StatShot

Finding 'edge' through Data Curation

When we talk of ‘edge’ in trading it essentially means, what is it different that you or your system has which would lead to an ‘alpha’ in terms of returns. One way to extend that question is to ask ourselves, as to what are the sources of that edge?

To me having a more deeper and nuanced understanding of the contexts and setups that I trade, exponentially adds to my edge. To get a better sense of my contexts, one of the practices that has massively helped me in my trading, is curating setup specific data. This is a lot of work, let me tell you. Sometimes it’s very difficult to train the computer to do what we humans can do intuitively, which means a lot of it is manual labor.

Let me try to give you a sense of what I mean, if you follow my blog you would know that, these are the three setups that I trade.

NIFTY - Scalping Set-up - 01 - Opening Spikes & Opening Drive

NIFTY - Scalping Set-up - 02 - Mid-Day Mean Reversion

NIFTY - Scalping Set-up - 03 - Afternoon Range Extension

Now each of these setups have their nuances and details, like

a) At what time did the entry get signaled? Is that time range bound? Is there a seasonal skew to it?

b) Range breakouts on VIX? Time and Amplitude.

c) What is the average size of the pullbacks that in the setups you trade?

d) What is the ideal holding time for your some of your setups, based on the length of the trends?

A lot of these computations are possible only if you have specific data. Therefore, this is a practice that we follow in-house and for our clients - i.e. to capture such data so as to run tests on it.

Here is a mini snapshot of the data. (Disclaimer : The snapshot may make no sense whatsoever without context)

This practice when followed over long periods of time can give you a gold mine of data, which (I believe) can add to your trading expectancy.

ATR (Average True Range) vs. ADR (Average Day Range) | What they don't tell you

I was having this conversation with a coachee of mine, who was bent on using ATR instead of ADR as a reference for trading and I had to help him understand the difference and the context as to what is relevant where, and why I lean towards ADR. Below is an excerpt of what I told him.

The case for ADR

First things first - What is ADR - ADR is simply the average of intraday (High-Low) value. This excludes Gaps.

So - What is ATR? - Here is a better explanation. Essentially ATR is a range calculation which includes Gaps as it calculates from PDC (Previous Day Close).

So it essentially boils down to the significance of Gaps.

Let’s digress a bit to understand why do we use Range as a reference.

To me Range is a good indication (of / or a proxy for) volatility. You will see that for yourself, if you follow VIX, as VIX increases, so does range (Ref. the plot below). By including Gap in the calculation we may get an incorrect and irrelevant view of the intraday volatility.

Based on Past 60 Days’ data - NIFTY Futures and INDIAVIX

As an Intraday trader I am concerned only about what happens between the Open and Close. That is what is my playing field. I am not a positional trader to take advantage of or get affected by Gaps.

So the next question to ask is? Is there a correlation between the size of the gap and the ADR for the day? This would determine if including GAP data helps us in any way.

I did a quick math by calculating the Correlation Coefficient with Gap size as an independent variable and (ADR) Range as a dependent variable and I get a score of 0.36. Take a look at the scatter-plot below.

Based on Past 60 Days’ data - NIFTY Futures

As you can see there is no linearity in there.

So given that there seems to be no correlation between Gaps and ADR, I would recommend using ADR and not ATR,

ATR is relevant in markets or products were Gaps have a correlation with Range, which does not seem to be the case with NIFTY.

However, I do keep an eye on the Gaps, but that is more from a perspective of understanding if there is a visible change in the market structure, more on it later.

Here is a snapshot of the data that I reference during the day. It gives me a clear sense of the developing range with references of Previous day and a 20 day Look back period.

Snapshot of NS-RangeByTime Indicator for NT8

NiftyScalper | StatShot - 02 | Volatility Clusters

StatShot 01 - Here

In this post let's look at the time segments within the day when NIFTY is more volatile. If you ask what is the relevance of this? Then perhaps you would need to dig a bit on the Internet to understand the relevance of volatility in the context of day trading, but for those in the know, here are the time segments which offer you the "meat" of the day. 

The idea is, your return on time invested would increase exponentially if you avoid trading in time segments that are not volatile. 

A through F stand for an hour of trading time and G for 15 mins. 9:15 to 15:30

A through F stand for an hour of trading time and G for 15 mins. 9:15 to 15:30

The light blue colour represents moves of 3 points / min; light pink represents 5 points / min; and dark blue represents 10 points / min. 

If you are wondering as to why segment "D" has the maximum number of 10 point moves well, here is some hint.

 All standard disclaimers apply, trade safe!

NiftyScalper | StatShot - 01 | High/Low Time Segment

Starting a new series here on the blog called "StatShot". The idea is to bring to you statistics based snapshots of NIFTY. Be warned that you cannot use these in isolation for your trading, it has to be used in a specific context while creating your strategy or setups.

Coming to our first 'StatShot' in the series. Let's took at what I call the 'High/Low Time Segment'. As an Intraday trader one of the key things is to determine your supports or resistances, and to do that you need to know when (at what time i.e.) does an index mark its high or low during the day. Of course you could slice and dice the data to get to more nuanced stats and probabilities.

So heres's how it looks like for NIFTY

Based on NIFTY Data (2014 to 2017) - A through F constitute 1 hour each and G is 15 min

Based on NIFTY Data (2014 to 2017) - A through F constitute 1 hour each and G is 15 min

With this info. you can take a probabilistic call as to where your supports and resistances lie.

Hope this helps.

Trade Safe.