Ziyu W.

Experience

• 

  Postal Savings Bank Of China (PSBC)

  Jul 2018 - Aug 2018

  Business Department Intern

  1. Verified 800 pages accounting records of individual customer’s annual revenue and costs and figured out the total net cash flow to evaluate financial operations to judge whether it is a high quality customer.2. Based on enterprise’s financials, calculated asset-liability ratio , quick-action current ratio to conduct business assessment 3. Wrote annual reports and pointed out the improvements in asset-liability match

• 

  FinFabrik

  Jan 2019 - Feb 2019

  Research Intern | Research on Bitcoin Quantitative Timing Based on Time Series

  1. Constructed strategies and factors in the quantitative backtesting system by using the data of bitcoin hour line: -Adopted FRactal Adaptive Moving Average (FRAMA) based on the traditional moving average in order to filter noise, preserve the tendency of data and solve the problem of hysteresis quality -Applied OLS strategy to build operators. By choosing the closing price for the last 24 hours as variables, predict the price for the next hour. If the predicted price is higher, then send a buy signal -Added BOP index and William index, formed multi-factor backtest by equal-weight strategy -To prevent overfitting, used LASSO to select variables and optimize the model2. Data cleaning, and visualized the final yield curve Show less

• 

  Shanghai Luoshu Investment Co. Ltd

  Jun 2019 - Oct 2019

  Quant Intern | Quant Trading

  1. Based on 10 years’ daily data of Chinese A-share market, constructed a regression model on four factors (market, size, value and momentum) to obtain the residual as idiosyncratic volatility factor, then based on it, built a multi-factor backtest and calculated Sharpe ratio, turnover and maximum drawdown, achieving 33.84% return in the top portfolio2. Analyzed the correlation between regional brightness indexes and economic indicators, such as GDP, CPI, and IVA etc: -Determined whether satellite data can serve as a significant feature to predict the economy (Global, U.S.A., China). -Further studied regional differences by classifying data in terms of province, year and city etc. -Calculated year-on-year brightness indexes to find outliers, compared datasets from different companies by outliers, and gave suggestions on choosing datasets and potential companies to work with -Concluded that datasets were more reliable in major cities and recent years, both with higher R2 and t-values2. Summarized highlights from 50+ research reports and tested some of them using datasets crawled from public resources Show less

• 

  ZHONGZHOU Futures Economy Co., ltd.

  Apr 2021 - Aug 2021

  Asset Management Intern

  1. Conducted time series analysis on the oil futures -Analyzed the seasonality of the spread of soybean oil and soybean meal, soybean oil and palm oil using Python -Implemented ARIMA model to fit one-year data of palm oil basis, under the lags of 20, determined the optimal model ARIMA(11,1,11) to predict 2. Constructed strategies and factors in the quantitative backtesting system by using the data of Chinese commodity market -Used basis factor to back-test based on 10 years’ data and yielded 153.37% excess return with 16.24% drawdown -Formed multi-factor model including basis, roll yield, warehouse receipt, the change rate of net long position and skewness, used Sharpe ratio of each factor as weight, obtained one-year excess return 31.26% -Applied Dual Thrust strategy to back-test palm oil future, obtained the optimal parameter of buyline and sellline, with 25.6% annual return, 7.3% drawdown and Sharpe of 1.53 3. Gathered fundamental data and made balance sheets of oil and oilseeds futures Show less

• 

  Valkyrie Trading

  May 2022 - Aug 2022

  Derivatives Trader Intern

  1. Traded crude oil options as a market maker and achieved positive total PnL-Retreated option prices after trading options and combos like risk reversals and butterflies and hedged through futures -Made money from edges between bid and ask price (scalping), Greeks including theta and vega and gamma hedging2. Created a dashboard using Streamlit where users input portfolio name and trading date to their desire-Calculated PnL from gamma hedging (hedge deltas on a preset hedging frequency if delta exceeds hedging threshold)-Obtained PnL attributed to matching ATM straddles, fences and strangles by applying ATM volatility changes, changes of (volatility of 25 delta - volatility of 75 delta) to Retreat Generating Function to get new Volatility Curves Show less

• 

  Columbia University in the City of New York

  Aug 2022 - now

  Teaching Assistant

  Currently working as a teaching assistant in the Stochastic Applications course of Professor Lars Nielsen.

• 

  Sumitomo Mitsui Banking Corporation

  Mar 2023 - now

  Model Validation Analyst