condition set by median

src/auxiliary_functions.py

@@ -255,7 +255,6 @@ def find_spread(mid_price,a_PNLTICK,a_TICKSIZE, b_PNLTICK,b_TICKSIZE):
         try:
             spread[i+1] = (((a_series[i+1] - a_series[i])*atick) - ((b_series[i+1] - b_series[i])*btick)) +spread[i]    
         except:
-            print("haata")
             pass
     #return pd.np.trim_zeros(pd.Series(data=spread,index=a_series.index,name='spread'))
     return pd.Series(data=spread,index=a_series.index,name='spread')

src/condition.py

@@ -6,6 +6,7 @@
 
 import pandas as pd
 from datetime import time
+import src.daily_data as dt
 
 # In[ ]:
 
@@ -43,7 +44,7 @@ def get_conditon(df_line, min_duration=None, min_amplitude=None):
         else:
             return False
     else:
-        if df_line['duration'] <= min_duration and abs(df_line['amplitude']) <= min_amplitude:
+        if df_line['duration'] <= min_duration and abs(df_line['amplitude']) <= abs(min_amplitude):
             return True
         else:
             return False
@@ -52,7 +53,7 @@ def get_conditon(df_line, min_duration=None, min_amplitude=None):
 # In[3]:
 
 
-def _conditionally_scan(df,min_dur,min_amp=None):
+def _conditionally_scan(df,medyan):
     """Bir dataframe'i verilen koşullara göre yeniden düzenler.
     Parameters:
         df (dataframe): pd.Dataframe nesnesi
@@ -67,6 +68,8 @@ def _conditionally_scan(df,min_dur,min_amp=None):
         koşulu sağlayıp sağlamadığını kontrol eder. Sağlanıyorsa aynı işlem devam eder. Sağlanmıyorsa veriler kaydedilerek
         bir sonraki dalga için işlemler tekrar yapılır ve diğer dalga hesaplanır.
     """
+    min_dur = None
+    min_amp=None
     df=df[['duration','amplitude']].dropna()
     index = 0
     cout = 1
@@ -78,18 +81,19 @@ def _conditionally_scan(df,min_dur,min_amp=None):
     first = True
     second = True
     while(cout<len(df)):
-        current_dur = df.iloc[index,:]['duration']
+        current_dur = df.iloc[index,:]['duration']        
         current_amp = df.iloc[index,:]['amplitude']
-        current_sign = find_sign(current_amp)       
+        current_sign = find_sign(current_amp)
+
+        if current_sign==1:
+            min_dur = medyan['pozitive'].duration
+            min_amp = medyan['pozitive'].amplitude
+        else:
+            min_dur = medyan['negative'].duration
+            min_amp = medyan['negative'].amplitude
 
         second = True
         while(second and cout<len(df)+1):
-            if(current_sign==0):
-                # dur_list.append(df.iloc[index,:]['duration'])
-                # index = cout                        
-                # cout=cout+1
-                # second = False
-                continue
             try:
                 next_sign = find_sign(df.iloc[cout,:]['amplitude'])
 
@@ -100,11 +104,14 @@ def _conditionally_scan(df,min_dur,min_amp=None):
 
                 else:
                     if get_conditon(df.iloc[cout,:],min_dur,min_amp):
+
                         dur_list.append(df.iloc[cout,:]['duration'])
                         amp_list.append(df.iloc[cout,:]['amplitude'])
                         cout+=1
 
                     else:      
+                        print("count: ", cout," -index: ",index)
+                        print("d",min_dur,"a",min_amp)
                         dur_list.append(df.iloc[index,:]['duration'])
                         amp_list.append(df.iloc[index,:]['amplitude'])
                         index = cout
@@ -121,7 +128,7 @@ def _conditionally_scan(df,min_dur,min_amp=None):
                 dlist.append(sum(dur_list))
                 alist.append(sum(amp_list))
                 time_list.append(df.index[cout-1])
-                second = False            
+                second = False
 
     new_df = pd.DataFrame({'duration':dlist,'amplitude':alist},index=time_list)
     return new_df
@@ -141,7 +148,8 @@ def scan(df):
     main_index = df.index.levels[0].tolist()    
     for index in main_index:
         current_df = df.loc[index]
-        df_list.append(_conditionally_scan(df=current_df,min_dur=current_df.duration.median()))
+        current_medyan = dt.divide(current_df)
+        df_list.append(_conditionally_scan(df=current_df,medyan=current_medyan))
 
     combin_df = pd.concat(df_list,keys=main_index)
     return combin_df

src/daily_data.py

@@ -23,21 +23,22 @@ class DailyData:
     time: List[pd.Series] = field(default_factory=list)
     spread: List[pd.Series] = field(default_factory=list)
     change: List[pd.Series] = field(default_factory=list)
-    duration: List[pd.Series] = field(default_factory=list)
-    amplitude: List[pd.Series] = field(default_factory=list)
+    duramp: List[pd.DataFrame] = field(default_factory=list)
+    # duration: List[pd.Series] = field(default_factory=list)
+    # amplitude: List[pd.Series] = field(default_factory=list)
     df: pd.DataFrame = field(default=pd.DataFrame)
 
     def spread_hesapla(self):
         for i in range(23):
             self.time.append(self.mid_price_list[i]['time'])
             self.spread.append(aux.find_spread((self.mid_price_list[i].iloc[:,1],self.mid_price_list[i].iloc[:,2]),a_PNLTICK=10,a_TICKSIZE=0.0001,b_PNLTICK=6.25,b_TICKSIZE=0.0001))
             self.change.append(aux.find_change(self.spread[i]))
-            self.duration.append(aux.find_duration(self.change[i]))
-            self.amplitude.append(aux.find_amplitude(self.change[i],self.duration[i]))
+            self.duramp.append(aux.find_duramp(self.change[i]))
+        #    self.amplitude.append(aux.find_amplitude(self.change[i],self.duration[i]))
 
 
     def get_df(self,hour_slice):
-        return pd.DataFrame(data=[self.spread[hour_slice],self.change[hour_slice],self.duration[hour_slice],self.amplitude[hour_slice]]).transpose()
+        return pd.DataFrame(data=[self.spread[hour_slice],self.change[hour_slice],self.duramp.duration[hour_slice],self.duramp.amplitude[hour_slice]]).transpose()
 
 
 
@@ -103,8 +104,8 @@ def get_all_data(full_data):
         part_amplitude = []
 
         for data in full_data:
-            part_duration.extend(data.duration[hour])
-            part_amplitude.extend(data.amplitude[hour])
+            part_duration.extend(data.duramp[hour].duration)
+            part_amplitude.extend(data.duramp[hour].amplitude)
 
         df_amp = pd.DataFrame(part_amplitude)
         df_amp = df_amp.reset_index(drop=True)        
@@ -124,4 +125,4 @@ def get_all_data(full_data):
 
 
 def divide(df):
-    return {'pozitive':df[df['amplitude']>0],'negative':df[df['amplitude']<0]}
+    return {'pozitive':df[df['amplitude']>0].median(),'negative':df[df['amplitude']<0].median()}