速寫 Day51

來源：赤倉

在6/23-6/26，
跟大學同學去宜花畢業旅行。
真的是累死了，
一直唱歌喉嚨好痛。
然後又每天聊天聊到一兩點。

因為疫情的關係大家好久沒有聚在一塊，
現在好不容易聚在一起，
反而已經是最後一次了。
畢業之際，雖然有許多不錯的回憶，
但心中真的還有許多滿滿的不捨。

之後照片整理整理後再寫一篇文章吧！

現在真的沒辦法。
一回來就好多東西要忙==
例如實習前要研讀的許多教程資料；
7/3要準備考的日檢N1；
或是日本交換線上課程準備要考期末；
更甚至最近家裡裝潢完成需要開始搬東西。
~~或是要再去新竹領畢業證書之類的==~~

總之就是很多東西。

–

然後最近水電工、第四台、裝冰箱等等工人，
一直來，
我最近的下午完全沒辦法念書。

雖然覺得他們真的很累很辛苦，
可是……
我真的要開天窗啦！
啊啊啊啊啊啊啊！

然後還要碩士註冊前體檢，
插座還因為水電師傅接錯，
印表機不小心燒掉？

太多事了吧！
我真的要崩潰了喔？

日文總複習

すら ─ 就連
だに ─ 光是
たにとも…ない ─ 就連1次、1秒…都沒有
がてら ─ 順便
かたがた ─ (正式) 順便
かたわら ─ 另一方面
ずくめ ─ 滿是
まみれ ─ 渾身(負面)
が早いか ─ 一…就…
や否や ─ 一…就…
そばから ─ 才剛…就…(負面)
～なり ─ 一…的同時
～んがために ─ 為了…
～とばかりに ─ 彷彿機會來了、像是在說…的樣子
～とばかりに ─ (程度很大)
（に）～かのごとく（き） ─ 彷彿(強調，但沒實際做！)
てしかるべき ─ (不滿)本來就應該
べからず ─ 不能、禁止
（ある）まじき ─ 身為…(的立場)是不能…的
Aもさることながら…Bも ─ A不用說，但B也非常…(強調)
はおろか ─ 不用說…就連(負面)
にもまして ─ 比…更(以前)
に越したことはない ─ 雖然這樣做比較好，但…
には及ばない ─ 不必
にはあたらない ─ 不值得
までもない ─ 不必特地
を皮切りに ─ …為開始
を限りに ─ 1. …為最後 2. 盡全力
をもって ─ 就是で(通常接時間後面)
ばこそ ─ 正因為
ならでは ─ 特有的
あっての ─ 有著…(才有的)
をおいて…ない ─ 除了…其他人都不能
～なり～なり ─ 都可以喔
～つ～つ ─ 重複
～であれ～であれ ─ 不管、無論
～といい～といい ─ 不管A還是B，全部都
～うが～うが ─ 不管A還是B(同種類)，都與我無關
～うと～まいが ─ 不管A還是B(正反面)，都與我無關
～うが～なかろうか ─ 不管A還是B，都可以…
たところで ─ 就算(現在開始…) 也沒辦法…
いかんだ ─ 根據
をよそに ─ 不顧 (周圍的意見)
をものともせず ─ 不顧 (糟糕的狀況)
たるもの ─ 身為(…立場) 必須…
ともあろうものが ─ (批評高位階的人)
ゆえに ─ 因為
と思いきや ─ 原以為…卻 (正面)
といえども ─ 原以為…但 (不能)
からある ─ (數量)以上
からなる ─ 組成
めく／めいた ─ 春、謎題、說教
ぶる ─ 假裝、刻意
てならない ─ 非常(殘念、擔心)
てやまない ─ 無比(感激、願望)
極まりない ─ 極為
の至り ─ 無比
を禁じ得ない ─ 不禁
に堪えない ─ 忍不住
を余儀なくされる ─ 不得不
ずにはいられない ─ 不由得、不得不

稍微的筆記

Deep Learning with Python - Chollet

Optimization: the process of adjusting a model to get the best performance possible on the training data.
Generalization: how well the trained model performs on data it has never seen before.
Regulization: the process of fighting overfitting.

Preventing overfitting:

Get more training data.
Reduce the capacity of the network.
Add weight regularization.
Add dropout. (introducing noice)

–

Finding Alphas: A Quantitative Approach to Building Trading Strategies - Tulchinsky et al.

Introduction to Alpha Design

Alpha design: A forecast of the return on each of the financial securities, to make predictions about future movements.
Data \(\rightarrow\) (e.g. historical proces, volume, shares traded, etc.) \(\rightarrow\) Alpha \(\rightarrow\) Price Prediction
The existence of alphas is a result of the imperfect flow of information among market participants with competing objectives.
Rules in trading are called alphas.
Cutting losses: Letting go of rules that no longer work.
Applying all rules simultaneously is the key to success. No single rule can ever be relied upon completely, so it’s necessary to come up with a strategy for using rules simultaneously. (Some rule should be cut immediately when emergency happens)

Design and Evaluation

Alpha design

Typically, an alpha is utilized as a component of a trading strategy, which converts the alpha’s predictions into actual trading decisions. The strategy also considers practical issues such as transaction costs and portfolio risk.

Alpha Types

1. Intraday alphas: rebalanced during trading hours of the day.
  (1) Rebalance at each interval. e.g. 1 min/5 min.
  (2) Rebalance triggered by some events such as ticks/orders/fills or predefined events.
1. Daily alphas: rebalance every day.
  (1) Delay N: use data of N days ago.
  (2) Delay 0 snapshot: use data before a certain time snapshot.
  (3) MOO/MOC: alphas trade at market open/close auction session.
1. Weekly/Monthly alphas: rebalanced every week/month.

Development of an alpha:

Using public information by searching signals/patterns:
(1) Price/Volume.
(2) Fundamentals.
(3) Macro data.
(4) Text.
(5) Multimedia. (Extract text information from the video/audio).
Other information from which alphas are not derived directly but may be used to improve the performance of alphas:
(1) Risk factor models. (risk mitigation.)
(2) Relationship models. (lead or lag due to correlation.)
(3) Microstructure models. (real trading.)

Practical alpha evaluaion

Run a simulation (backtest) to measure characteristics:
(1) Information ratio: The mean of the alpha’s returns divided by the standard deviation of the returns.
-> Measure Consistency.
-> which combined with the length of the observation peroid can be used to determine how confident we are to determine that the alpha is not some random noise.
(2) Margin: The amount of profit made by the alpha in the simulation divided by the amount of trading that was done.
-> Measure Sensitivity to transaction costs.
(3) Uniqueness: The maximum correlation of the alpha to others in the pool of alphas.
-> Lower correlation means more valuable alpha.

Develop an alpha:

1. Collect information.
1. Come up with an idea.
1. Translate into a mathematical expression. (Positive alpha \(\rightarrow\) long position, vice versa.)
1. Transform the raw expression by applying operations.
1. Final robust alpha.
1. Translate into positions in a financial instrument.
1. Check for robustness, such as High In-sample / Good Out-of-sample Information Ratio, less fitting, small drawdown, etc.

Improving alpha:

Risk mitigation: requiring our portfolios to be always long/short balanced within each industry.
Relative size is more accurate as a predictor: Concept of rank.
Using Decay: averaging alpha signal within a time window. (To reduce turnover.)

今日其他進度：

日文N1文法、N1題目
一堆的動畫

我會繼續努力的。