notes by PDI

  Twitterの分析を始めました。 アカウントの投稿頻度などを集計して見やすいチャートにします。 ご希望の集計対象があればリクエストにお答えします サンプルは以下のとおりです。 Started Twitter analysis. pull together certain subject's tweets and visualize them. provide upon request. please see below sample figures

  疑問 これ とかいろいろなところでレバレッジが高すぎると逆にリターンが減るという話が言われているのだけれども、理屈はともかく実際に過去の値動きで検証したらどうだろう？ 方法 2011年以降直近までの日経平均のデータをとりレバレッジを変えてシミュレーションを行う。 手数料や税金は考慮しない。 結果 下の図のとおりレバレッジを上げすぎると実際の数字で確認してもリターンが下がる レバレッジ1倍だとリターンは+140%程度 レバレッジを引き上げるにつれてリターンも増加していくが2倍から2.5倍くらいでピークを迎えてそれ以降低下していく 4倍にすると+80%程度になりレバレッジ1倍も下回る

  this post is the sequel to the previous post ; there I grouped S&P 500's daily performance by dates from previous trading day; and found '2days' group had higher return. below is the risk and return plot of 4 group (1day to 4 days). as mentioned, '2days' group has higher return than other groups then distribution of return of '2days' group is shown below chart. added '1day' group for the comparison. difference of average performance came from performance around zero percent. 2Days group has more dates with performance between '0% to 1%' than 1day group. 1Day group has more dates with performance between '-1% to 0%' than 2days group. these differences apparently led to 2Days group's higher performance.

 Purpose: News flow continuously, events happen regardless markets' schedule; For example, markets digest '1 day information from Tuesday to Friday; On Monday markets have to chew up 3 days information (Saturday, Sunday, and Monday); As such, price movements on Monday must be bigger than other weekdays, is my idea. Method: to see this I go over S&P 500 index from 1970; Sort daily (business day) close price change into groups by calendar day difference; calculate average, and standard deviation for each group. Results: Below figure shows the summary; Standard deviation increased along with number of days up to 3. In date-wise, 2 days standard deviation should be square root of 2 times bigger than 1 days deviation, but not that much (1.09% to 1.03%, 1.06 x); and 3 days should be square root of 3 times bigger, but the results were 1.25% to 1.03%, 1.21; There are less events and economic news may be behind this; below is the chart for standard deviation. Average index change va...