Forex pairs denominated in currencies other than US Dollars. To perform multi-currency position sizing, the function looks up the price of the Forex pair that’s a combination of your account’s base currency and the base currency of the Forex pair being backtested. For example, if you’re backtesting CHFJPY, the base currency would be JPY. If your account base currency is USD, the function would look up the price of USDJPY, to calculate the price of JPY in US Dollars, which then allows it to calculate the position size for CHFJPY in US Dollars. Therefore, the following data is required for multi-currency position sizing to work:

  1. You must enter the correct Point Value for each instrument you’re backtesting. Consult the AmiBroker documentation for an explanation of Point Value.
  2. You must enter the Margin Deposit for each instrument you’re backtesting. You can enter a fixed margin deposit as a positive number, or as a percentage of the contract’s cash value, by entering the percentage as a negative number. For example, if your broker offers 20:1 leverage, the margin deposit would be 5%. So you’d enter -5 as the margin deposit.
  3. For any instrument NOT denominated in your account base currency, you must enter the base currency in the Currency field. Using the example of CHFJPY, you’d enter JPY in the Currency field for that instrument.
  4. You must have quotes in your AmiBroker database for the ticker symbol XXXOOO or OOOXXX, where XXX represents your account base currency and OOO represents the instrument base currency. If your account base currency appears first in the symbol (XXXOOO), the function will automatically invert the price of the currency pair. Again, using the example of CHFJPY, if your account base currency is USD, you’d need to have quotes for USDJPY in your database. The symbol must be formatted exactly as XXXOOO or OOOXXXX, NOT XXX-OOO or some other variation. If your data provider uses a non-standard ticker symbol format, you can map the standard format to the non-standard format in your database by creating a variable named after the standard format and assigning the non-standard format as the value. Again, using the above example, if your database contained the symbol USD-JPY, you could map USDJPY to USD-JPY using the following code:
USDJPY = "USD-JPY";

Alternatively, you may set a fixed conversion value by setting the variable equal to a numerical value. For instance:

USDJPY = 108.68;
  • Setting a fixed conversion value overrides any values in your database. This is mostly useful when you don’t have quotes in your database for the conversion currency.
  • You must create variables for all your symbol maps before calling the PrescientPositionSize function.
  • The PrescientPositionSize function ignores AmiBroker’s built-in currency conversion settings. Thus, when using this function, you can disregard the settings in the Preferences->Currencies screen. We actually recommend that you disable AmiBroker’s dynamic currency conversion functions by removing the symbols from the Dynamic Rate Symbol column in this screen. This is because AmiBroker’s built-in dynamic currency conversion creates a problem when the conversion currency has a shorter quote history than the instrument being analyzed. In this instance, the position size will be set to zero for any dates outside the range of the conversion currency’s quote history. Removing the symbols from the Dynamic Rate Symbol column disables dynamic currency conversion, which fixes this problem. If your conversion currency’s quote history is shorter than the the quote history of the instrument being analyzed, the PresicentPositionSize function will use the price from the last bar of the conversion currency, for any bars outside the range of the conversion currency’s quote history.
Was this article helpful?
Dislike 0
Views: 1367
FOREVER FREE
Daily Trading Signals
by PrescientSignals
SUBSCRIBE
Subscribe to receive FREE trading signals delivered to your email inbox daily
Free signals are delayed by one-week. This allows you to evaluate our PrescientSignals service risk-free, by comparing the previous week's predictions with the actual results.
FOREVER FREE
Daily Trading Signals by PrescientSignals
SUBSCRIBE
Subscribe to receive FREE trading signals delivered to your email inbox daily
Free signals are delayed by one-week. This allows you to evaluate our PrescientSignals service risk-free, by comparing the previous week's predictions with the actual results.