How to read an academic paper - b - Week 2 - Strategy

The first part of the paper after the title,

you'll have something called abstract.

Now abstract, what it contains?

Basically, the entire paper is summarized in about 150 to 200 words.

Sometimes, less than 100 words.

Abstract will tell you the main question of the paper.

What the paper does, and what results they get, the authors get.

So these are summarized briefly.

So if you just read the abstract,

unless you have a understanding of the literature, unless you have an idea of

what the paper is trying to do, you may not understand much.

You can just read the abstract.

It's not too taxing, it's just 100, 200 words, as I've told you before.

But I don't

think you will get a full sense of what's going on by just reading the abstract.

Once you move on, the next 4 to 5 pages of the paper is the introduction.

Now, what the interaction does, this section.

If you have some background in the literature,

if you read the interaction you'll understand the entire paper.

What introduction does is it gives you a motivation for the paper.

Why the authors actually started this inquiry?

What is the background?

What had happened in the past?

And how others contribute to the literature, and

of course, their main results.

They also briefly talk about data sources.

They will also talk about what methodology they employ that's very important.

As I have told you the referee usually asks for

you know says asks for robustness test.

They have to, the authors have to show that their results are robust to

alternative explanations.

So lot of these things are briefly explaining the interruption.

Another question is purely from a trading point of view,

do you need to read the interruption or not?

I encourage you to read the interruption.

And those parts which are technical, where you have these statistical

terminologies where there is a lot of math you might just ignore.

What do you need to understand is that what is the trading strategy

that is being employed here?

And from where do you get the data to replicate these training strategies?

If you get a broad idea of this and

what are the kind of returns this training strategy has generated in the past.

And also what is economic reason why this trading strategy works?

If you understand this much, after the reading the introduction,

I think you are in good shape.

After this, you have institutional background.

No, institutional background basically describes the institutional

set up of the country from which the authors use the data.

Most of the papers are obviously from the United States, but

there are papers where data from other countries are also used.

Now one important check that you have to do here is A,

are the authors using some proprietary data.

Sometimes what happens is authors go to some government agency or

a central bank and get some data.

Now if this trading strategy depends on

having access to this data you can just forget that paper.

It may not be possible for an individual investor to get this kind of data.

Whereas academic researchers will have access.

So you should ensure that the trading strategy is not a result of a typical

institutional feature which is applicable to a particular country.

Then, the trading strategy also of course applies to only that country.

If that institutional actually is not available elsewhere,

then you can't do anything.

Suppose I'll give you an example.

Suppose the training strategy involves shouting and

the country where you want to use this strategy, shouting is not allowed.

Then there no point in going through that paper.

When shouting is not allowed, of course there is no way you can replicate

the strategy in the country where you want to trade.

So from that point of view, you should be able to read institutional structure.

Institutional structure typically is not very, very technical.

It explains the market structure in that country in brief.

So it's not very dense you can glance through the structure.

Then, the next part usually is about data sources.

This is very, very important.

I've already given you hint when you read the institutional structure itself.

You should have an idea that whether this is something which is typical to

a particular country data sources as I've told you before.

You should look whether the authors are using some proprietary data or

it's a public data.

If it's proprietary data, stop.

Forget that paper.

You only look at papers where data source is public, anyone can access.

By public, I don't mean free.

Sometimes, you may have to pay, but since you're going to trade and if you

are confident about the strategy, it's not a bad idea to you an access paid data.

But most of the data sources are free, are freely available.

Once you understand the data source.

The author also describes how he arranges data.

How he or she organizes data, all that you can have a look at and understand.

Then, the next part is the most important part.

Once you get used to reading papers you can actually start from there.

If you actually understand the abstract fully,

you can just get into this next part that I'm going to describe.

This is about reading algorithm.

Now that is why we're interested in the paper.

From our trading point of view, our interest is more

In this algorithm rather than the knowledge that is coming from the paper.

Of course we are interested in knowledge but

the primary impressed is this algorithm.

This is the meat of the paper that a trader has to understand if you have to

implement.

So what does this section do?

Here the author describes the exact formula that he uses for trading.

So here a prospect of trader will go

line by line word by word, understand this fully.

What do I mean by understand this fully?

You should be able to generate this data and generate

this formula using whatever software that you're comfortable with.

And come out with the output that the author comes out.

In other words, you should be able to replicate the formula.

That's the level of comfort that one should have.

Now how do you get it?

What we will do in this course is that for some papers we will

take you through this formula in detail, step by step.

We will tell you what are the components of our formula.

Where do you get the data from you will actually show an illustration

with an actual company and show you how to create these kind of scores.

We will also show how do you trade and finally we will also present the results

that the author has obtained and since we are in India, so

what do you'll do is if you apply the same strategy on Indian stocks.

Then what would have been your results?

Even that we will show, for the kind of strategies.

What we encourage you to do is when the study is done, of course the easy way is

to just try this on stocks and see whether it works.

As I warned you before It may not work in the same form where

it has worked 20 years before, you need to improvise.

So the first thing is you should first try whether you can replicate on your own.

And then try to apply this on a market portfolio.

And once you're confident that this works and then only you should try trading,

and based on the insights that you obtained you should try to generalize.

That's our advice.

Now coming back to the sections of the paper, paper doesn't stop at there.

Once your trading strategy algorithm is done

then the author gets into the next part of the paper, which is hypothesis.

What does he do here?

Here the author tries to explain different

hypothesis as to why this certain strategy works.

And why he also presents tension in the hypothesis.

There will be reasons why this certain strategy works.

There will be reason why it should not work and then he shows

the purpose of the paper is to empirically test which of the hypothesis works.

As an assignment, please, remember you can never accept the hypothesis.

You can only reject something.

And the way the approach is author rejects all other explanations, and

then comes to a conclusion that possibly this must be the reason why it was.

And again, this is more foreign academic audience.

It's nice to know I won't repeat this once again.

I'm not discouraging you from reading the paper.

Don't think that after this module,

don't think that you should not read the entire paper.

No, that's not the purpose.

It's good.

Somebody who reads and

understands the entire will do better than someone who reads only a part of it.

But then the purpose of this model is to tell you

that you need not understand the entire paper.

But only from a trading point of view.

You just have to understand the trading strategy that comes out of the paper.

How to implement it, data sources and the expected results.

And the ways of back testing it.

That's the purpose of this module.

So purely from that utilitarian point of view,

you can try to ignore this hypothesis part.

You'll not loose much.

How to read an academic paper - b

Design your own trading strategy – Culminating Project

Reviews

Taught By

Ramabhadran Thirumalai

Explore our Catalog

Join for free and get personalized recommendations, updates and offers.