Files
QuanTAlib/Tests/Validations/EMA_Test.cs
T
Miha Kralj 58694a9600 refactoring
2022-11-26 22:04:26 -08:00

167 lines
5.1 KiB
C#

using System;
using QuanTAlib;
using Skender.Stock.Indicators;
using Tulip;
using Python.Runtime;
using Python.Included;
using TALib;
using Validations;
using Xunit;
namespace One.by.one;
public class Ema : IDisposable
{
private readonly GBM_Feed bars;
private readonly Random rnd = new();
private readonly int period, skip;
private readonly double precision;
private readonly IEnumerable<Quote> quotes;
private readonly double[] outdata;
private readonly double[] inopen;
private readonly double[] inhigh;
private readonly double[] inlow;
private readonly double[] inclose;
private readonly double[] involume;
private readonly string OStype;
private readonly dynamic np;
private readonly dynamic ta;
private readonly dynamic df;
public void Dispose() {
PythonEngine.Shutdown();
GC.SuppressFinalize(this);
}
public Ema()
{
bars = new(Bars: 1000, Volatility: 0.5, Drift: 0.0);
period = rnd.Next(30) + 5;
skip = period-1;
precision = 1e-8;
quotes = bars.Select(q => new Quote
{
Date = q.t,
Open = (decimal)q.o,
High = (decimal)q.h,
Low = (decimal)q.l,
Close = (decimal)q.c,
Volume = (decimal)q.v
});
outdata = new double[bars.Count];
inopen = bars.Open.v.ToArray();
inhigh = bars.High.v.ToArray();
inlow = bars.Low.v.ToArray();
inclose = bars.Close.v.ToArray();
involume = bars.Volume.v.ToArray();
// Checking the host OS and setting PythonDLL accordingly
OStype = Environment.OSVersion.ToString();
if (OStype == "Unix 13.1.0")
OStype = @"/usr/local/Cellar/python@3.10/3.10.8/Frameworks/Python.framework/Versions/3.10/lib/libpython3.10.dylib";
else
OStype = Path.GetFullPath(".") + @"\python-3.10.0-embed-amd64\python310.dll";
Installer.InstallPath = Path.GetFullPath(path: ".");
Installer.SetupPython().Wait();
Installer.TryInstallPip();
Installer.PipInstallModule(module_name: "pandas-ta");
Runtime.PythonDLL = OStype;
PythonEngine.Initialize();
np = Py.Import(name: "numpy");
ta = Py.Import(name: "pandas_ta");
string[] cols = { "open", "high", "low", "close", "volume" };
double[,] ary = new double[bars.Count, 5];
for (int i = 0; i < bars.Count; i++) {
ary[i, 0] = bars.Open[i].v;
ary[i, 1] = bars.High[i].v;
ary[i, 2] = bars.Low[i].v;
ary[i, 3] = bars.Close[i].v;
ary[i, 4] = bars.Volume[i].v;
}
df = ta.DataFrame(data: np.array(ary), index: np.array(bars.Close.t), columns: np.array(cols));
}
[Fact]
public void WeirdData() {
EMA_Series QL = new(source: bars.Close, period);
var lastData = bars.Close.Last();
var lastCalc = QL.Last();
QL.Add((DateTime.Today.AddDays(1), double.NaN), update: true);
Assert.NotEqual(lastCalc, QL.Last()); //value changed
QL.Add(lastData, update: true);
Assert.Equal(lastCalc, QL.Last()); // back to the same data
QL.Add((DateTime.Today.AddDays(-1), double.NegativeInfinity), update: true);
Assert.NotEqual(lastCalc, QL.Last()); //value changed
QL.Add(lastData, update: true);
Assert.Equal(lastCalc, QL.Last()); // back to the same data
QL.Add((new DateTime(), double.Epsilon), update: true);
Assert.NotEqual(lastCalc, QL.Last()); //value changed
QL.Add(lastData, update: true);
Assert.Equal(lastCalc, QL.Last()); // back to the same data
}
[Fact]
public void Updating() {
EMA_Series QL = new(source: bars.Close, period);
var lastData = bars.Close.Last();
var lastCalc = QL.Last();
int lastLen = QL.Count;
QL.Add((DateTime.Today, 0), update: true);
Assert.NotEqual(lastCalc, QL.Last()); //value changed
QL.Add(lastData, update: true);
Assert.Equal(lastLen, QL.Count); // same size
Assert.Equal(lastCalc, QL.Last()); // same data
}
[Fact]
public void Skender_Test()
{
EMA_Series QL = new(bars.Close, period, false);
var SK = quotes.GetEma(period).Select(i => i.Ema.Null2NaN()!);
for (int i = QL.Length; i > skip; i--)
{
double QL_item = QL[i - 1].v;
double SK_item = SK.ElementAt(i - 1);
Assert.InRange(SK_item! - QL_item, -precision, precision);
}
}
[Fact]
public void TALIB_Test()
{
EMA_Series QL = new(bars.Close, period, false);
Core.Ema(inclose, 0, bars.Count - 1, outdata, out int outBegIdx, out _, period);
for (int i = QL.Length - 1; i > skip; i--)
{
double QL_item = QL[i].v;
double TA_item = outdata[i - outBegIdx];
Assert.InRange(TA_item! - QL_item, -precision, precision);
}
}
[Fact]
public void Tulip_Test() {
double[][] arrin = { inclose };
double[][] arrout = { outdata };
EMA_Series QL = new(bars.Close, period: period, useNaN: false, useSMA: false);
Tulip.Indicators.ema.Run(inputs: arrin, options: new double[] { period }, outputs: arrout);
for (int i = QL.Length - 1; i > skip; i--) {
double QL_item = QL[i].v;
double TU_item = arrout[0][i];
Assert.InRange(TU_item! - QL_item, -precision, precision);
}
}
[Fact]
void PandasTA_Test() {
EMA_Series QL = new(bars.Close, period, false);
var pta = df.ta.ema(close: df.close, length: period);
for (int i = QL.Length; i > skip; i--) {
double QL_item = QL[i - 1].v;
double PanTA_item = (double)pta[i - 1];
Assert.InRange(PanTA_item! - QL_item, -1e-5, 1e-5);
}
}
}