Cfrtjryk

I'm in the process of creating a forex trading algorithm and wanted to try my shot at calculating EMA (Exponential Moving Averages). My results appear to be correct (compared to the calculations I did by hand) so I believe the following method works, but just wanted to get an extra set of eyes to makes sure i'm not missing anything.

Note that this just returns the EMA for the latest price, it doesn't return an array of EMA's as that isn't what I need for my application.

I am using this link as a reference: Exponential Moving Average

class Indicators:



    def sma(self, data, window):

        """

        Calculates Simple Moving Average

        http://fxtrade.oanda.com/learn/forex-indicators/simple-moving-average

        """

        if len(data) < window:

            return None

        return sum(data[-window:]) / float(window)



    def ema(self, data, window, position=None, previous_ema=None):

        """

        Calculates Exponential Moving Average

        http://fxtrade.oanda.com/learn/forex-indicators/exponential-moving-average

        """

        if len(data) < window + 2:

            return None

        c = 2 / float(window + 1)

        if not previous_ema:

            return self.ema(data, window, window, self.sma(data[-window*2 + 1:-window + 1], window))

        else:

            current_ema = (c * data[-position]) + ((1 - c) * previous_ema)

            if position > 0:

                return self.ema(data, window, position - 1, current_ema)

            return previous_ema





# Sample close prices for GBP_USD currency pair on the 2 hour timeframe

close_prices = [1.682555, 1.682545, 1.682535, 1.682655, 1.682455, 1.682685, 1.68205, 1.683245, 1.68405, 1.68401, 1.68506, 1.685825, 1.685955, 1.686595, 1.686325, 1.686375, 1.68701, 1.684995, 1.687245, 1.686135, 1.686205, 1.68724, 1.68753, 1.687775, 1.688245, 1.687745, 1.68699, 1.687285, 1.686325, 1.686295, 1.683945, 1.683035, 1.68401, 1.68327, 1.685185, 1.684755, 1.685265, 1.685325, 1.68625, 1.685645, 1.684355, 1.68387, 1.68413, 1.68416, 1.683425, 1.68481, 1.683245, 1.683645, 1.68325, 1.682745, 1.680385, 1.680655, 1.680875, 1.679995, 1.680445, 1.68064, 1.67937, 1.677735, 1.67769, 1.67777, 1.677525, 1.677435, 1.67766, 1.677835, 1.678005, 1.67823, 1.67902, 1.678605, 1.678425, 1.67876, 1.678555, 1.678505, 1.679085, 1.678755, 1.678125, 1.677495, 1.67677, 1.676205, 1.67716, 1.67741, 1.677135, 1.679295, 1.68054, 1.68143, 1.68115, 1.68111, 1.68055, 1.680495, 1.680565, 1.681375, 1.68244, 1.673395, 1.670885, 1.67156, 1.669525, 1.66906, 1.66903, 1.668935, 1.668805, 1.667895, 1.667905, 1.668485, 1.666345, 1.66832, 1.668005, 1.668615, 1.669305, 1.668415, 1.66891, 1.66843, 1.66855, 1.66834, 1.668725, 1.66952, 1.668075, 1.66859, 1.669, 1.669685, 1.668575, 1.66909, 1.66957, 1.669375, 1.671655, 1.67186, 1.67244, 1.6729, 1.672965, 1.673405, 1.67284, 1.67256, 1.67216, 1.67193, 1.673265, 1.67295, 1.672705, 1.67224, 1.67221, 1.67222, 1.67254, 1.670105, 1.66501, 1.663845, 1.66201, 1.661935, 1.661725, 1.66189, 1.661605, 1.661925, 1.66215, 1.66049, 1.660185, 1.66233, 1.66374, 1.66491, 1.665195, 1.663225, 1.66267, 1.65927, 1.659415, 1.65998, 1.6583, 1.656825, 1.65741, 1.659025, 1.658355, 1.659355, 1.65871, 1.65887, 1.658595, 1.65768, 1.657965, 1.657855, 1.657415, 1.658125, 1.65816, 1.659125, 1.658245, 1.65773, 1.658585, 1.65732, 1.657825, 1.65731, 1.65725, 1.65433, 1.654875, 1.65508, 1.656205, 1.656185, 1.6567, 1.658865, 1.658805, 1.65879, 1.6584, 1.65806, 1.658145, 1.65706, 1.656925, 1.65885, 1.65917, 1.659, 1.65794, 1.65797, 1.65711, 1.658675, 1.656915, 1.65474, 1.65455, 1.654135, 1.65467, 1.65473, 1.65543, 1.65465, 1.65721, 1.65717, 1.65927, 1.65895, 1.65724, 1.65812, 1.657435, 1.657395, 1.65755, 1.65975, 1.65983, 1.658975, 1.658855, 1.65814, 1.65838, 1.65797, 1.65785, 1.657795, 1.658915, 1.65888, 1.65888, 1.65869, 1.65851, 1.658195, 1.659985, 1.65933, 1.65842, 1.65836, 1.658435, 1.657605, 1.660225, 1.65991, 1.65908, 1.659065, 1.659605, 1.659555, 1.660535, 1.663025, 1.662295, 1.661525, 1.662735, 1.661335, 1.660895, 1.660905, 1.66093, 1.661425, 1.65934, 1.658235, 1.658305, 1.657035, 1.652785, 1.653185, 1.65176, 1.650105, 1.648505, 1.64713, 1.646975, 1.646815, 1.646575, 1.645355, 1.646425, 1.646365, 1.648295, 1.646245, 1.646305, 1.645075, 1.644875, 1.646035, 1.64602, 1.646025, 1.645615, 1.646135, 1.645585, 1.645695, 1.646195, 1.642865, 1.64237, 1.634805, 1.634575, 1.634475, 1.631665, 1.629265, 1.631115, 1.63094, 1.631775, 1.632175, 1.631775, 1.629345, 1.632785, 1.631155, 1.631765, 1.632865, 1.6327, 1.618735, 1.621365, 1.622655, 1.620755, 1.617995, 1.616985, 1.611595, 1.61411, 1.615785, 1.613975, 1.611155, 1.610865, 1.60935, 1.609255, 1.610085, 1.607585, 1.608405, 1.610095, 1.611495, 1.610465, 1.609775, 1.608715, 1.608615, 1.612435, 1.610495, 1.612275, 1.612555, 1.611785, 1.612515, 1.612945, 1.609495, 1.612515, 1.616155, 1.613295, 1.618215, 1.621225, 1.62018, 1.619885, 1.619565, 1.620435, 1.619375, 1.624325, 1.625165, 1.625185, 1.621845, 1.622345, 1.623795, 1.621875, 1.627455, 1.624845, 1.623875, 1.623625, 1.623295, 1.625575, 1.626125, 1.622445, 1.622145, 1.624155, 1.626055, 1.625755, 1.62671, 1.627055, 1.625875, 1.625055, 1.623925, 1.624645, 1.625215, 1.624725, 1.624025, 1.624515, 1.624205, 1.623755, 1.623325, 1.62273, 1.622535, 1.6242, 1.623045, 1.62169, 1.618415, 1.618185, 1.619605, 1.621425, 1.627035, 1.628145, 1.62778, 1.6271, 1.626485, 1.626335, 1.627615, 1.627965, 1.63094, 1.630125, 1.632065, 1.633775, 1.632895, 1.63064, 1.627885, 1.625845, 1.62667, 1.626805, 1.626695, 1.631185, 1.629635, 1.63067, 1.63367, 1.63908, 1.63709, 1.637255, 1.63738, 1.64403, 1.642545, 1.650745, 1.65183, 1.64764, 1.646825, 1.639945, 1.634085, 1.633615, 1.631255, 1.63123, 1.62993, 1.628745, 1.629105, 1.63096, 1.63417, 1.635245, 1.634745, 1.633755, 1.63316, 1.633325, 1.63464, 1.63394, 1.635555, 1.636435, 1.636235, 1.63692, 1.638125, 1.63869, 1.637795, 1.6323, 1.638925, 1.640955, 1.63767, 1.63686, 1.636575, 1.63977, 1.63909, 1.63945, 1.64001, 1.641005, 1.63986, 1.63838, 1.64039, 1.64047, 1.636, 1.63434, 1.634115, 1.633895, 1.633725, 1.63255, 1.633225, 1.63228, 1.632915, 1.63046, 1.630275, 1.628565, 1.63377, 1.631165, 1.630405, 1.63149, 1.63178, 1.63308, 1.63234, 1.630675, 1.630235, 1.63027, 1.632255, 1.630505, 1.626665, 1.625325, 1.624565, 1.624355, 1.62497, 1.62389, 1.62394, 1.62399, 1.622855, 1.621865, 1.62358, 1.62292, 1.623685, 1.624135, 1.62672, 1.624515, 1.624305, 1.624215, 1.62416, 1.623665, 1.6259, 1.625805, 1.626625, 1.62005, 1.618425, 1.62162, 1.62192, 1.620865, 1.62121, 1.621525, 1.621475, 1.619475, 1.619145, 1.619835, 1.620235, 1.6204, 1.618875, 1.622535, 1.62144, 1.617695, 1.61798, 1.61831, 1.618825, 1.61982, 1.62336, 1.621535, 1.61987, 1.616985, 1.6134, 1.61441, 1.6139, 1.61428, 1.61376, 1.61498, 1.615715, 1.612955, 1.61323, 1.61406, 1.6102, 1.606695, 1.60757, 1.59774, 1.59611, 1.597425, 1.597505, 1.59687, 1.59683, 1.596235, 1.59762, 1.59792, 1.59878, 1.596685, 1.598745, 1.59928, 1.60067, 1.602755, 1.603465, 1.607645, 1.608225, 1.60736, 1.60442, 1.604255, 1.60657, 1.60907, 1.604735, 1.607615, 1.61128, 1.607135, 1.60798, 1.60935, 1.60968, 1.60865, 1.607105, 1.60607, 1.606545, 1.60638, 1.607575, 1.60701, 1.60822, 1.606605, 1.604175, 1.617025, 1.615945, 1.616205, 1.61726, 1.61868, 1.618035, 1.62082, 1.620575, 1.62089, 1.61883, 1.61219, 1.61243, 1.61167, 1.61194, 1.61212, 1.61281, 1.61193, 1.61268, 1.606455, 1.60555, 1.60459, 1.60322, 1.604705, 1.60562, 1.606145, 1.6077, 1.60683, 1.60916, 1.611945, 1.61187, 1.611335, 1.60832, 1.609145, 1.60955, 1.608575, 1.60676, 1.606755, 1.60695, 1.607395, 1.606405, 1.6076, 1.606815, 1.60695, 1.604905, 1.59545, 1.59164, 1.59162, 1.592925, 1.59173, 1.590465, 1.590475, 1.588995, 1.58925, 1.590845, 1.590575, 1.589605, 1.59287, 1.59246, 1.597345, 1.596035, 1.591425, 1.59756, 1.60024, 1.59879, 1.600055, 1.598305, 1.597, 1.59925, 1.596045, 1.598845, 1.600635, 1.606405, 1.60702, 1.609275, 1.607365, 1.609575, 1.60851, 1.60739, 1.607985, 1.60689, 1.60864, 1.61119, 1.606205, 1.60851, 1.61039, 1.6088, 1.609185, 1.609595, 1.609035, 1.609775, 1.61074, 1.61063, 1.61041, 1.612855, 1.612635, 1.61363, 1.613635, 1.61695, 1.61705, 1.615905, 1.615515, 1.61577, 1.617205, 1.618045, 1.616225, 1.61466, 1.61568, 1.61528, 1.613335, 1.613045, 1.611435, 1.61178, 1.611265, 1.612395, 1.612615, 1.61215, 1.607975, 1.604285, 1.60507, 1.60358, 1.606845, 1.606225, 1.605045, 1.60427, 1.60436, 1.604135, 1.60491, 1.60554, 1.603425, 1.60145, 1.602715, 1.602035, 1.603575, 1.60334, 1.602125, 1.602895, 1.602555, 1.60353, 1.603785, 1.60398, 1.603185, 1.60395, 1.605205, 1.608145, 1.6097, 1.608285, 1.60858, 1.609015, 1.608575, 1.609035, 1.61034, 1.61067, 1.61045, 1.610075, 1.609925, 1.609565, 1.61126, 1.61328, 1.612295, 1.61265, 1.611675, 1.61242, 1.61272, 1.61275, 1.61212, 1.612105, 1.610675, 1.611365, 1.617255, 1.61567, 1.613815, 1.61384, 1.613175, 1.61411, 1.6132, 1.613675, 1.61394, 1.613675, 1.612405, 1.61159, 1.61244, 1.6149, 1.609405, 1.600625, 1.60129, 1.600285, 1.597765, 1.59804, 1.597085, 1.59792, 1.598775, 1.598545, 1.60051, 1.602205, 1.599575, 1.599565, 1.600345, 1.59987, 1.599305, 1.599525, 1.597605, 1.599295, 1.59902, 1.600385, 1.59634, 1.59984, 1.599365, 1.599665, 1.59966, 1.597265, 1.593855, 1.59653, 1.59713, 1.59792, 1.59974, 1.60036, 1.599825, 1.598095, 1.598495, 1.59798, 1.597485, 1.59773, 1.597355, 1.5986, 1.599495, 1.599755, 1.60003, 1.600025, 1.600375, 1.60105, 1.598955, 1.600155, 1.599765, 1.600475, 1.60022, 1.6006, 1.60181, 1.596045, 1.5943, 1.588815, 1.59068, 1.596245, 1.59832, 1.59755, 1.59771, 1.59605, 1.595625, 1.59563, 1.597925, 1.599085, 1.59813, 1.594745, 1.593165, 1.592695, 1.586095, 1.58439, 1.583355, 1.583495, 1.58396, 1.58395, 1.58188, 1.58351, 1.58259, 1.583445, 1.582, 1.58423, 1.584275, 1.58594, 1.58744, 1.58719, 1.588185, 1.58738, 1.589525, 1.590055, 1.59015, 1.588425, 1.590905, 1.589435, 1.587295, 1.585705, 1.585945, 1.584915, 1.584655, 1.585055, 1.585295, 1.58395, 1.58466, 1.584475, 1.58468, 1.585585, 1.586555, 1.588415, 1.59241, 1.591835, 1.591695, 1.590885, 1.591405, 1.590985, 1.591665, 1.592275, 1.5882, 1.581655, 1.580375, 1.58148, 1.57864, 1.578555, 1.57667, 1.577125, 1.577305, 1.57743, 1.577365, 1.577185, 1.57641, 1.574255, 1.57483, 1.57164, 1.570785, 1.57102, 1.5706, 1.568675, 1.567595, 1.56684, 1.56692, 1.56813, 1.567345, 1.565315, 1.560175, 1.565545, 1.568455, 1.567155, 1.566805, 1.566615, 1.567495, 1.57258, 1.572635, 1.571035, 1.56638, 1.56362, 1.564205, 1.56323, 1.564425, 1.56413, 1.564065, 1.56356, 1.56443, 1.565565, 1.565335, 1.565155, 1.56566, 1.565865, 1.564555, 1.564785, 1.564695, 1.56344, 1.5631, 1.56226, 1.561195, 1.56147, 1.560665, 1.562395, 1.56057, 1.56928, 1.566655, 1.56624, 1.566875, 1.56932, 1.56767, 1.56817, 1.567015, 1.567355, 1.56741, 1.56635, 1.565175, 1.566865, 1.570025, 1.57282, 1.56816, 1.570325, 1.56959, 1.56924, 1.56901, 1.570075, 1.569705, 1.56823, 1.56393, 1.56667, 1.56727, 1.56499, 1.56707, 1.564855, 1.566205, 1.56555, 1.564845, 1.565205, 1.56587, 1.56643, 1.56677, 1.564145, 1.56529, 1.56839, 1.568565, 1.569955, 1.569735, 1.570485, 1.57035, 1.569595, 1.568, 1.567995, 1.568395, 1.56889, 1.567615, 1.56646, 1.57027, 1.57135, 1.57154]

• Recursion is a good tool for the right job, but here it is used to accomplish simple looping. As such the code...
  
  
  
  • is more difficult to read and reason about.
  
  
  • is slower because much of the code in ema only needs to run once.
  
  
  • will fail with large enough value of window due to overflowing Python's call stack.
  
  
  
  


• Please document at least the parameters of each function, eg. that window is the length of the window, and that position counts backwards from the end of data. (In fact things would be clearer if position were a normal forward index into data)


• Raise an exception when you find a parameter has an invalid value. Returning None instead will only cause a more confusing exception later. In fact, if I try Indicators().ema(close_prices, 600) I get infinite recursion because sma returns None, which makes ema call sma over and over again.


• The previous point also reveals that if len(data) < window + 2 is not the right validity check.


• The + 1 in data[-window*2 + 1:-window + 1] don't seem correct to me. I suppose you want data[-window*2:-window]


• The statement return previous_ema is in an odd place because at that point you have calculated a new current_ema. This is the base case of the recursion, and it is customary to handle the base case first.

My proposal for ema:

def ema(self, data, window):

    if len(data) < 2 * window:

        raise ValueError("data is too short")

    c = 2.0 / (window + 1)

    current_ema = self.sma(data[-window*2:-window], window)

    for value in data[-window:]:

        current_ema = (c * value) + ((1 - c) * current_ema)

    return current_ema

Here's a short implementation:
https://pythonprogramming.net/advanced-matplotlib-graphing-charting-tutorial/

def ExpMovingAverage(values, window):

    """ Numpy implementation of EMA

    """

    weights = np.exp(np.linspace(-1., 0., window))

    weights /= weights.sum()

    a =  np.convolve(values, weights, mode='full')[:len(values)]

    a[:window] = a[window]

    return a

Note that it uses Numpy. I performed some tests and its answers match talib - that's another good option if you want to (and can) leverage C libraries for speed.

Pretty shallow review :

You don't need to write a class for what you are doing (and I suggest you have a look at this video). Your class does not encapsulate any data and you just use it to have your functions in a the same entity. I guess things would easier to understand if you were to define classmethod to make it obvious that you won't really rely on any instance whatsoever. However, an even better option would be to just define functions in a indicator module.