DolphinDB/modules/indicatorCalculator.dos

module fundit::indicatorCalculator

use fundit::dataPuller
use fundit::returnCalculator
use fundit::navCalculator

/*
 *  Annulized multiple
 */
def get_annulization_multiple(freq) {

  ret = 1;
  
  if (freq == 'd') {
    ret = 252; // We have differences here between Java and DolphinDB, Java uses 365.25 days
  } else if (freq == 'w') {
    ret = 52;
  } else if (freq == 'm') {
    ret = 12;
  } else if (freq == 'q') {
    ret = 4;
  } else if (freq == 's') {
    ret = 2;
  } else if (freq == 'a') {
    ret = 1;
  }
  
  return ret;
}

/*
 *   将VaR包裹一层，使之成为系统认可的聚集函数 
 *   @param returns <DOUBLE VECTOR>: 非空收益率
 *   @param method <STRING>: 'normal', 'logNormal', 'historical', 'monteCarlo'
 *   @param confidenceLevel <DOUBLE>: 置信水平，取值区间(0, 1)
 * 
 */
defg aggVaR(returns, method, confidenceLevel) {

	return returns.VaR(method, confidenceLevel);
}

/*
 *   将CVaR包裹一层，使之成为系统认可的聚集函数 
 *   @param returns <DOUBLE VECTOR>: 非空收益率
 *   @param method <STRING>: 'normal', 'logNormal', 'historical', 'monteCarlo'
 *   @param confidenceLevel <DOUBLE>: 置信水平，取值区间(0, 1)
 * 
 */
defg aggCVaR(returns, method, confidenceLevel) {

	return returns.CVaR(method, confidenceLevel);
}

/*
 *   回撤
 * 
 * 
 */
defg maxDrawdown(navs) {

	return max(1 - navs \ cummax(navs));
}


/*
 *   Trailing Monthly Return, Standard Deviation, Skewness, Kurtosis, Max Drawdown, VaR, CVaR, Calmar Ratio
 *     
 *   @param entity_info <TABLE>: xxx_information表，NEED COLUMNS entity_id, inception_date
 *   @param ret <TABLE>: 收益表，需要有 entity_id, price_dat, end_date, nav
 *   @param trailing_month <STRING>: trailing X month or ytd, incep
 *     
 *   NOTE: standard deviation of Java version is noncompliant-GIPS annulized number
 *     
 *   Create:  20240904                                                                     Joey
 *            TODO: SQL is wrong for max drawdowns
 *            TODO: var, cvar, calmar are off; std dev, skewness, kurtosis are slightly off
 *            TODO: SQL is missing for portfolio since inception date return
 *            TODO: Java calculates max drawdown even there is no nav
 *            TODO: Java ytd worst month could be wrong (i.e. portfolio 166002, 2024-03)
 *     
 */
def cal_basic_performance(entity_info, ret, trailing_month) {

    // accumulate 版的 skewness, kurtosis, var, cvar 似乎都不对劲，只好找个笨办法来实现
    if(trailing_month == 'incep') {
        // 需要至少6个数才计算标准差、峰度、偏度
        t0 = SELECT price_date.max() AS price_date,
                    prod(1+ret)-1 AS trailing_ret, 
                    prod(1+ret)-1 AS trailing_ret_a,
                    iif(count(entity_id) > 5, std(ret), null) AS std_dev,
                    iif(count(entity_id) > 5, skew(ret, false), null) AS skewness,
                    iif(count(entity_id) > 5, kurtosis(ret, false), null)-3 AS kurtosis,
                    min(ret) AS wrst_month
             FROM ret
             WHERE ret > -1
             GROUP BY entity_id
             CGROUP BY end_date
             ORDER BY end_date;

        // 年化收益（给后面计算Calmar用）
        UPDATE t0 
            SET trailing_ret_a = (1+trailing_ret).pow(12\(t0.end_date - ei.inception_date.month()))-1
        FROM ej(t0, entity_info ei, 'entity_id')
        WHERE t0.end_date > ei.inception_date.month() + 12;
        
        // 不会用上面的办法算最大回撤, VaR, CVaR
        t_var = SELECT entity_id, end_date, ret,
                       cummax(1 - nav \ cummax(nav)) AS drawdown,
                       - cumpercentile(ret, 5, 'linear') AS var
                FROM ret WHERE ret > -1
                CONTEXT BY entity_id;

        t_cvar = SELECT entity_id, end_date, drawdown, var,
                        - cumavg(iif(ret <= -var, ret, null)) AS cvar
                 FROM t_var
                 CONTEXT BY entity_id;

        t1 = SELECT t0.*, t_cvar.drawdown, t_cvar.var, t_cvar.cvar
             FROM t0 LEFT JOIN t_cvar ON t0.entity_id = t_cvar.entity_id AND t0.end_date = t_cvar.end_date
             ORDER BY t0.entity_id, t0.end_date;

    } else if(trailing_month == 'ytd') {

        t1 = SELECT entity_id, end_date, price_date.cummax() AS price_date,
                    cumprod(1+ret)-1 AS trailing_ret,
                    cumprod(1+ret)-1 AS trailing_ret_a, // no need annulization for ytd
                    iif(cumcount(entity_id) > 5, cumstd(ret), null) AS std_dev,
                    iif(cumcount(entity_id) > 5, tmoving(skew{, false}, end_date, ret, 12), null) AS skewness,
                    iif(cumcount(entity_id) > 5, tmoving(kurtosis{, false}, end_date, ret, 12)-3, null) AS kurtosis,
                    cummin(ret) AS wrst_month,
                    cummax(1 - nav \ cummax(nav)) AS drawdown
        FROM ret WHERE ret > -1
        CONTEXT BY entity_id, end_date.year()
        ORDER BY entity_id, end_date;

    // trailing x month
    } else {
        	
        win = trailing_month$INT;
        
        t1 = SELECT entity_id, end_date, price_date.mmax(win) AS price_date,
                    mprod(1+ret, win)-1 AS trailing_ret,
                    iif(trailing_month > 12,
                        mprod(1+ret, win).pow(12\win)-1,
                        mprod(1+ret, win)-1) AS trailing_ret_a,
                    mstd(ret, win) AS std_dev,
                    mskew(ret, win, false) AS skewness,
                    mkurtosis(ret, win, false) - 3 AS kurtosis,
                    mmin(ret, win) AS wrst_month,
                    moving(maxDrawdown, nav, win) AS drawdown,
                    moving(aggVaR{, 'historical', 0.95}, ret, win) AS var,
                    moving(aggCVaR{, 'historical', 0.95}, ret, win) AS cvar
        FROM ret WHERE ret > -1
        CONTEXT BY entity_id
        ORDER BY entity_id, end_date;
    }

    t1.addColumn('calmar', DOUBLE);
    UPDATE t1 SET calmar =  iif(drawdown == 0, null, trailing_ret_a\drawdown);

    return t1;

}


/*
 *   Lower Partial Moment
 *   NOTE: risk free rate is used as Minimal Accepted Rate (MAR) here
 * 
 */
def cal_LPM(ret, risk_free, trailing_month) {
    
    t = SELECT *, cumcount(entity_id) AS cnt FROM ret WHERE ret > -1 CONTEXT BY entity_id;

    if(trailing_month == 'incep') {

        lpm = SELECT entity_id, end_date,
                    iif(cumcount(end_date) > 5, (cumsum  (iif(rfr.ret > t.ret, rfr.ret - t.ret, 0)) \ cumcount(end_date)).pow(1\1), null) AS lpm1, 
                    iif(cumcount(end_date) > 5, (cumsum2 (iif(rfr.ret > t.ret, rfr.ret - t.ret, 0)) \ cumcount(end_date)).pow(1\2), null) AS lpm2, 
                    iif(cumcount(end_date) > 5, (cumsum3 (iif(rfr.ret > t.ret, rfr.ret - t.ret, 0)) \ cumcount(end_date)).pow(1\3), null) AS lpm3
              FROM t 
              INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
              CONTEXT BY entity_id
              ORDER BY entity_id, end_date;
    	
    } else if(trailing_month == 'ytd') {

        lpm = SELECT entity_id, end_date,
                    iif(cumcount(end_date) > 5, (cumsum  (iif(rfr.ret > t.ret, rfr.ret - t.ret, 0)) \ cumcount(end_date)).pow(1\1), null) AS lpm1, 
                    iif(cumcount(end_date) > 5, (cumsum2 (iif(rfr.ret > t.ret, rfr.ret - t.ret, 0)) \ cumcount(end_date)).pow(1\2), null) AS lpm2, 
                    iif(cumcount(end_date) > 5, (cumsum3 (iif(rfr.ret > t.ret, rfr.ret - t.ret, 0)) \ cumcount(end_date)).pow(1\3), null) AS lpm3
              FROM t 
              INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
              CONTEXT BY entity_id, end_date.year()
              ORDER BY entity_id, end_date;

    } else {

        win = trailing_month$INT;
        lpm = SELECT t.entity_id, t.end_date,
                     (msum (iif(rfr.ret > t.ret, rfr.ret - t.ret, 0), win) \ mcount(end_date, win)).pow(1\1) AS lpm1, 
                     (msum2(iif(rfr.ret > t.ret, rfr.ret - t.ret, 0), win) \ mcount(end_date, win)).pow(1\2) AS lpm2, 
                     (moving(sum3, iif(rfr.ret > t.ret, rfr.ret - t.ret, 0), win) \ mcount(end_date, win)).pow(1\3) AS lpm3
              FROM t 
              INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
              CONTEXT BY t.entity_id
              ORDER BY entity_id, end_date;

    }
    

    return lpm;
}

/*
 *     Downside Devision, Omega Ratio, Sortino Ratio, Kappa Ratio
 * 
 *     TODO: Java version of Downside Deviation (LPM2) uses cnt-1 as denominator to calculate mean excess return, which might be wrong
 *           Java version of Omega could be wrong because Java uses annualized returns and cnt-1 
 *           Java'version of Kappa could be very wrong
 *           
 */
def cal_omega_sortino_kappa(ret, risk_free, trailing_month) {

    lpm = cal_LPM(ret, risk_free, trailing_month);

    if(trailing_month == 'incep') {

        tb = SELECT t.entity_id, t.end_date,
                l.lpm2 AS ds_dev,
                (t.ret - rfr.ret ).cumavg() \ l.lpm1 + 1 AS omega,
                (t.ret - rfr.ret ).cumavg() \ l.lpm2 AS sortino,
                (t.ret - rfr.ret ).cumavg() \ l.lpm3 AS kappa
              FROM ret t
              INNER JOIN lpm l ON t.entity_id = l.entity_id AND t.end_date = l.end_date
              INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
              WHERE t.ret > -1
              CONTEXT BY t.entity_id;

    } else if(trailing_month == 'ytd') {

        tb = SELECT t.entity_id, t.end_date,
                l.lpm2 AS ds_dev,
                (t.ret - rfr.ret ).cumavg() \ l.lpm1 + 1 AS omega,
                (t.ret - rfr.ret ).cumavg() \ l.lpm2 AS sortino,
                (t.ret - rfr.ret ).cumavg() \ l.lpm3 AS kappa
              FROM ret t
              INNER JOIN lpm l ON t.entity_id = l.entity_id AND t.end_date = l.end_date
              INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
              WHERE t.ret > -1
              CONTEXT BY t.entity_id, t.end_date.year();

    } else {

        win = trailing_month$INT;
        
        tb = SELECT t.entity_id, t.end_date,
                l.lpm2 AS ds_dev,
                (t.ret - rfr.ret ).mavg(win) \ l.lpm1 + 1 AS omega,
                (t.ret - rfr.ret ).mavg(win) \ l.lpm2 AS sortino,
                (t.ret - rfr.ret ).mavg(win) \ l.lpm3 AS kappa
              FROM ret t
              INNER JOIN lpm l ON t.entity_id = l.entity_id AND t.end_date = l.end_date
              INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
              WHERE t.ret > -1
              CONTEXT BY t.entity_id;
    }

    return tb;
}


/*
 *    Winning Ratio, Tracking Error, Information Ratio
 *    
 *    NOTE: mcount is very unique in mFun, because it doesn't support minPeriods(BUG?), while others default minPeriods = window.
 *          As a result, we have to delete records having winrate but no tracking error and info ratio for the sake of consisence
 *   
 *    TODO: Win Rate incept is off, because Java incorrectly takes all end_date as denominator even when benchmark has no price
 *          Information Ratio is way off!
 *          Not sure how to describe a giant number("inf"), for now 999 is used
 */
def cal_benchmark_tracking(ret, benchmarks, bmk_ret, trailing_month) {

    if(trailing_month == 'incep') {

        t0 = SELECT t.entity_id, t.end_date, t.price_date,
                    t.ret, bmk.ret AS ret_bmk,
                    t.entity_id.cumcount() AS cnt,
                    t.ret - bmk.ret AS exc_ret, bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id
             WHERE t.ret > -1
               AND bmk.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id;

        t = SELECT entity_id, end_date, benchmark_id,
                   iif(cnt > 5, cumcount(iif(exc_ret >= 0, 1, null)) \ cnt, null) AS winrate,
                   iif(cnt > 5, exc_ret.cumstd(), null) AS track_error, 
                   iif(cnt > 5, iif(exc_ret.cumstd() == 0, null, exc_ret.cumavg() \ exc_ret.cumstd()), 5) AS info
            FROM t0
            CONTEXT BY entity_id, benchmark_id
            ORDER BY entity_id, end_date, benchmark_id;

    } else if(trailing_month == 'ytd') {

        t0 = SELECT t.entity_id, t.end_date, t.price_date,
                    t.ret, bmk.ret AS ret_bmk,
                    t.entity_id.cumcount() AS cnt, t.ret - bmk.ret AS exc_ret, bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id
             WHERE t.ret > -1
               AND bmk.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id, t.end_date.year();

        t = SELECT entity_id, end_date, benchmark_id,
                   iif(cnt > 5, cumcount(iif(exc_ret >= 0, 1, null)) \ cnt, null) AS winrate,
                   iif(cnt > 5, exc_ret.cumstd(), null) AS track_error, 
                   iif(cnt > 5, iif(exc_ret.cumstd() == 0, null, exc_ret.cumavg() \ exc_ret.cumstd()), null) AS info
            FROM t0
            CONTEXT BY entity_id, benchmark_id, end_date.year()
            ORDER BY entity_id, end_date, benchmark_id;
    } else {

        win = trailing_month$INT;

        t0 = SELECT t.entity_id, t.end_date, t.price_date,
                    t.ret, bmk.ret AS ret_bmk,
                    t.entity_id.mcount(win) AS cnt,
                    t.ret - bmk.ret AS exc_ret, bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id
             WHERE t.ret > -1
               AND bmk.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id;

        t = SELECT entity_id, end_date, benchmark_id,
                   iif(cnt > 5, mcount(iif(exc_ret >= 0, 1, null), win) \ cnt, null) AS winrate, 
                   iif(cnt > 5, mstd(exc_ret, win), null) AS track_error, 
                   iif(cnt > 5, iif(mstd(exc_ret, win) == 0, null, mavg(exc_ret, win) \ mstd(exc_ret, win)), null) AS info
            FROM t0
            CONTEXT BY entity_id, benchmark_id
            ORDER BY entity_id, end_date, benchmark_id;
    }

    return t; //SELECT * FROM t WHERE track_error IS NOT NULL;
}

/*
 *   Alpha & Beta
 *   NOTE: alpha of Java version is wrong because it doesn't use risk free rate
 */
def cal_alpha_beta(ret, benchmarks, bmk_ret, risk_free, trailing_month) {

    t = SELECT t.entity_id, t.end_date, t.ret, bm.benchmark_id, bmk.ret AS ret_bmk
        FROM ret t
        INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
        INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id
        WHERE t.ret > -1
          AND bmk.ret > -1;

    if(trailing_month == 'incep') {

        beta = SELECT entity_id, end_date, benchmark_id,
                      iif(cumcount(end_date) > 5, ret.cumbeta(ret_bmk), null) AS beta
               FROM t CONTEXT BY entity_id, benchmark_id;

        alpha = SELECT t.entity_id, t.end_date, t.benchmark_id, beta.beta AS beta, 
                       (t.ret - rfr.ret).cumavg() - beta.beta * (t.ret_bmk - rfr.ret).cumavg() AS alpha
                FROM t 
                INNER JOIN beta beta ON t.entity_id = beta.entity_id AND t.benchmark_id = beta.benchmark_id AND t.end_date = beta.end_date
                INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                CONTEXT BY t.entity_id, t.benchmark_id
                ORDER BY t.entity_id, t.end_date, t.benchmark_id;

    } else if(trailing_month == 'ytd') {

        beta = SELECT entity_id, end_date, benchmark_id,
                      iif(cumcount(end_date) > 5, ret.cumbeta(ret_bmk), null) AS beta
               FROM t CONTEXT BY entity_id, benchmark_id, end_date.year();

        alpha = SELECT t.entity_id, t.end_date, t.benchmark_id, beta.beta AS beta, 
                       (t.ret - rfr.ret).cumavg() - beta.beta * (t.ret_bmk - rfr.ret).cumavg() AS alpha
                FROM t 
                INNER JOIN beta beta ON t.entity_id = beta.entity_id AND t.benchmark_id = beta.benchmark_id AND t.end_date = beta.end_date
                INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                CONTEXT BY t.entity_id, t.benchmark_id, t.end_date.year()
                ORDER BY t.entity_id, t.end_date, t.benchmark_id;

    } else {

        win = trailing_month$INT;

        beta = SELECT entity_id, end_date, benchmark_id, 
                      iif(mcount(end_date, win) > 5, ret.mbeta(ret_bmk, win), null) AS beta 
               FROM t CONTEXT BY entity_id, benchmark_id;

        alpha = SELECT t.entity_id, t.end_date, t.benchmark_id, beta.beta AS beta, 
                       (t.ret - rfr.ret).mavg(win) - beta.beta * (t.ret_bmk - rfr.ret).mavg(win) AS alpha
                FROM t 
                INNER JOIN beta beta ON t.entity_id = beta.entity_id AND t.benchmark_id = beta.benchmark_id AND t.end_date = beta.end_date
                INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                CONTEXT BY t.entity_id, t.benchmark_id
                ORDER BY t.entity_id, t.end_date, t.benchmark_id;
    }

    return alpha;
}


/*
 *  Upside/Down Capture Return/Ratio
 *  
 *  TODO: trailing x month values are way off!
 * 
 */
def cal_capture_ratio(ret, benchmarks, bmk_ret, trailing_month) {

    if(trailing_month == 'incep') {

        t1 = SELECT t.entity_id, t.end_date,
                    (1 + iif(bmk.ret >= 0, t.ret, 0)).cumprod() AS upside_ret,
                    (1 + iif(bmk.ret >= 0, bmk.ret, 0)).cumprod() AS bmk_upside_ret,
                    cumcount(iif(bmk.ret >= 0, 1, null)) AS bmk_upside_cnt,
                    (1 + iif(bmk.ret < 0, t.ret, 0)).cumprod() AS downside_ret,
                    (1 + iif(bmk.ret < 0, bmk.ret, 0)).cumprod() AS bmk_downside_ret,
                    cumcount(iif(bmk.ret < 0, 1, null)) AS bmk_downside_cnt,
                    bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id AND t.end_date = bmk.end_date
             WHERE t.ret > -1
               AND bmk.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id;


    } else if(trailing_month == 'ytd') {

        t1 = SELECT t.entity_id, t.end_date,
                    (1 + iif(bmk.ret >= 0, t.ret, 0)).cumprod() AS upside_ret,
                    (1 + iif(bmk.ret >= 0, bmk.ret, 0)).cumprod() AS bmk_upside_ret,
                    cumcount(iif(bmk.ret >= 0, 1, null)) AS bmk_upside_cnt,
                    (1 + iif(bmk.ret < 0, t.ret, 0)).cumprod() AS downside_ret,
                    (1 + iif(bmk.ret < 0, bmk.ret, 0)).cumprod() AS bmk_downside_ret,
                    cumcount(iif(bmk.ret < 0, 1, null)) AS bmk_downside_cnt,
                    bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id AND t.end_date = bmk.end_date
             WHERE t.ret > -1
               AND bmk.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id, t.end_date.year();


    } else {

        win = trailing_month$INT;

        t1 = SELECT t.entity_id, t.end_date,
                    (1 + iif(bmk.ret >= 0, t.ret, 0)).mprod(win) AS upside_ret,
                    (1 + iif(bmk.ret >= 0, bmk.ret, 0)).mprod(win) AS bmk_upside_ret,
                    mcount(iif(bmk.ret >= 0, 1, null), win) AS bmk_upside_cnt,
                    (1 + iif(bmk.ret < 0, t.ret, 0)).mprod(win) AS downside_ret,
                    (1 + iif(bmk.ret < 0, bmk.ret, 0)).mprod(win) AS bmk_downside_ret,
                    mcount(iif(bmk.ret < 0, 1, null), win) AS bmk_downside_cnt,
                    bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id AND t.end_date = bmk.end_date
             WHERE t.ret > -1
               AND bmk.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id;

    }

    t = SELECT entity_id, end_date, benchmark_id,
                    t1.upside_ret.pow(1 \ t1.bmk_upside_cnt)-1 AS upside_capture_ret,
                   (t1.upside_ret.pow(1 \ t1.bmk_upside_cnt)-1)/(t1.bmk_upside_ret.pow(1 \ t1.bmk_upside_cnt)-1) AS upside_capture_ratio,
                    t1.downside_ret.pow(1 \ t1.bmk_downside_cnt)-1 AS downside_capture_ret,
                   (t1.downside_ret.pow(1 \ t1.bmk_downside_cnt)-1)/(t1.bmk_downside_ret.pow(1 \ t1.bmk_downside_cnt)-1) AS downside_capture_ratio
             FROM t1 
             ORDER BY entity_id, benchmark_id, end_date;

    return t;
}

/*
 *    Sharpe Ratio
 *    NOTE: Java version is noncompliant-GIPS annulized number
 */
def cal_sharpe(ret, std_dev, risk_free, trailing_month) {

    if(trailing_month == 'incep') {

        sharpe = SELECT t.entity_id, t.end_date, (t.ret - rfr.ret).cumavg() / std.std_dev AS sharpe
                 FROM ret t
                 INNER JOIN std_dev std ON t.entity_id = std.entity_id AND t.end_date = std.end_date
                 INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                 WHERE std.std_dev <> 0 AND t.ret > -1
                 CONTEXT BY t.entity_id;
                 
    } else if(trailing_month == 'ytd') {

        sharpe = SELECT t.entity_id, t.end_date, (t.ret - rfr.ret).cumavg() / std.std_dev AS sharpe
                 FROM ret t
                 INNER JOIN std_dev std ON t.entity_id = std.entity_id AND t.end_date = std.end_date
                 INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                 WHERE std.std_dev <> 0 AND t.ret > -1
                 CONTEXT BY t.entity_id, t.end_date.year();
    } else {

        win = trailing_month$INT;

        sharpe = SELECT t.entity_id, t.end_date, (t.ret - rfr.ret).mavg(win) / std.std_dev AS sharpe
                 FROM ret t
                 INNER JOIN std_dev std ON t.entity_id = std.entity_id AND t.end_date = std.end_date
                 INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                 WHERE std.std_dev <> 0 AND t.ret > -1
                 CONTEXT BY t.entity_id;
    }

    return sharpe;
}

/*
 *    Treynor Ratio = annulized excess return / beta
 *    
 *    TODO: ytd is off because Java uses non-GIPS rule to annulize return
 */
def cal_treynor(ret, risk_free, beta, trailing_month) {

    if(trailing_month == 'incep') {

        t = SELECT *, cumcount(entity_id) AS cnt 
            FROM ret t 
            INNER JOIN risk_free rfr ON t.end_date = rfr.end_date 
            WHERE t.ret > -1
              AND rfr.ret > -1
            CONTEXT BY t.entity_id;
           
        treynor = SELECT t.entity_id, t.end_date, beta.benchmark_id,
                        ((1 + t.ret).cumprod().pow(12\iif(t.cnt<12, 12, t.cnt)) - (1 + t.rfr_ret).cumprod().pow(12\iif(t.cnt<12, 12, t.cnt))) / beta.beta AS treynor
                  FROM t
                  INNER JOIN beta AS beta ON t.entity_id = beta.entity_id AND t.end_date = beta.end_date
                  CONTEXT BY t.entity_id, beta.benchmark_id;

    } else if(trailing_month == 'ytd') {

        t = SELECT *, cumcount(entity_id) AS cnt 
            FROM ret t 
            INNER JOIN risk_free rfr ON t.end_date = rfr.end_date 
            WHERE t.ret > -1
              AND rfr.ret > -1
            CONTEXT BY t.entity_id, t.end_date.year();
           
        treynor = SELECT t.entity_id, t.end_date, beta.benchmark_id,
                        ((1 + t.ret).cumprod().pow(12\iif(t.cnt<12, 12, t.cnt)) - (1 + t.rfr_ret).cumprod().pow(12\iif(t.cnt<12, 12, t.cnt))) / beta.beta AS treynor
                  FROM t
                  INNER JOIN beta AS beta ON t.entity_id = beta.entity_id AND t.end_date = beta.end_date
                  CONTEXT BY t.entity_id, beta.benchmark_id, t.end_date.year();
    } else {

        win = trailing_month$INT;

        t = SELECT *, mcount(entity_id, win) AS cnt 
            FROM ret t 
            INNER JOIN risk_free rfr ON t.end_date = rfr.end_date 
            WHERE t.ret > -1
              AND rfr.ret > -1
            CONTEXT BY t.entity_id;
           
        treynor = SELECT t.entity_id, t.end_date, beta.benchmark_id,
                        ((1 + t.ret).mprod(win).pow(12\iif(t.cnt<12, 12, t.cnt)) - (1 + t.rfr_ret).mprod(win).pow(12\iif(t.cnt<12, 12, t.cnt))) / beta.beta AS treynor
                  FROM t
                  INNER JOIN beta AS beta ON t.entity_id = beta.entity_id AND t.end_date = beta.end_date
                  CONTEXT BY t.entity_id, beta.benchmark_id;
    }

    return treynor;
}

/*
 *    Jensen's Alpha
 *    TODO: the result is slightly off
 */
def cal_jensen(ret, bmk_ret, risk_free, beta, trailing_month) {

    if(trailing_month == 'incep') {
    	
        jensen = SELECT t.entity_id, t.end_date, t.ret.cumavg() - rfr.ret.cumavg() - beta.beta * (bmk.ret.cumavg() - rfr.ret.cumavg()) AS jensen, beta.benchmark_id
                 FROM ret t
                 INNER JOIN beta beta ON t.entity_id = beta.entity_id AND t.end_date = beta.end_date
                 INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND beta.benchmark_id = bmk.benchmark_id
                 INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                 WHERE t.ret > -1
                 CONTEXT BY t.entity_id, beta.benchmark_id;

    } else if(trailing_month == 'ytd') {

        jensen = SELECT t.entity_id, t.end_date, t.ret.cumavg() - rfr.ret.cumavg() - beta.beta * (bmk.ret.cumavg() - rfr.ret.cumavg()) AS jensen, beta.benchmark_id
                 FROM ret t
                 INNER JOIN beta beta ON t.entity_id = beta.entity_id AND t.end_date = beta.end_date
                 INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND beta.benchmark_id = bmk.benchmark_id
                 INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                 WHERE t.ret > -1
                 CONTEXT BY t.entity_id, beta.benchmark_id, t.end_date.year();

    } else {

        win = trailing_month$INT;

        jensen = SELECT t.entity_id, t.end_date, t.ret.mavg(win) - rfr.ret.mavg(win) - beta.beta * (bmk.ret.mavg(win) - rfr.ret.mavg(win)) AS jensen, beta.benchmark_id
                 FROM ret t
                 INNER JOIN beta beta ON t.entity_id = beta.entity_id AND t.end_date = beta.end_date
                 INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND beta.benchmark_id = bmk.benchmark_id
                 INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
                 WHERE t.ret > -1
                 CONTEXT BY t.entity_id, beta.benchmark_id;

    }

    return jensen;
}

/*
 *     Modigliani Modigliani Measure (M2)
 *     NOTE: M2 = sharpe * std(benchmark) + risk_free_rate
 *     NOTE: Java version is noncompliant-GIPS annulized number 
 */
def cal_m2(ret, benchmarks, bmk_ret, risk_free, trailing_month) {

    if(trailing_month == 'incep') {
    	
        m2 = SELECT t.entity_id, t.end_date,
                    iif(t.entity_id.cumcount() > 5, (t.ret - rfr.ret).cumavg() / t.ret.cumstd() * bmk.ret.cumstd() + rfr.ret.cumavg(), null) AS m2, bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id
             INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
             WHERE t.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id;

    } else if(trailing_month == 'ytd') {

        m2 = SELECT t.entity_id, t.end_date, 
                    iif(t.entity_id.cumcount() > 5, (t.ret - rfr.ret).cumavg() / t.ret.cumstd() * bmk.ret.cumstd() + rfr.ret.cumavg(), null) AS m2, bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id
             INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
             WHERE t.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id, t.end_date.year();

    } else {

        win = trailing_month$INT;

        m2 = SELECT t.entity_id, t.end_date,
                    iif(t.entity_id.mcount(win) > 5, (t.ret - rfr.ret).mavg(win) / t.ret.mstd(win) * bmk.ret.mstd(win) + rfr.ret.mavg(win), null) AS m2, bm.benchmark_id
             FROM ret t
             INNER JOIN benchmarks bm ON t.entity_id = bm.entity_id AND t.end_date = bm.end_date
             INNER JOIN bmk_ret bmk ON t.end_date = bmk.end_date AND bm.benchmark_id = bmk.benchmark_id
             INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
             WHERE t.ret > -1
             CONTEXT BY t.entity_id, bm.benchmark_id;

    }

    return m2;
}

/*
 *    Morningstar Return, Morningstar Risk-Adjusted Return
 *    
 *    TODO: Tax and loads are NOT taken care of
 *    TODO: Assume Chinese methodology using 3, 5, 10 as number of traling years
 *    TODO: need verify with reliable results
 *    
 *    NOTE: Morningstar methodology requires monthly return for calculation, so that "12" is hard-coded here
 * 
 * 
 */
def cal_ms_return(ret, risk_free, trailing_month) {

    win = trailing_month$INT;

    r = SELECT t.entity_id, t.end_date,
               ((1 + t.ret)\(1 + rfr.ret)).mprod(win).pow(12\(t.end_date.mmax(win) - t.end_date.mmin(win)))-1 AS ms_ret_a,
               (1 + t.ret).pow(-2).mavg(win).pow(-12/2)-1 AS ms_rar_a
        FROM ret t
        INNER JOIN risk_free rfr ON t.end_date = rfr.end_date
        WHERE t.ret > -1
        CONTEXT BY t.entity_id;

    return r;
}

/*
 *   有效主体-基准对应表
 * 
 *   @param benchmarks <TABLE>: entity-benchmark 的对应关系表 NEED COLUMNS: entity_id, end_date, benchmark_id
 *   @param end_day <DATE>:
 *   @param trailing_month <STRING>: 
 *   @param isEffectiveOnly <BOOL>: false时与Java相同； true：多了个限制条件：如果区间内有效基准数少于1/2，不做计算
 *                                  比如过去12个月中某BFI只出现2次，小于需要的6次，此BFI不参与 trailing 1 year 计算
 * 
 */
def get_effective_benchmarks(benchmarks, end_day, trailing_month, isEffectiveOnly) {

    min_pct = 0.5;

    if(isEffectiveOnly) {

      	t_dates = SELECT DISTINCT entity_id, end_date FROM benchmarks WHERE end_date <= end_day.month();

        if(trailing_month == 'incep') {

        	t = SELECT entity_id, end_date, end_date.cumcount() AS cnt FROM t_dates CONTEXT BY entity_id;
        	
        	bmk = SELECT bmk.* FROM benchmarks bmk
        	      INNER JOIN t ON bmk.entity_id = t.entity_id AND bmk.end_date = t.end_date
        	      CONTEXT BY bmk.entity_id, bmk.benchmark_id
        	      HAVING bmk.end_date.cumcount() >= t.cnt * min_pct;

        } else if(trailing_month == 'ytd') {

        	t = SELECT entity_id, end_date, end_date.cumcount() AS cnt FROM t_dates CONTEXT BY entity_id, end_date.year();
        	
        	bmk = SELECT bmk.* FROM benchmarks bmk
        	      INNER JOIN t ON bmk.entity_id = t.entity_id AND bmk.end_date = t.end_date
        	      CONTEXT BY entity_id, benchmark_id, end_date.year()
        	      HAVING bmk.end_date.cumcount() >= t.cnt * min_pct;
        } else {

        	win = trailing_month$INT;

        	t = SELECT entity_id, end_date, end_date.mcount(win) AS cnt FROM t_dates CONTEXT BY entity_id;
        	
        	bmk = SELECT bmk.* FROM benchmarks bmk
        	      INNER JOIN t ON bmk.entity_id = t.entity_id AND bmk.end_date = t.end_date
        	      CONTEXT BY entity_id, benchmark_id
        	      HAVING bmk.end_date.mcount(win) >= t.cnt * min_pct;
        }

    } else {

    	bmk = SELECT * FROM benchmarks WHERE end_date <= end_day.month();
    }

    return bmk;
}


/*
 *   Calculation for monthly indicators which need benchmark
 *   
 *   @param entity_info <TABLE>: xxx_information表，NEED COLUMNS entity_id, inception_date
 *   @param benchmark_mapping <TABLE>: entity-benchmark mapping table, NEED COLUMNS entity_id, end_date, benchmark_id
 *   @param end_day <DATE>;
 *   @param tb_ret <TABLE>: 收益表，NEED COLUMNS entity_id, price_dat, end_date, nav
 *   @param index_ret <TABLE>: historical benchmark return table, NEED COLUMNS fund_id, end_date, ret
 *   @param risk_free <TABLE>: historical risk free rate table, NEED COLUMNS fund_id, end_date, ret
 *   @param month <INT>: trailing x month
 *   
 *   @return: indicators table
 *    
 *   
 *   Create  20240904  模仿Java & python代码在Dolphin中实现，具体计算逻辑可能会有不同                          Joey
 *                     TODO: some datapoints require more data, we need a way to disable calculation for them
 *
 */
def cal_indicators_with_benchmark(entity_info, benchmark_mapping, end_day, tb_ret, index_ret, risk_free, month) {

    if(entity_info.isVoid() || entity_info.size() == 0 || benchmark_mapping.isVoid() || benchmark_mapping.size() == 0 ) return null;
    if(tb_ret.isVoid() || tb_ret.size() == 0 || index_ret.isVoid() || index_ret.size() == 0 || risk_free.isVoid() || risk_free.size() == 0 ) return null;

    // sorting for correct first() and last() value
    ret = SELECT * FROM tb_ret WHERE ret > -1 AND end_date <= end_day.month() ORDER BY entity_id, price_date;

    // get the effective benchmarks
    benchmarks = get_effective_benchmarks(benchmark_mapping, end_day, month, true);

    if(ret.isVoid() || ret.size() == 0 || benchmarks.isVoid() || benchmarks.size() == 0) return null;

    // alpha, beta
    alpha_beta = cal_alpha_beta(ret, benchmarks, index_ret, risk_free, month);

    // 胜率、跟踪误差、信息比率
    bmk_tracking = cal_benchmark_tracking(ret, benchmarks, index_ret, month);

    // 特雷诺
    treynor = cal_treynor(ret, risk_free, alpha_beta, month);

    // 詹森指数
    jensen = cal_jensen(ret, index_ret, risk_free, alpha_beta, month);

    // M2
    m2 = cal_m2(ret, benchmarks, index_ret, risk_free, month);

    // 上下行捕获率、收益
    capture_r = cal_capture_ratio(ret, benchmarks, index_ret, month);

    r = SELECT * FROM bmk_tracking a1
                 LEFT JOIN alpha_beta ON a1.entity_id = alpha_beta.entity_id AND a1.benchmark_id = alpha_beta.benchmark_id AND a1.end_date = alpha_beta.end_date
                 LEFT JOIN treynor ON a1.entity_id = treynor.entity_id AND a1.benchmark_id = treynor.benchmark_id AND a1.end_date = treynor.end_date
                 LEFT JOIN jensen ON a1.entity_id = jensen.entity_id AND a1.benchmark_id = jensen.benchmark_id AND a1.end_date = jensen.end_date
                 LEFT JOIN m2 ON a1.entity_id = m2.entity_id AND a1.benchmark_id = m2.benchmark_id AND a1.end_date = m2.end_date
                 LEFT JOIN capture_r ON a1.entity_id = capture_r.entity_id AND a1.benchmark_id = capture_r.benchmark_id AND a1.end_date = capture_r.end_date;

    // 年化各数据点
    // GIPS RULE: NO annulization for data less than 1 year
    plainAnnu = get_annulization_multiple('m');
    sqrtAnnu = sqrt(get_annulization_multiple('m'));

    r.addColumn(['alpha_a', 'jensen_a', 'track_error_a', 'info_a', 'm2_a'],
                [DOUBLE, DOUBLE, DOUBLE, DOUBLE, DOUBLE]);

    UPDATE r 
      SET alpha_a = alpha * iif(end_date - inception_date.month() > 12, plainAnnu, 1),
          jensen_a = jensen * iif(end_date - inception_date.month() > 12, plainAnnu, 1),
          track_error_a = track_error * iif(end_date - inception_date.month() > 12, sqrtAnnu, 1),
          info_a = info * iif(end_date - inception_date.month() > 12, sqrtAnnu, 1),
          m2_a = m2 * iif(end_date - inception_date.month() > 12, plainAnnu, 1)
    FROM ej(r, entity_info, 'entity_id');
    
    return r;
}

/*
 *   Monthly standard indicator calculation
 *   
 *   @param entity_info <TABLE>:
 *   @param benchmarks <TABLE>: entity-benchmark mapping table
 *   @param end_day <DATE>:
 *   @param tb_ret <TABLE>: 收益表，NEED COLUMNS entity_id, price_dat, end_date, nav 
 *   @param benchmark_ret <TABLE>: historical benchmark return table, NEED COLUMNS fund_id, end_date, ret
 *   @param risk_free <TABLE>: historical risk free rate table, NEED COLUMNS fund_id, end_date, ret
 *   @param month <STRING>:
 *   
 *   @return: indicators table
 *   
 *   
 *   Create  20240904  模仿Java & python代码在Dolphin中实现，具体计算逻辑可能会有不同                          Joey
 *
 */
def cal_indicators(entity_info, benchmarks, end_day, tb_ret, benchmark_ret, risk_free, month) {

    if(entity_info.isVoid() || entity_info.size() == 0 || benchmarks.isVoid() || benchmarks.size() == 0 ) return null;
    if(tb_ret.isVoid() || tb_ret.size() == 0 || benchmark_ret.isVoid() || benchmark_ret.size() == 0 || risk_free.isVoid() || risk_free.size() == 0 ) return null;

    // sorting for correct first() and last() value
    ret = SELECT * FROM tb_ret WHERE end_date <= end_day.month() ORDER BY entity_id, price_date;

    // 收益、标准差、偏度、峰度、最大回撤、VaR, CVaR、卡玛比率
    rtn = cal_basic_performance(entity_info, ret, month);

    // 夏普
    sharpe = cal_sharpe(ret, rtn, risk_free, month);

    // 整合后的下行标准差、欧米伽、索提诺、卡帕
    lpms = cal_omega_sortino_kappa(ret, risk_free, month);

    // 需要基准的指标们
    indicator_with_benchmark = cal_indicators_with_benchmark(entity_info, benchmarks, end_day, ret, benchmark_ret, risk_free, month);

    r = SELECT * FROM rtn a1
                 LEFT JOIN sharpe ON a1.entity_id = sharpe.entity_id AND a1.end_date = sharpe.end_date
                 LEFT JOIN lpms ON a1.entity_id = lpms.entity_id AND a1.end_date = lpms.end_date
                 LEFT JOIN indicator_with_benchmark bmk ON a1.entity_id = bmk.entity_id AND a1.end_date = bmk.end_date;


    // 晨星收益和风险
    if(month$STRING in ['36', '60', '120']) {
        ms = cal_ms_return(ret, risk_free, month);

        r = SELECT * FROM r LEFT JOIN ms ON r.entity_id = ms.entity_id AND r.end_date = ms.end_date;
    }

    // 年化各数据点
    // GIPS RULE: NO annulization for data less than 1 year
    plainAnnu = get_annulization_multiple('m');
    sqrtAnnu = sqrt(get_annulization_multiple('m'));

    r.addColumn(['std_dev_a', 'ds_dev_a', 'sharpe_a', 'sortino_a'],
                [DOUBLE, DOUBLE, DOUBLE, DOUBLE]);

    UPDATE r 
      SET std_dev_a = std_dev * iif(end_date - inception_date.month() > 12, sqrtAnnu, 1),
          ds_dev_a = ds_dev * iif(end_date - inception_date.month() > 12, sqrtAnnu, 1),
          sharpe_a = sharpe * iif(end_date - inception_date.month() > 12, sqrtAnnu, 1),
          sortino_a = sortino * iif(end_date - inception_date.month() > 12, sqrtAnnu, 1)
    FROM ej(r, entity_info, 'entity_id');
    
    return r;
}


/*
 *   Calculate trailing 6m, ytd, 1y, 2y, 3y, 4y, 5y, 10y and since inception datapoints
 *   
 *   @param: func <FUNCTION>: the calculation function
 *   @param: entity_info <TABLE>: basic information of entity, NEED COLUMNS entity_id, inception_date
 *   @param benchmarks <TABLE>: entity-benchmark mapping table
 *   @param: end_day <DATE>: 计算截止日期
 *   @param: ret <TABLE>: 收益表，NEED COLUMNS entity_id, price_dat, end_date, nav 
 *   @param bmk_ret <TABLE>: historical benchmark return table, NEED COLUMNS fund_id, end_date, ret
 *   @param risk_free <TABLE>: historical risk free rate table, NEED COLUMNS fund_id, end_date, ret
 *   @param periods <BOOL VECTOR>: 是否计算的区间向量，分别对应 incep, ytd, 6m, 1y, 2y, 3y, 4y, 5y, 10y
 *   
 * 
 */
def cal_trailing(func, entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate) {

    r_incep = null;
    r_ytd = null;
    r_6m = null;
    r_1y = null;
    r_2y = null;
    r_3y = null;
    r_4y = null;
    r_5y = null;
    r_10y = null;

    // incep
    r_incep = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, 'incep');

    // ytd
    r_ytd = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, 'ytd');

    // 6m
    r_6m = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, '6');

    // 1y
    r_1y = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, '12');

    // 2y
    r_2y = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, '24');

    // 3y
    r_3y = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, '36');

    // 4y
    r_4y = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, '48');

    // 5y
    r_5y = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, '60');

    // 10y
    r_10y = func(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate, '120');

    return r_incep, r_ytd, r_6m, r_1y, r_2y, r_3y, r_4y, r_5y, r_10y;
}


/*
 *   Calculate trailing ytd, 6m, 1y, 2y, 3y, 4y, 5y, 10y and since inception standard indicators
 *   
 *   @param: entity_info <TABLE>: basic information of entity, NEED COLUMNS entity_id, inception_date
 *   @param benchmarks <TABLE>: entity-benchmark mapping table
 *   @param: ret <TABLE>: 收益表，NEED COLUMNS entity_id, price_dat, end_date, nav 
 *   @param: end_day <DATE>: 计算截止日期
 *   @param bmk_ret <TABLE>: historical benchmark return table, NEED COLUMNS fund_id, end_date, ret
 *   @param risk_free <TABLE>: historical risk free rate table, NEED COLUMNS fund_id, end_date, ret
 * 
 */
def cal_trailing_indicators(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate) {
	
    return cal_trailing(cal_indicators, entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate);

}

/*
 *   Calculate trailing 6m, ytd, 1y, 2y, 3y, 4y, 5y, 10y and since inception bfi indicators
 *   
 *   @param: entity_info <TABLE>: basic information of entity, NEED COLUMNS entity_id, inception_date
 *   @param benchmarks <TABLE>: entity-benchmark mapping table
 *   @param: ret <TABLE>: 收益表，NEED COLUMNS entity_id, price_dat, end_date, nav 
 *   @param: end_day <DATE>: 计算截止日期
 *   @param bmk_ret <TABLE>: historical benchmark return table, NEED COLUMNS fund_id, end_date, ret
 *   @param risk_free <TABLE>: historical risk free rate table, NEED COLUMNS fund_id, end_date, ret
 *   
 * 
 */
def cal_trailing_bfi_indicators(entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate) {

    return cal_trailing(cal_indicators_with_benchmark, entity_info, benchmarks, end_day, tb_ret, bmk_ret, risk_free_rate);
}


/*
 *   通用月度指标计算
 * 
 *   @param entity_type <STRING>:
 *   @param indicator_type <STRING>: PBI, BFI
 *   @param monthly_returns <TABLE>: NEED COLUMN: entity_id, end_date, price_date, nav, ret
 * 
 *   @return <DICT TABLE>: ['PBI-INCEP', 'PBI-YTD', 'PBI-6M', 'PBI-1Y', 'PBI-2Y', 'PBI-3Y', 'PBI-4Y', 'PBI-5Y', 'PBI-10Y', 'MS-3Y', 'MS-5Y', 'MS-10Y']
 * 
 */
def cal_monthly_indicators(entity_type, indicator_type, monthly_returns) {

    if(find(['MF', 'HF', 'PF'], entity_type) < 0) return null;

    if(monthly_returns.isVoid() || monthly_returns.size() < 1) return null;

    oldest_date = EXEC price_date.min() FROM monthly_returns;

    v_entity_ids = (SELECT DISTINCT entity_id FROM monthly_returns).entity_id;
    
    entity_info = get_entity_info(entity_type, v_entity_ids);
    
    if(entity_info.isVoid() || entity_info.size() == 0) { return null };
    
    end_day = today();

    // 取基金和基准的对照表
    if(indicator_type == 'BFI') {

        benchmark = SELECT fund_id AS entity_id, end_date.temporalParse('yyyy-MM') AS end_date, factor_id AS benchmark_id 
                    FROM get_fund_bfi_factors(v_entity_ids, oldest_date.temporalFormat('yyyy-MM'), end_day.temporalFormat('yyyy-MM'));

    } else {
        // 主基准， 对应 xxx_info 中的 primary_benchmark_id
        benchmark = SELECT entity_id, end_date, iif(benchmark_id.isNull(), 'IN00000008', benchmark_id) AS benchmark_id 
                    FROM get_entity_primary_benchmark(entity_type, v_entity_ids, oldest_date.temporalFormat('yyyy-MM'), end_day.temporalFormat('yyyy-MM')) ;
    	
    }

    // 取所有出现的基准月收益
    bmk_ret = get_benchmark_return(benchmark, end_day);

    if(bmk_ret.isVoid() || bmk_ret.size() == 0) { return null; }

    // TODO: risk free指数月收益存在fund_performance表，所以先将就用 fund_id 表示。之后统一改为更准确的名字
    risk_free_rate = SELECT entity_id AS fund_id, temporalParse(end_date, 'yyyy-MM') AS end_date, ret FROM get_risk_free_rate(oldest_date, end_day);

    if(risk_free_rate.isVoid() || risk_free_rate.size() == 0) { return null; }

    // 指标计算
    if(indicator_type == 'BFI') {

        t0 = cal_trailing_bfi_indicators(entity_info, benchmark, end_day, monthly_returns, bmk_ret, risk_free_rate);

        v_table_name = ['BFI-INCEP', 'BFI-YTD', 'BFI-6M', 'BFI-1Y', 'BFI-2Y', 'BFI-3Y', 'BFI-4Y', 'BFI-5Y', 'BFI-10Y'];

    } else {

        t0 = cal_trailing_indicators(entity_info, benchmark, end_day, monthly_returns, bmk_ret, risk_free_rate);

        v_table_name = ['PBI-INCEP', 'PBI-YTD', 'PBI-6M', 'PBI-1Y', 'PBI-2Y', 'PBI-3Y', 'PBI-4Y', 'PBI-5Y', 'PBI-10Y'];
    }

    return dict(v_table_name, t0);
	
}
/*
 *   Calculate historcial fund trailing indicators 
 * 
 *   @param fund_type <STRING>: MF, HF
 *   @param fund_ids <STRING>: 逗号和单引号分隔的fund_id
 *   @param end_day <DATE>: 要计算的日期
 *   @param isFromNav <BOOL>: 用净值实时计算还是从表中取月收益
 *   @param isFromSQL <BOOL>: TODO: 从MySQL还是本地DolphinDB取净值/收益数据
 *   
 *   @return <DICT TABLE>: ['PBI-INCEP', 'PBI-YTD', 'PBI-6M', 'PBI-1Y', 'PBI-2Y', 'PBI-3Y', 'PBI-4Y', 'PBI-5Y', 'PBI-10Y', 'MS-3Y', 'MS-5Y', 'MS-10Y']
 *   
 * 
 *   Example: cal_fund_indicators('HF', "'HF000004KN','HF000103EU','HF00018WXG'", 2024.06.28, true);
 * 
 */
def cal_fund_indicators(fund_type, fund_ids, end_day, isFromNav) {

    very_old_date = 1990.01.01;

    if(isFromNav == true) {

        // 从净值开始计算收益
        tb_ret = SELECT * FROM cal_fund_monthly_returns(fund_type, fund_ids, true) WHERE price_date <= end_day;
        tb_ret.rename!(['fund_id', 'cumulative_nav'], ['entity_id', 'nav']);

    } else {

        // 从fund_performance表里读月收益
        tb_ret = get_monthly_ret(fund_type, fund_ids, very_old_date, end_day, true);
    
        v_end_date = tb_ret.end_date.temporalParse('yyyy-MM');
        tb_ret.replaceColumn!('end_date', v_end_date);

    }

    if(tb_ret.isVoid() || tb_ret.size() == 0) { return null; }

    // 标准的指标
    d = cal_monthly_indicators(fund_type, 'PBI', tb_ret);

    return d;

}

/*
 *   Calculate historcial fund trailing BFI indicators 
 * 
 *   @param fund_type <STRING>: MF, HF
 *   @param fund_ids <STRING>: 逗号和单引号分隔的fund_id
 *   @param end_day <DATE>: 要计算的日期
 *   @param isFromNav <BOOL>: 用净值实时计算还是从表中取月收益
 *   @param isFromSQL <BOOL>: TODO: 从MySQL还是本地DolphinDB取净值/收益数据
 *   
 *   @return <DICT TABLE>: ['BFI-INCEP', 'BFI-YTD', 'BFI-6M', 'BFI-1Y', 'BFI-2Y', 'BFI-3Y', 'BFI-4Y', 'BFI-5Y', 'BFI-10Y']
 *   
 * 
 *   Example: cal_fund_bfi_indicators('MF', "'MF00003PW2', 'MF00003PW1', 'MF00003PXO'", 2024.08.31, true);
 * 
 */
def cal_fund_bfi_indicators(fund_type, fund_ids, end_day, isFromNav) {

    very_old_date = 1990.01.01;

    if(isFromNav == true) {

        // 从净值开始计算收益
        tb_ret = SELECT * FROM cal_fund_monthly_returns(fund_type, fund_ids, true) WHERE price_date <= end_day;
        tb_ret.rename!(['fund_id', 'cumulative_nav'], ['entity_id', 'nav']);

    } else {

        // 从fund_performance表里读月收益
        tb_ret = get_monthly_ret(fund_type, fund_ids, very_old_date, end_day, true);
        tb_ret.rename!(['fund_id'], ['entity_id']);
    
        v_end_date = tb_ret.end_date.temporalParse('yyyy-MM');
        tb_ret.replaceColumn!('end_date', v_end_date);

    }

    if(tb_ret.isVoid() || tb_ret.size() == 0) { return null; }

    // BFI指标
    d = cal_monthly_indicators(fund_type, 'BFI', tb_ret);

    return d;
}

/*
 *   Calculate historcial portfolio trailing indicators 
 *  
 *  @param portfolio_ids <STRING>: comma-delimited portfolio ids
 *  @param end_day <DATE>: the date
 *  @param cal_method <INT>: calculate based on cumulative nav (1) or nav (2)
 *  @param isFromNav <BOOL>: calculate returns from NAV on-the-fly (true) or get from monthly return table (false)
 *  
 *  Example: cal_portfolio_indicators('166002,166114', 2024.08.31, 1, true);
 * 
 */
def cal_portfolio_indicators(portfolio_ids, end_day, cal_method, isFromNav) {

    very_old_date = 1990.01.01;
    start_month = very_old_date.month();

    portfolio_info = get_portfolio_info(portfolio_ids);

    if(portfolio_info.isVoid() || portfolio_info.size() == 0) { return null };
    
    portfolio_info.rename!('portfolio_id', 'entity_id');

    if(isFromNav == true) {
        // 从净值开始计算收益
        tb_raw_ret = SELECT * FROM cal_portfolio_nav(portfolio_ids, very_old_date, cal_method) WHERE price_date <= end_day;

        if(tb_raw_ret.isVoid() || tb_raw_ret.size() == 0) return null;

        // funky thing is you can't use "AS" for the grouping columns?
        tb_ret = SELECT portfolio_id, price_date.month(), price_date.last() AS price_date, (1+ret).prod()-1 AS ret, nav.last() AS nav
                 FROM tb_raw_ret
                 WHERE price_date <= end_day
                 GROUP BY portfolio_id, price_date.month();
        tb_ret.rename!(['portfolio_id', 'month_price_date'], ['entity_id', 'end_date']);

    } else {
        // 从pf_portfolio_performance表里读月收益
        tb_ret = get_monthly_ret('PF', portfolio_ids, very_old_date, end_day, true);
        tb_ret.rename!(['portfolio_id'], ['entity_id']);
        v_end_date = tb_ret.end_date.temporalParse('yyyy-MM');
        tb_ret.replaceColumn!('end_date', v_end_date);
    }

    if(tb_ret.isVoid() || tb_ret.size() == 0) return null;

    // 混合因子做基准，同SQL保持一致
    t_dates = table(start_month..end_day.month() AS end_date);
    primary_benchmark = SELECT ei.entity_id, dt.end_date, 'FA00000VNB' AS benchmark_id 
                        FROM portfolio_info ei JOIN t_dates dt
                        WHERE dt.end_date >= ei.inception_date.month();

    if(primary_benchmark.isVoid() || primary_benchmark.size() == 0) { return null; }

    // 取所有出现的基准月收益
    bmk_ret = get_benchmark_return(primary_benchmark, end_day);

    if(bmk_ret.isVoid() || bmk_ret.size() == 0) { return null; }

    // TODO: risk free指数月收益存在fund_performance表，所以先将就用 fund_id 表示。之后统一改为更准确的名字
    risk_free_rate = SELECT entity_id AS fund_id, temporalParse(end_date, 'yyyy-MM') AS end_date, ret FROM get_risk_free_rate(very_old_date, end_day);

    if(risk_free_rate.isVoid() || risk_free_rate.size() == 0) { return null; }

    t0 = cal_trailing_indicators(portfolio_info, primary_benchmark, end_day, tb_ret, bmk_ret, risk_free_rate);

    v_table_name = ['PBI-INCEP', 'PBI-YTD', 'PBI-6M', 'PBI-1Y', 'PBI-2Y', 'PBI-3Y', 'PBI-4Y', 'PBI-5Y', 'PBI-10Y'];
    
    return dict(v_table_name, t0);

}


/*
 *   Calculate historcial portfolio trailing BFI indicators 
 *  
 *  @param portfolio_ids <STRING>: comma-delimited portfolio ids
 *  @param end_day <DATE>: the date
 *  @param cal_method <INT>: calculate based on cumulative nav (1) or nav (2)
 *  @param isFromNav <BOOL>: calculate returns from NAV on-the-fly (true) or get from monthly return table (false)
 *  
 *  TODO: intergrate with cal_portfolio_indicators
 *   
 *  Example: cal_portfolio_bfi_indicators('166002,166114', 2024.08.31, 1, true);
 * 
 */
def cal_portfolio_bfi_indicators(portfolio_ids, end_day, cal_method, isFromNav) {

    very_old_date = 1990.01.01;

    start_month = 1990.01M;

    portfolio_info = get_portfolio_info(portfolio_ids);

    if(portfolio_info.isVoid() || portfolio_info.size() == 0) { return null };
    
    portfolio_info.rename!('portfolio_id', 'entity_id');

    if(isFromNav == true) {
        // 从净值开始计算收益
        tb_raw_ret = SELECT * FROM cal_portfolio_nav(portfolio_ids, very_old_date, cal_method) WHERE price_date <= end_day;

        if(tb_raw_ret.isVoid() || tb_raw_ret.size() == 0) return null;
       
        // funky thing is you can't use "AS" for the grouping columns?
        tb_ret = SELECT portfolio_id, price_date.month(), price_date.last() AS price_date, (1+ret).prod()-1 AS ret, nav.last() AS nav
                 FROM tb_raw_ret
                 WHERE price_date <= end_day
                 GROUP BY portfolio_id, price_date.month();
        tb_ret.rename!(['portfolio_id', 'month_price_date'], ['entity_id', 'end_date']);

    } else {
        // 从pf_portfolio_performance表里读月收益
        tb_ret = get_monthly_ret('PF', portfolio_ids, very_old_date, end_day, true);
        tb_ret.rename!(['portfolio_id'], ['entity_id']);
        v_end_date = tb_ret.end_date.temporalParse('yyyy-MM');
        tb_ret.replaceColumn!('end_date', v_end_date);
    }

    if(tb_ret.isVoid() || tb_ret.size() == 0) return null;

    // 取组合和基准的对照表
    bfi_benchmark = SELECT portfolio_id AS entity_id, end_date.temporalParse('yyyy-MM') AS end_date, factor_id AS benchmark_id 
                    FROM get_portfolio_bfi_factors(portfolio_ids, start_month.temporalFormat('yyyy-MM'), end_day.temporalFormat('yyyy-MM'));

    if(bfi_benchmark.isVoid() || bfi_benchmark.size() == 0) { return null; }

    bmk_ret = get_benchmark_return(bfi_benchmark, end_day);

    if(bmk_ret.isVoid() || bmk_ret.size() == 0) { return null; }

    // TODO: risk free指数月收益存在fund_performance表，所以先将就用 fund_id 表示。之后统一改为更准确的名字
    risk_free_rate = SELECT entity_id AS fund_id, temporalParse(end_date, 'yyyy-MM') AS end_date, ret FROM get_risk_free_rate(very_old_date, end_day);

    if(risk_free_rate.isVoid() || risk_free_rate.size() == 0) { return null; }
    
    t0 = cal_trailing_bfi_indicators(portfolio_info, bfi_benchmark, end_day, tb_ret, bmk_ret, risk_free_rate);

    v_table_name = ['PBI-INCEP', 'PBI-YTD', 'PBI-6M', 'PBI-1Y', 'PBI-2Y', 'PBI-3Y', 'PBI-4Y', 'PBI-5Y', 'PBI-10Y'];
    
    return dict(v_table_name, t0);

}