from IPython.display import Markdown, display
from tulip.analysis.equity_related.analytics import *
from tulip.analysis.equity_related.visualization import *
from tulip.utils.notebook_related import *
from tulip.plots import *
import pycountry

NOTEBOOK_TITLE = "Japan - MSCI JP"

CTRY = "JPN"
CTRY_EQ_INDEX = "MXJP Index"  # 'TPX Index'
CTRY_GOV = "GTJPY10Y Govt"
CTRY_BEI = "JYGGBE10 Index"
CURRENCY = "JPY Curncy"
GOLDMAN_CTRY = "Japan (SA)"

ERP = 0.02
DIV_PO = 0.5  # Crazy different results for some geographies
INTERCEPT_IN_MODEL_1 = False

# Parameters
NOTEBOOK_TITLE = "Japan - MSCI JP"
CTRY = "JPN"
CTRY_EQ_INDEX = "MXJP Index"
CTRY_GOV = "GTJPY10Y Govt"
CTRY_BEI = "JYGGBE10 Index"
CURRENCY = "JPY Curncy"
GOLDMAN_CTRY = "Japan (SA)"
ERP = 0.02
DIV_PO = 0.5
INTERCEPT_IN_MODEL_1 = False

BASE_CURRENCY = bb.bdp(CURRENCY, fields=["BASE_CRNCY"])["base_crncy"].squeeze()
PY_CTRY = pycountry.countries.get(alpha_3=CTRY)
if PY_CTRY:
    CTRY_2 = PY_CTRY.alpha_2
else:
    if CURRENCY == "EUR Curncy":
        CTRY_2 = "EU"

Japan EWJ¶

## ----- Get Data -----
# Get historical EPS and SPS data
BBERG_EPS_FIELD = (
    "IS_DIL_EPS_CONT_OPS" if CTRY in ["USA", "CAN"] else "IS_EARN_BEF_XO_ITEMS_PER_SH"
)
historical_eps = get_historical_eps(
    CTRY_EQ_INDEX, start_date="1992-01-01", eps_field=BBERG_EPS_FIELD
)
historical_sps = get_historical_sps(CTRY_EQ_INDEX, start_date="1992-01-01")

# Get GDP data
NGDP, RGDP = get_gdp_data(CTRY)

# Get daily index data
daily_index_df = get_daily_index_data(CTRY_EQ_INDEX).droplevel(0, axis=1)
daily_index_df["EPS_Delta"] = (
    daily_index_df["BEST_EPS"] - daily_index_df["TRAIL_12M_EPS"]
)
daily_index_df["EPS_Delta_Pct"] = (
    daily_index_df["EPS_Delta"] / daily_index_df["TRAIL_12M_EPS"]
)

# Get yield data
nominal_yield_df, bei_df = get_yield_data(CTRY_GOV, CTRY_BEI)

data_per_q_df = calculate_data_per_q(
    NGDP=NGDP.ts.rename("NGDP"),
    RGDP=RGDP.ts.rename("RGDP"),
    EPS=historical_eps,
    SPS=historical_sps,
    currency=CURRENCY,
    daily_index_df=daily_index_df,
    base_currency=BASE_CURRENCY,
)

nominal_gdp_forecast = get_NGDP_forecast(GOLDMAN_CTRY)

# ----- Compute Reference Dates -----
last_fwd_eps = daily_index_df["BEST_EPS"].tail(1).squeeze()
last_pe = daily_index_df["BEST_PE_RATIO"].tail(1).squeeze()
last_eps = daily_index_df.loc[:, "TRAIL_12M_EPS"].tail(1).squeeze()
last_valid_gdp_q = NGDP.info.observation_end
last_actual_eps = data_per_q_df["EPS"].last_valid_index()
horizon_date = last_actual_eps + pd.offsets.MonthEnd(12)

## ----- Compute IEG -----
ieg = compute_ieg(
    pe_ratio=daily_index_df["BEST_PE_RATIO"],
    nominal_yield=nominal_yield_df.iloc[:, 0],
    bei=bei_df.iloc[:, 0],
    equity_risk_premium=ERP,
    dividend_payout_ratio=DIV_PO,
)

ieg.loc[:, ["NY_10Y", "BEI_10Y", "DY"]] = ieg.loc[:, ["NY_10Y", "BEI_10Y", "DY"]].apply(
    fill_forward
)

ieg_summary = organize_ieg_summary(ieg, covid_start_date, covid_end_date)

Valuation¶

default_x_range = (
    ieg.index[-1] - pd.offsets.MonthEnd(60),
    ieg.index[-1] + pd.offsets.MonthEnd(9),
)
pe_fig = plot_lines(
    daily_index_df["BEST_PE_RATIO"],
    default_x_range=default_x_range,
    title=f"{CTRY} | {CTRY_EQ_INDEX}: FWD PE Ratio",
    source="Bloomberg, Note: Uses the BF methodology",
)

ieg_fig = plot_lines(
    ieg[["NIEG", "RIEG"]].rename(columns={"NIEG": "Nominal", "RIEG": "Real"}),
    tick_format="0.0%",
    title=f"{CTRY} | {CTRY_EQ_INDEX}: Implied Earnings Growth",
    default_x_range=default_x_range,
    source="Bloomberg, Kate Capital",
)

erp_fig = plot_lines(
    ieg["EY-NY"],
    tick_format="0.0%",
    title=f"{CTRY} | {CTRY_EQ_INDEX}: Earnings yield - Nominal yield",
    default_x_range=default_x_range,
    source="Bloomberg, Kate Capital",
)

display_table_and_chart(
    table=style_ieg_summary(ieg_summary),
    chart=pe_fig,
    table_width="50%",
    chart_width="50%",
)

display_two_charts(
    erp_fig,
    ieg_fig,
    chart1_width="50%",
    chart2_width="50%",
    container_width="100%",
)

Earnings Expectations¶

eps_summary = calculate_eps_summary(daily_index_df=daily_index_df)
styled_eps_summary = style_eps_summary(
    eps_summary, caption=f"{CTRY} | {CTRY_EQ_INDEX}: EPS Summary"
)

eps_fig = plot_lines(
    daily_index_df[["TRAIL_12M_EPS", "BEST_EPS", "EPS_Delta"]],
    title=f"{CTRY} | {CTRY_EQ_INDEX}: EPS Evolution",
    default_x_range=default_x_range,
    axis_title="EPS",
    source="Bloomberg, Kate Capital",
)

switch_trace_to_secondary_axis(
    eps_fig,
    "EPS_Delta",
    secondary_axis_title="Delta",
    default_y2_range=(0, daily_index_df["EPS_Delta"].max()),
    as_area=True,
)

display_table_and_chart(
    table=styled_eps_summary, chart=eps_fig, table_width="30%", chart_width="70%"
)

Positioning¶

# todo

EPS Forecasts¶

MODEL 0: We forecast EPS extrapolating historical growth (Naive Baseline)¶

import tulip.analysis.equity_related.eps_forecasts.model_0 as eps_model_0

model_0, act_vs_pred_df, rsq, n_year_cagr = eps_model_0.fit_model(
    actual_q_df=data_per_q_df,
    actual_daily_q=daily_index_df,
    years_lookback=10,
)

model_0_forecast_df = eps_model_0.make_forecast(fitted_model=model_0)

model_0_fig = eps_model_0.plot_forecast(
    forecast_df=model_0_forecast_df, act_vs_pred_df=act_vs_pred_df, figsize=(12.8, 4.8)
)

model_0_summary = eps_model_0.summarize_forecast(
    forecast_df=model_0_forecast_df, rsq=rsq, horizon=horizon_date
)
display(model_0_summary.to_frame("EPS_MODEL_0").T)

MODEL 1: We forecast EPS_YoY growth quaterly using YoY Historical Nominal GDP Growth and GS Forecasts¶

import tulip.analysis.equity_related.eps_forecasts.model_1 as eps_model_1

model_1, act_vs_pred_df, rsq = eps_model_1.fit_model(
    df=data_per_q_df,
    target="EPS_YoY",
    predictors=["NGDP_YoY"],
    clean_outliers=True,
    add_intercept=INTERCEPT_IN_MODEL_1,
)

model_1_forecast_df = eps_model_1.make_forecast(
    fitted_model=model_1,
    data_per_q_df=data_per_q_df,
    nominal_gdp_forecast=nominal_gdp_forecast,
)

model_1_plot = eps_model_1.plot_model_output(
    model_1_forecast_df, data_per_q_df, act_vs_pred_df
)

coefficients = eps_model_1.summarize_coefficients(fitted_model=model_1)
model_1_summary = eps_model_1.summarize_forecast(
    forecast_df=model_1_forecast_df, rsq=rsq
)
display(coefficients)
display(model_1_summary.to_frame("EPS_MODEL_1").T)

MODEL 2: We forecast EPS by first forecasting SPS and Net Margins using YoY Historical Nominal GDP Growth and GS Forecasts¶

The evolution of SPS tends to follow changes in NGDP more closely than EPS

import tulip.analysis.equity_related.eps_forecasts.model_2 as eps_model_2

fitted_sps_model, fitted_nim_model, act_vs_pred_df, rsq = eps_model_2.fit_model(
    df=data_per_q_df, clean_outliers=True, add_intercept=False
)

model_2_forecast_df = eps_model_2.make_forecast(
    fitted_sps_model=fitted_sps_model,
    fitted_nim_model=fitted_nim_model,
    data_per_q_df=data_per_q_df,
    nominal_gdp_forecast=nominal_gdp_forecast,
)

model_2_plot = eps_model_2.plot_output(
    forecast_df=model_2_forecast_df,
    actual_df=data_per_q_df,
    act_vs_pred_df=act_vs_pred_df,
    clean_outliers=True,
    figsize=(12.8, 9.6),
)

model_2_summary = eps_model_2.summarize_forecast(
    forecast_df=model_2_forecast_df, rsq=rsq
)
display(model_2_summary.to_frame("EPS_MODEL_2").T)

EPS Forecast Results:¶

summary = pd.DataFrame(
    {
        "Model 0: Extrapolation": model_0_summary,
        "Model 1: Based on EPS YoY": model_1_summary,
        "Model 2: Based on SPS and Margins": model_2_summary,
    }
).T
summary.insert(0, "EPS Expectation", last_fwd_eps)
summary

from tulip_mania.notebook_related import notebook_updated

notebook_updated()