Merge pull request #16 from viventriglia/develop

Advanced Stats

src/pages/1_📈_Basic_Stats.py

@@ -5,6 +5,7 @@
     PLT_CONFIG,
     PLT_CONFIG_NO_LOGO,
     FAVICON,
+    DICT_GROUPBY_LEVELS,
 )
 from utils import (
     aggregate_by_ticker,
@@ -86,16 +87,11 @@
         options=["Macro Asset Classes", "Asset Classes", "Ticker"],
         horizontal=True,
     )
-    dict_group_by = {
-        "Macro Asset Classes": "macro_asset_class",
-        "Asset Classes": "asset_class",
-        "Ticker": "ticker",
-    }
     df_pivot_ = get_portfolio_pivot(
         df=df_j,
         df_dimensions=df_anagrafica,
         pf_actual_value=pf_actual_value,
-        aggregation_level=dict_group_by[group_by],
+        aggregation_level=DICT_GROUPBY_LEVELS[group_by],
     )
     df_pivot_.index += 1
     st.table(

src/pages/2_👩‍🔬_Advanced_Stats.py

@@ -1,17 +1,14 @@
 import streamlit as st
 
-from var import (
-    GLOBAL_STREAMLIT_STYLE,
-    # PLT_CONFIG,
-    # PLT_CONFIG_NO_LOGO,
-    FAVICON,
+from var import GLOBAL_STREAMLIT_STYLE, PLT_CONFIG_NO_LOGO, FAVICON, DICT_GROUPBY_LEVELS
+
+from utils import (
+    get_max_common_history,
+    write_disclaimer,
+    get_daily_returns,
 )
 
-# from utils import (
-#     sharpe_ratio,
-#     get_risk_free_rate_last_value,
-#     get_risk_free_rate_history,
-# )
+from plot import plot_correlation_map
 
 st.set_page_config(
     page_title="PFN | Advanced Stats",
@@ -21,15 +18,70 @@
 )
 
 st.markdown(GLOBAL_STREAMLIT_STYLE, unsafe_allow_html=True)
-st.warning("Work in progress! Please, come back later", icon="🚧")
 
-# df_common_history = get_max_common_history(ticker_list=ticker_list)
+if "data" in st.session_state:
+    df_transactions = st.session_state["data"]
+    df_registry = st.session_state["dimensions"]
+else:
+    st.error("Oops... there's nothing to display. Go through 🏠 first to load the data")
+    st.stop()
+
+ticker_list = df_transactions["ticker_yf"].unique().tolist()
+df_common_history = get_max_common_history(ticker_list=ticker_list)
+
+st.markdown("## Correlation of daily returns")
+
+col_l_up, col_r_up = st.columns([1, 1], gap="small")
+
+level = col_l_up.radio(
+    label="Aggregate by:",
+    options=["Macro Asset Classes", "Asset Classes", "Ticker"],
+    horizontal=True,
+    index=2,
+)
+
+enhance_corr = col_r_up.radio(
+    "Kind of correlation to enhance:",
+    options=["Positive", "Null", "Negative"],
+    horizontal=True,
+)
+
+st.markdown("")
+col_l_mid, col_r_mid = st.columns([1, 0.3], gap="small")
 
-# weights = [
-#     df_pivot[df_pivot["ticker_yf"].eq(x_)]["weight_pf"].values[0]
-#     for x_ in df_common_history.columns
-# ]
-# weighted_average = pd.Series(np.average(df_common_history, weights=weights, axis=1))
+first_transaction = df_transactions["transaction_date"].sort_values().values[0]
+first_day, last_day = col_l_mid.select_slider(
+    "Select a time slice:",
+    options=df_common_history.index,
+    value=[first_transaction, df_common_history.index[-1]],
+    format_func=lambda value: str(value)[:10],
+    label_visibility="collapsed",
+)
+
+df_daily_rets = get_daily_returns(
+    df=df_common_history.loc[first_day:last_day, :],
+    df_registry=df_registry,
+    level=DICT_GROUPBY_LEVELS[level],
+)
+
+fig = plot_correlation_map(
+    df=df_daily_rets.corr(),
+    enhance_correlation=enhance_corr.lower(),
+    lower_triangle_only=True,
+)
+st.plotly_chart(fig, use_container_width=True, config=PLT_CONFIG_NO_LOGO)
+
+# st.markdown("## Distribution of Daily Returns")
+
+# diff_prev_day = get_wealth_history(
+#     df_transactions=df_transactions,
+#     df_prices=get_max_common_history(ticker_list=ticker_list),
+# )['diff_previous_day']
+
+# import plotly.express as px
+# st.plotly_chart(px.histogram(diff_prev_day), use_container_width=True, config=PLT_CONFIG_NO_LOGO)
+
+#################################
 
 # returns = np.log(weighted_average.div(weighted_average.shift(1))).fillna(0)
 
@@ -46,3 +98,5 @@
 # )
 
 # st.write(sr)
+
+write_disclaimer()

src/plot.py

@@ -53,11 +53,11 @@ def plot_pnl_by_asset_class(
 
 def plot_wealth(df: pd.DataFrame) -> go.Figure:
     fig = px.area(
-        data_frame=df,
-        x=df.index,
-        y="ap_daily_value",
+        data_frame=df, x=df.index, y="ap_daily_value", custom_data=["diff_previous_day"]
+    )
+    fig.update_traces(
+        hovertemplate="%{x}: <b>%{y:,.0f}€</b> <extra>Gain/Loss on previous day: %{customdata:,.0f}€</extra>"
     )
-    fig.update_traces(hovertemplate="%{x}: <b>%{y:,.0f}€</b>")
     fig.update_layout(
         autosize=False,
         height=550,
@@ -70,3 +70,45 @@ def plot_wealth(df: pd.DataFrame) -> go.Figure:
         showlegend=False,
     )
     return fig
+
+
+def plot_correlation_map(
+    df: pd.DataFrame,
+    enhance_correlation: Literal["positive", "null", "negative"],
+    lower_triangle_only: bool = False,
+):
+    if lower_triangle_only:
+        mask = np.triu(np.ones_like(df, dtype=bool))
+
+    if enhance_correlation == "positive":
+        cuts = [-1, 0, 1]
+        colors = ["white", "white", "darkred"]
+    elif enhance_correlation == "null":
+        cuts = [-1, -0.3, 0, 0.3, 1]
+        colors = ["white", "white", "darkgreen", "white", "white"]
+    elif enhance_correlation == "negative":
+        cuts = [-1, 0, 1]
+        colors = ["darkblue", "white", "white"]
+    colorscale = [[(cut_ + 1) / 2, col_] for cut_, col_ in zip(cuts, colors)]
+
+    fig = go.Figure(
+        go.Heatmap(
+            z=df.mask(mask) if lower_triangle_only else df,
+            x=df.columns,
+            y=df.columns,
+            colorscale=colorscale,
+            zmin=-1,
+            zmax=1,
+        )
+    )
+    fig.update_traces(hovertemplate="<b>%{z:.2f} <extra></extra>")
+    fig.update_layout(
+        height=500,
+        hoverlabel_font_size=PLT_FONT_SIZE,
+        margin=dict(l=0, r=0, t=30, b=0),
+        yaxis=dict(autorange="reversed", showgrid=False),
+        # paper_bgcolor="rgba(0,0,0,0)",
+        # plot_bgcolor="rgba(0,0,0,0)",
+    )
+
+    return fig

src/utils.py

@@ -274,12 +274,14 @@ def get_wealth_history(
         df_asset_allocation.loc[data, ticker] += total_shares
     df_asset_allocation = df_asset_allocation.cumsum()
 
-    df_wealth = (
+    df_wealth = pd.DataFrame(
         df_asset_allocation.multiply(df_prices.loc[begin_date:])
         .fillna(method="ffill")
         .sum(axis=1)
         .rename("ap_daily_value")
     )
+    df_wealth["diff_previous_day"] = df_wealth.diff()
+
     return df_wealth
 
 
@@ -341,3 +343,25 @@ def write_disclaimer() -> None:
         </span> </center>',
         unsafe_allow_html=True,
     )
+
+
+@st.cache_data(ttl=CACHE_EXPIRE_SECONDS, show_spinner=False)
+def get_daily_returns(
+    df: pd.DataFrame,
+    df_registry: pd.DataFrame,
+    level: Literal["ticker", "asset_class", "macro_asset_class"],
+):
+    df_daily_rets = df.pct_change()[1:]
+
+    if level == "ticker":
+        return df_daily_rets
+    else:
+        classes = df_registry[level].unique()
+        df_daily_rets_classes = pd.DataFrame(columns=classes)
+        for class_ in classes:
+            cols_to_sum = df_registry[df_registry[level].eq(class_)][
+                "ticker_yf"
+            ].to_list()
+
+            df_daily_rets_classes[class_] = df_daily_rets[cols_to_sum].sum(axis=1)
+        return df_daily_rets_classes

src/var.py

@@ -8,7 +8,7 @@
     os.path.join(*split_script_running_path[0 : len(split_script_running_path) - 1])
 )
 
-APP_VERSION = "0.1.0"
+APP_VERSION = "0.1.1"
 
 # Data/images
 
@@ -44,3 +44,8 @@
 # Others
 
 TRADING_DAYS_YEAR = 252
+DICT_GROUPBY_LEVELS = {
+    "Macro Asset Classes": "macro_asset_class",
+    "Asset Classes": "asset_class",
+    "Ticker": "ticker",
+}