Fractals

What I learnt from Victor Neiderhoffer | Article Review

I have been a fan of Victor Niederhoffer for a long time. What a talented multifaceted man. If you are into trading or any kind of speculative business you've got to read his books.

I read this (https://www.newyorker.com/magazine/2007/10/15/the-blow-up-artist) article about him and thought of sharing some interesting snippets.

1. Experience in some sort of Performance Sport - I do think it helps to have had played a sport or done any performance related thing like Music etc. The reason I think its relevant is from a skill development perspective. All forms of sports or crafts require a progression of skill levels. It also requires us to observe, reflect, deconstruct, attempt and practice tasks. 

Capture_1-Sport.PNG

2. Discipline, avoiding noise and focusing on Short term moves - Being on time I guess is the first aspect of discipline, alludes to the idea of routines and their importance in success. The second aspect is avoiding noise, as a trader you need to have confidence in your system and approach and you need to avoid noise i.e informational noise. Thankfully I quit TV a decade back, and have never gone back. Lastly, look at that focus on "Short term moves". I have always been against the Random Walk/Fama school, and have been more of a Mandelbrot follower. Good to see the same beliefs at play here. 

Capture_2-Trading Style.PNG

3. Market Statistics as an Anchor - It always helps to know how things worked in a similar context in the past. The reason it helps is because prices in the short term are nothing but a reflection of the emotions of people, and that does not change.

Capture_3-Trading Style_Statistics.PNG

4. I told you so! If someone tells you that day trading does not work, its because most traders do not have the skills to make it work, but by itself patterns in shorter periods are far easy to predict than in longer periods.

Capture_4-Trading Style_shortterm predictability.PNG

5. RIP - Efficient Market Hypothesis and Random Walk Theory - The moment you shorten your trading time frame, you will start to see the inefficiencies. If you are a trader reading this, think about Open Close as a separate event and High Low range formation as a different event, and you will see the inefficiencies clearly.

Capture_5-against RMH.PNG

Lastly, understand where risk comes from, I mean risk of ruin. While there is always a defined risk with which we take trades, risk of ruin typically comes from the possibility of an unforeseen event happening and you having no control over it. Read that statement again, the second part especially.

Now think about how to avoid being in that situation, that should be good enough to keep you going in this business.

Happy Trading!

Meditations on Screen-time - Part 2

In the first part, over here - Meditations on Screen-time - Part 1 we talked about what is it that one needs to do while looking at charts, the first aspect that we focused on was the difference between Seeing and Observing. 

In this post we will move further and see what is it that we can do with what we have observed.

By observation we collect data or information about whatever we are observing, we may be looking at a 1 min time frame chart or a 5 min time frame chart, or perhaps a footprint chart, it could be anything that is contextually relevant to you. Over time with note taking and reflection, you would have a lot of this information in your mind, what the psychologists call "Crystallized Intelligence". 

The next step is to put this observed information to use, by making inferences (Fluid Intelligence). There are two ways to look at the concept of "Inference" one is - as a plausible explanation about an observation (The Why), the other way to look at inference is the way Bayesians use it, for them Inference-ing is about thinking in terms of probability distributions, and updating the values we assign to the distribution as we come across new evidence. If that is too much to handle, let me try to simplify it with an example.

Let's say you have been observing the breakout in NIFTY the moment it crosses Day High, there are few ways to cognitively process this information -

a) You can think of "causes", meaning why is the price shooting at the point. You may deduce that perhaps stops are getting triggered etc. - This is an example of the first type of inference. No harm in doing this, the only limitation is, a lot of times its more difficult to find causes than correlations, and as a trader you know whats more important.

b) You can think in terms of "averages" - You may deduce that usually it shoots up by 10 points and keep that as a reference. This is good but then "averages" are not that precise as we know, and more importantly does for help build a framework/scaffold on which one could build a sort of model as more data/evidence is observed. That brings us to the third way.

c) You can think in terms of "Probability Distributions" and "Conditional Probabilities" - Every time you observe the breakout you could create an imaginary distribution.

For instance keep one can keep the following exponential distribution in mind (look at it more from a visual perspective rather than for it's mathematical accuracy)   

Now imagine, putting all your observations on the curve, perhaps adjusting the curve based on observations. That would get you to think about the outcomes more from precisely. You may be able to tell yourself, if Day High crosses we will get 2 points for sure, I am 90 % confident about it. 

Now add to this some "Conditional Probabilities" - Lets say you also observe that on days were you get a range beyond X in the first half of the day the above distribution no longer remains valid.  Now what you have done is you have found an exception to the above rule. Over time you would be able to "cognitively" add several such conditional exceptions as well to your inference. 

I hope the above example helps you visualize outcomes in a more structured way. If you are more interested in the Bayesian way of thinking about situations and outcomes you may want to start here. 

https://www.farnamstreetblog.com/2012/12/thomas-bayes-and-bayess-theorem/

So at the end of it, all the time you spend in front of the screen, watching price action and charts should help you create a probabilistic framework in your mind, and the more you gather evidence the more you should be able to refine your "confidence levels" of the outcomes.

As always happy to hear more form you. 

Meditations on Screen-time - Part 1

Recently, I gave an assignment to a novice trader to observe certain parameters and trade locations, on live NIFTY charts, as the market develops. So, at the end of the week, we had a catch up session to exchange notes. I was eager to know what he had observed. 

I asked the question "So what did you observe in the past 5 days every time when NIFTY crosses the VWAP?" - He said, "NIFTY either goes above VWAP or below it, and that keeps happening, I don't know what else you wanted me to watch, (he also added) I actually watched it for a couple of days and then stopped, as it didn't make much sense to me"

I was not really surprised, as this was not the first time such a thing was happening. So over the long weekend I thought of reflecting on what is it that I really expect them to do while watching the charts, so here is a two part series on Screen-time, in the first part we will focus on the idea of "Observation" and the next would focus on "Making Inferences"

When I started trading, I actually didn't know what "Observing" was all about, I used to see the charts. It was only a year into trading, and much reflection later that I realized what I was doing. So first things first.

Seeing vs. Observing 

Read this instance from Sherlock Homes's "Scandal in Bohemia"

http://bigthink.com/artful-choice/dont-just-see-observe-what-sherlock-holmes-can-teach-us-about-mindful-decisions

In the context of trading, here is how I would translate what Sherlock is trying to say.

Observation includes the following

1) Capturing Statistical Data - When observing, are you looking at data elements like, time taken for the price to move from point A to B. A being a reference point, how many times does the price B get revisited, and things like this. While you are doing this the next point should also happen in the background.

2) Creating a Mental Baseline - After observing the charts for a few days, across the same reference points, you should be able to create mental baseline, for instance, If the price moves beyond this point, it does go up by X points on an average. This baseline would get refined over time, but this "program" has to be running in the background of you mind all the time.

Now that we hopefully have some idea of what does it mean to "Observe" lets looks at some ways to enhance the process of observation.

How to get better at Observing?

1. Focus - Imagine you had two tasks to choose from - a) To look at every passing vehicle on the road to see if it,s registration number starts with odd or even b) To look at every passing two wheeler and see if its registration number starts with odd or even?. In all probability your performance on the task (b) would be better. Same logic here. If you look at too many things, too many indexes or stocks, its difficult to observe something specific. The other side of focus is also distractions - It could be in the form of other people in the immediate environment, browsing something else, phone calls etc. That too needs to be controlled.

2. Perspective - Taking the same example as above imagine you are given two more choices now -  a) To look at every passing two wheeler and see if its registration number starts with odd or even?. b) Sit in a control room and look for all the passing two wheelers, both from back and front - again in all probability your performance on the task (b) would be better. This is akin to watching both the CE and PE option strikes of an instrument. You can also build perspective by looking at the same instrument in different time frames, same instrument with different types of charts, for instance you can juxtapose both a foot-print chart a candlestick chart. Or may be a PnF chart and a Candlestick chart. The idea is to get multiple perspectives of the same instrument.

3. Technology -  Well, here technology means, more screen real-estate. One of the reason a multi monitor set-up helps is it, helps you simultaneously view the price action,  which is very different from flipping across tabs. It does not (at least to me) give the same visual perspective. Take at look at my set up below.

On the top are footprint charts of NIFTY and BANKNIFTY along with PnF charts + The bottom two monitors are dedicated to slightly ITM CE and PE

In a way on a single side/pane (without having to turn my head) I get to see multiple views/perspectives of the same instrument. Personally I would say, this has been one of the best investments that I have made in my trading business, and it has enhanced my understanding of NIFTY quite a lot.

4. Documentation - Note Taking, also plays a very vital role, I have seldom seen a high performing trader without a note book on his/her trading desk. How to make notes perhaps could be a post by itself. But at the moment we will keep it simple - so do jot down, things you may want to refer, back test, or explore further later. Essentially what note taking does is, it eases the job of your memory for more important tasks and there by improving your cognitive capacity. 

All the above steps should help you increase your ability to capture "cognitively"
speaking much more than what you otherwise would.

In the next part we will explore, what is it that we can do with this observation. The logical next step would be to "Make Inferences".

Also you may want to look at this old post on visual perception, which obviously is related to Screen time. - Learning to see data - NYT Article Summary & More

Learning to see data - NYT Article Summary & More

Article Link - https://www.nytimes.com/2015/03/29/sunday-review/learning-to-see-data.html?_r=1

I came across this article sometime back when I was researching on "Perceptual Learning". Perceptual learning basically is about how we learn and create patterns in our brain using all of our senses.

Here's an excerpt from the article which explains it further 

 

In the context of markets, it is closely linked to the idea of "Fractals" and concept of "Thin Slicing" from psychology.

When we are looking at charts and trading, esp. seasoned traders get an intuitive feeling of what may happen next. I have seen this with several traders. They see a pattern emerging and they know how its going to develop further. Obviously its not fool proof, but even if it's better than a coin toss probability, its good enough, in reality I have seen people who have a probability of getting it right more often. But then, it may not just be their refined visual perceptual abilities, there could be other variables that they may be tacitly processing, perhaps some moving averages or some other data. Its only when their brains kind of processes all these elements together that they get their predictions right.   

Also, since price movements are fractal in nature, traders are further able to apply the same sense to different time frames, and perhaps benefit from larger swings as well.  

But then how do we develop this intuition, unfortunately there are no shortcuts. Sufficient "Screen Time", "Focus" and "Reflection" seem to be the only ways get there.

BTW, if you are more interested in the subject of forecasting there is this awesome book by Philip E. Tetlock and Dan Gardner called "Superforecasting: The Art and Science of Prediction", here is a nice summary.