This course will provide back test results for all the strategies in developed and emerging markets. The learner will also be taught scientific ways of back testing without succumbing to either look ahead (or) survival bias. You will learn various methods of building a robust back testing system for the strategies discussed in the previous course. You will be taught how to differentiate between mere data mining and results based on solid empirical or theoretical foundation. Next, you will learn the ways and means of back testing the results and subjecting the back test results to stress tests. After which, you will learn the various ways in which transaction costs and other frictions could be incorporated in the back testing algorithm. Finally, you will learn techniques for measuring a strategies' performance and the concept of risk adjusted return. You will use some of the famous measures for risk adjusted returns such as Sharpe ratio, Treynor's Ratio and Jenson's Alpha. You will see how to pick an appropriate benchmark for a proposed fund.
G Score - Strategy
Loading...
来自 Indian School of Business 的课程
Advanced Trading Algorithms
119 个评分
从本节课中
Strategy - G Score
After completing this module you will be able to understand the meaning of G Score, build a trading strategy based on G Score and test the strategy that you have built.
与讲师见面
Prasanna TantriSenior Associate Director
Center for Analytical Finance
[MUSIC]
Learning outcome.
After watching this video, you will be able to understand the G score strategy
and you will also be able to implement the G score strategy.
[MUSIC]
Let us spend a minute and know the G score strategy.
You use eight financial signals to calculate it, G score.
Based on the values of the G score,
you can separate the firms into expected winners and expected losers.
Firms with high G score earn substantially higher returns than firms
with low G score.
So you long the firms with high G score and short the firms with low G score.
A long short strategy based on the G score earns significant excess returns,
and most of these come from the short side of the portfolio.
The best thing about this strategy is that it relies entirely on publicly available
historical financials of a firm.
It does not use market indicators or other information such as analyst forecasts
which rely implicitly on non financial or private sources of information.
Now the steps in the strategy.
The first step in the strategy is sample selection.
You should keep the following in mind when preparing the sample.
First, you have to calculate the low book to market portfolio.
For this, you take all the forms in the market and
calculate the book to market ratios for the current year.
Current year we denote it by t, and we also calculate the book to market ratios
for the previous years which we denote by t-1.
We do not use the firms with the negative book to market ratios in our analysis.
So from your sample, delete the firms with the negative book to market ratios.
We then arrange the firms in ascending order according to their book to market
ratios based on year t-1, please remember it is t-1 and not t.
And then divide them into quintiles.
The firms in the lowest quintile of book to market ratios are the growth firms,
and the firms in the highest quintile of book to market ratios are the value firms.
Now, you'll have to load the cut offs of these quintile groups.
We take the cutoff for lowest 20th percentile, and
compare this value with the book to market value of all the firms for the year t.
All the firms whose book to market value for
the year t falls below the cutoff value are taken.
This set of firms is our low book to market portfolio.
The calculation of many of the signals used in G score metric
requires comparison with industry medians.
So you require that a firm have at least three other firms
in the same industry in the same year.
The industry in this case is defined by if you remember the two-digit SIC code.
In case sufficient firms are not available in a particular industry,
you do not consider that industry.
So, we impose the condition
of having a minimum of four firms per two digit SIC code.
The second step in the strategy.
The second step is to gather all the financial information for
these low book to market firms.
We collect the quarterly financial information for these firms.
If adequate quarterly information is not available,
the information from the latest physical year is used.
This ensures that there is no look ahead bias in the computation of the signals.
Even if it means that the data may be sometimes outdated.
In relation to all these criteria, we only consider those firms which have earnings
and cash flow information available.
The third step.
The third step involves calculation of the signals.
The three signals relating to profitability and cash flows, that is,
G1, G2, and G3, are created using the annualized financials.
The two signals related to name extrapolation, that is,
earnings variability and sales growth variability.
Which we call G4 and G5 are generated from quarterly financials of the past four
years with the constraint that at least six quarters information be available.
You'll measure earnings variability as the variance of a firm's return on assets in
the past four years using quarterly information.
Now, I know what you're thinking.
You're thinking that quarterly information might induce variability
owing to seasonality.
That is the reason why we use firms in this same industry.
The industry adjustment should mitigate the problems caused by seasonality.
We measure sales growth variability as the variance of a firm's year over
year sales growth or the past 40 years using quarterly information.
I may be saying year over year here, but
we will calculate the values using quarterly information.
First quarter of this year compared to the last quarter of the previous year,
second quarter of this year is compared with the first quarter, and so on.
You get the idea.
Three signals alert to conservatism, which requires G6, G7, and
G8, are also created using the annual list financial statements information.
Now for the fourth step.
In the fourth step,
we analyze the calculated values and assign values to the signals.
Once values are assigned to the signals, we will add them up and
get the G Score value.
Based on the G Score value, we decide winners and losers and trade them.
So this is the strategy.
In the next video, with the help of a sample firm,
I'll show you the calculation of G score.
