Alpha Type:
- Delay (delay-1, delay-0, intraday)
- Universe (Top3000,Top 2000,Top1000, Top500)
Please start with US delay-1, Top 2000
- PnL
daily_pnl = sum of (position * daily_return)
(the daily return is per instrument (today’s close / yesterday’s close) – 1.0)
- Information Ratio (IR)
Prediction ability of a model: mean(daily_pnl)/std(daily_pnl).
Sharpe ratio is the annual IR: IR * sqrt(252)
(the 252 represents the estimated number of trading days in a year)
- Turnover (tvr)
Value traded / value held
- Drawdown
Percentage of the largest loss
DATA NAME DESCRIPTION
open Daily open price
close Daily close price
high Daily high price
low Daily low price
vwap Daily volume weighted price
volume Daily volume
returns Daily returns
adv20 Average daily volume in the past 20 days
sharesout Daily outstanding shares
OPERATOR DESCRIPTION
+, -, *, /, ^ Add, Subtract, Multiply, Divide, Power
<, <=, >, >=, ==, != Logic comparison operators
||, &&, ! Logical OR, AND, Negation
cond ? expr1 : expr2 If cond is true, expr1; else expr2. For example, close < open ? close : open
Rank(x) Rank the values in vector x and the return values are between 0.0 and 1.0
Min (x, y) Parallel minimum of vectors x and y (similar to the pmin function in R)
Max(x, y) Parallel maximum of vectors x and y (similar to the pmax function in R)
StdDev(x, n) Standard deviation of the values in vector x for the past n days.
Note that n must be less than *256
Correlation(x, y, n) Correlation of the values in vectors x and y for the past n days. Note that n
must be less than 256
Sum(x, n) Sum of the values in vector x for the past n days. Note that n must be less
than 256
Covariance(x, y, n) Covariance of the values in vectors x and y for the past n days. Note that n
must be less than 256