prepare equity simulation data

release_version.txt

@@ -1 +1 @@
-1.8.6,crypto_exch_stats docker run
+1.8.7,prepare equity simulation data

research/prepare_eqt_simdata.sh

@@ -0,0 +1,135 @@
+#!/bin/bash
+
+usage() {
+  echo "Usage: $0  -S <symbols> -d <YYYYMMDD Date> [-O <output dir (./) >]"
+  exit 1
+}
+
+# --------------------- Settings
+SourceHost=cloud21.cvtt.vpn
+SourceRootDir=/opt/store/cvtt/md_archive/equity/alpaca_md
+# --------------------- Settings
+
+
+is_business_day() {
+    dt=${1}
+    date=$(date -d "${dt}" +"%Y-%m-%d")
+
+    CalendarURL=http://cloud23.cvtt.vpn:8000/api/v1/markets/hours?mic=XNYS
+    URL="${CalendarURL}&start=${date}&end=${date}"
+    open_time=$(curl -s "${URL}" | jq '.[] | .open_time')
+    if [ -n  "${open_time}" ]; then
+        return 0
+    else
+        return 1
+    fi
+
+}
+export -f is_business_day
+
+
+while getopts ":d:S:O:" opt; do
+  case ${opt} in
+    d )
+      Date=$OPTARG
+      ;;
+    S )
+      SymList=$OPTARG
+      ;;
+    O )
+      OutputDir=$OPTARG
+      ;;
+    \? )
+      echo "Invalid option: -$OPTARG" >&2
+      usage
+      ;;
+    : )
+      echo "Option -$OPTARG requires an argument." >&2
+      usage
+      ;;
+  esac
+done
+
+if [ -z ${SymList}  ] ; then
+    echo "symbols are not specified"
+    usage
+fi
+if [ -z ${Date}  ] ; then
+    echo "date is not specified"
+    usage
+fi
+if [ -z ${OutputDir}  ] ; then
+    OutputDir=.
+fi
+mkdir -p ${OutputDir}
+
+if ! is_business_day ${Date}; then
+   echo "${Date} is not business day"
+   usage
+fi
+
+OLD_IFS=${IFS}
+IFS=","
+read -ra Symbols <<< "${SymList}"
+IFS=${OLD_IFS}
+
+echo "Date=${Date} Symbols=${Symbols[@]} OutputDir=${OutputDir}"
+echo Getting data from ${DataHost} ...
+
+year=$(date -d ${Date} +"%Y")
+for sym in ${Symbols[@]}; do
+    inst_id="STOCK-${sym}"
+    capital=${sym:0:1}
+    SourceDir="${SourceRootDir}/${year}/${capital}/${sym}"
+    SourceHbarFile="${SourceDir}/${Date}.${sym}.alpaca_1m_bars.db.gz"
+    SourceQatFile="${SourceDir}/${Date}.${sym}.alpaca_qat.db.gz"
+
+    for src_file in ${SourceHbarFile} ${SourceQatFile}; do
+        Cmd="rsync -ahv"
+        Cmd+=" ${SourceHost}:${src_file}"
+        Cmd+=" $OutputDir/"
+        echo ${Cmd}
+        eval ${Cmd}
+    done
+done
+
+Cmd="(cd ${OutputDir} && gunzip *.db.gz)"
+echo ${Cmd}
+eval ${Cmd}
+
+ResultDbFile="${OutputDir}/${Date}.alpaca_sim_md.db"
+echo "Creating Result Database File ${ResultDbFile}"
+sym=${Symbols[0]}
+for db_file in ${OutputDir}/${Date}.${sym}.alpaca_1m_bars.db ${OutputDir}/${Date}.${sym}.alpaca_qat.db ; do
+    Cmd="sqlite3"
+    Cmd+=" ${db_file} .schema"
+    Cmd+=" | sqlite3 ${ResultDbFile}"
+    echo ${Cmd}
+    eval ${Cmd}
+done
+
+set -f # not to expand *
+
+for sym in ${Symbols[@]}; do
+   src_hbar_db=${OutputDir}/${Date}.${sym}.alpaca_1m_bars.db
+   src_qat_db=${OutputDir}/${Date}.${sym}.alpaca_qat.db
+   for srcdb in ${src_hbar_db} ${src_qat_db} ; do
+
+       tables=$(sqlite3 ${srcdb} .tables)
+       for table in $(sqlite3 ${srcdb} .tables) ; do
+           echo "Loading from ${srcdb}.${table} ..."
+           sqlite3  ${ResultDbFile} <<EOF
+ATTACH '${srcdb}' AS source_db;
+BEGIN;
+INSERT INTO ${table} SELECT * FROM source_db.${table};
+COMMIT;
+DETACH source_db;
+EOF
+       done
+       Cmd="rm ${srcdb}"
+       echo ${Cmd}
+       eval ${Cmd}
+   done
+done
+
+echo Done $0 ${*}