25705 Financial Modelling and Analysis Spring Session 2025 Project #2: Modelling the Dynamics of Stock Volatility

联系我们： 手动添加方式: 微信>添加朋友>企业微信联系人>13262280223  或者 QQ: 1483266981

25705 Financial Modelling and Analysis
Spring Session 2025
Project #2: Modelling the Dynamics of Stock Volatility
I. Overview
Volatility is synonymous with risk. Volatility forecasting is an integral part of
investment decision and risk management. This project aims to reenforce the statistical
concepts and data analysisskills discussed in lectures. Students will develop forecasting
models, evaluate their performance, and use forecasts to make investment decisions.
II. Data
Analysis is based on the same stocks as in Project #1. Update your daily stock
data to 26/9/2025 before starting the analysis. The estimation period is from 3/1/2020
to 31/12/2024; the holdout period is from 2/1/2025 to 26/9/2025. The first worksheet
should be named “*** daily data” with *** being your stock code. [2 marks]
III. A benchmark model for predicting daily volatility
Create a new worksheet “Benchmark model”. The benchmark model for
forecasting daily volatility is
Vt = β0 + β1Vt-1 + β2 R
t 1 + β3R+
t 1 + β4ln(Vlmt-1) + εt
where Vt and Vt-1 are daily volatilities, R
t 1 and R+
t 1 are the lagged negative and
positive returns, and Vlmt-1 is the lagged trading volume in million shares. Follow
Lecture 9 Excel exercise to create these variables.
Estimate the benchmark model in the estimation period. Save regression
residuals and calculate its 1st-order autocorrelation. In about 300 words, discuss the sign
and significance of each coefficient, R2
, F statistic, and the residual’s 1st-order
autocorrelation. Draw comparison with Tesla in Lecture 9. Based on these comparisons,
does the benchmark model perform better for your stock than for Tesla  Explain why.
[2 marks for calculations, 3 marks for discussion]
IV. An enhanced model for predicting daily volatility
Create a new worksheet “Enhanced model”. Select a minimum of 6 and a
maximum of 8 variables to forecast daily volatility of your stock. You can include
some variables in the benchmark model. Potential variables include
 Sector and broad market index return and volatility.
 Daily exchange rates and volatilities if your company has significant sales overseas.
2
 The number of daily news on your stocks in Factiva available from library. It is a
proxy for daily public information flow on your stocks.
 The number of daily Google searches for your stock. It is a proxy for time-varying
investor attention to your stock.
 Historical company financials. Find the release date and use the change as the new
information. For example, if sales is released on Sept 15 and is 12% lower than the
last month, ΔSales = -12% on 15/9/2025 and 0 on other days of the month.
 Factset contains comprehensive global market data and company fundamentals. It
has a somewhat steep learning curve for extracting data. Email me if you want to
have access via UTS subscription.
You should also consider potential nonlinear relationships between daily
volatility and some forecasting variables, as well as interactions between forecasting
variables. Both are discussed in Lecture 11. Keep in mind that your model should be
linear in regression coefficients.
Estimate the enhanced model in the estimation period. If you include variables
in the benchmark model, explain any changes of their coefficients in the enhanced
model. Discuss the signs and significance of new variables. Explain the differences
between the enhanced and the benchmark models in terms of adjusted R2
, F statistic,
and the 1st-order residual autocorrelation. [4 marks for calculations, 5 marks for
discussion (~300 words)]
V. Rolling forecast
Create worksheet “Rolling forecast”. Copy the data of your regression variables
including the holdout period. Follow the instruction in Lecture 9 to setup the worksheet.
Use Linest function to conduct rolling forecasts in the holdout period. Even for a good
model, a small number of the forecasts can be very bad. Check if there are any “insane”
forecasts, e.g., σ t  σT, wt < 1 and you hold 1-wt in cash; if σ t  1 and you borrow cash to invest. We assume a borrowing limit 50% of
your portfolio value, i.e., your maximum weight wmax = 1.5. If wt > 1.5, it should be
replaced by 1.5.
Calculate your portfolio weight wt = σT
σ t
in column E, portfolio return rp,t = wtrt
in column F, and portfolio volatility σp,t = wtσt in column G. Tabulate the average,
standard deviation, min, and max of rt, wt, rp,t, and σp,t at the bottom of the daily data.
Report the Sharpe ratios of your stock SR = r
σ  and portfolio SRP = r
σ
p
p
.
1 Is rp > r  σ p  SR  Is Cor(rt, σt)  1)  Did you make or lose money when you
borrow  Are there extremely good or bad days strongly influencing SRP  How are the
signs of Cor(rt, σt), Cor(rt, σ t), and Cor(rt,wt) related to the success and failure of the
strategy  [6 marks for calculations, 5 marks for discussion (~450 words)]
1 This Sharpe ratio applies only if r > 0 and rp > 0. If           < 0, maximizing r σ p p leads to the perverse effect of higher, not lower, σ p. Assuming the benchmark return = 0 in Israelsen (2005, available on Canvas), his Eq (5) defines a modified Sharpe ratio as SR = rp/σ P Sign where Sign = rp/|rp|. When rp < 0, Sign = -1 and SR = rpσ p. If two portfolios have the same negative return, the one with lower risk has a higher (less negative) modified Sharpe ratio. VII. Video presentation Record a 5-minute video in MP4 format to discuss: The aim of the statistical analysis in Project #2. Your forecasting variables and justifications. Your forecasting model performance compared to the benchmark model. Your portfolio performance in the holdout period: return, volatility, Sharpe ratio, in comparison to the stock’s performance. You can but are not required to use ppt slides. The video carries 5 marks based on structure (the above topics in sequence), content (concise discussion of the above topics), and delivery (clear speech, confident tone, avoid reading slides). Content after the first 5 minutes will not be considered. VIII. Excel and video submission This project accounts for 40% of your final grade. You must complete this Project on an individual basis. Submissions are made on Canvas before 11.59 pm on 3 November 2025. It is your responsibility to ensure that submission is not delayed for any internet or technical problems. Excel files must be in xlsx format and must retain formula used in each cell. Files in other formats and answers without formulae in cells, i.e., with only numerical values, receive 0 mark. All instructions should be strictly followed, including the worksheet names. Failure to follow the instructions leads to deduction of marks. Reference (available on Canvas with this case instructions) Israelsen, C.L., 2005. A refinement to the Sharpe ratio and information ratio. Journal of Asset Management, 5(6), pp.423-427. Marra, S., 2015. Predicting volatility. Lazard Asset Management. 4

25705 Financial Modelling and Analysis Spring Session 2025 Project #2: Modelling the Dynamics of Stock Volatility最先出现在KJESSAY历史案例。