Skip to main content

Table 14 Cross-sectional return regressions with MAX and skewness

From: Lottery-like preferences and the MAX effect in the cryptocurrency market

MAX

BETA

SIZE

PRICE

MOM

REV

ILLIQ

TSKEW

ISKEW

E(TSKEW)

       

0.009

  
       

(2.004)**

  
        

0.008

 
        

(1.874)*

 

0.424

0.064

− 0.003

0.005

0.035

0.031

1.612

− 0.006

  

(7.298)***

(1.710)*

(− 1.226)

(1.160)

(1.841)*

(2.696)***

(0.186)

(− 0.635)

  

0.452

0.049

− 0.002

0.003

0.089

0.071

− 0.729

 

− 0.009

 

(6.069)***

(1.249)

(− 0.716)

(0.802)

(1.978)**

(2.552)***

(− 0.265)

 

(− 0.894)

 
         

0.038

         

(1.747)*

0.411

0.053

− 0.004

0.009

0.026

0.025

1.145

  

0.031

(6.973)***

(1.265)

(− 1.410)

(1.586)

(2.587)***

(2.650)***

(0.850)

  

(1.375)

  1. Each week from January 2014 to September 2020, we conduct a cryptocurrency-level cross-sectional regression of the return in that week on lagged explanatory variables, which are lagged TSKEW, ISKEW, E(TSKEW), and MAX as well as six other control variables - BETA, SIZE, PRICE, MOM, REV and ILLIQ. TSKEW refers to the total skewness measured for past month using daily returns. ISKEW is the skewness of regression residuals. E(TSKEW) refers to the expected total skewness fitted each week throughout the cross-sectional regressions. MAX refers to the maximum daily return within a month. BETA is the market beta calculated as the sum of coefficients on lag, current and lead returns on market portfolio. SIZE equals the natural logarithm of the cryptocurrency’s market capitalization by the end of previous week. PRICE is the natural logarithm of one plus the cryptocurrency’s price. MOM stands for momentum that equals the cumulative return over the previous three weeks skipping past week. REV stands for short-term reversal and equals the return in the past week. ILLIQ refers to Amihud illiquidity measure, which is calculated by the ratio of absolute daily cryptocurrency return to its mean dollar trading volume. Both dependent and independent variables are winsorized at 5% level from upper and lower tails. Table reports the time-series averages of the cross-sectional regression coefficients and corresponding t-statistics. Newey-West (1987) adjusted t-statistics are presented in parentheses
  2. ***, ** and * denote significance at 1%, 5% and 10%, respectively