Options trading and artificial intelligence

#1
I'm sharing this for those who are interested in the subject. https://deepmind.com/blog/alphago-zero-learning-scratch/


I'm sure many of us are familiar with AlphaGo, the AI program by DeepMind owned by Google, that managed to beat the best human Go players, a feat unimaginable just couple years ago. Now there is a new and improved version just one year later, called AlphaGo Zero, that managed to beat the older version by a margin of 100:0. The paper was published 2 days ago. I read this news with lots of excitement because I have a hunch this is where the future of options trading lies.

Ok - this is a bold statement, I agree and I have no evidence to support that statement, but I'm excited because I do see a path forward for using the same technologies in options trading. However, I'm not envisioning a future scenario of a robot trader trouncing human traders. Instead, I foresee a future (at least in the immediate future) where an AI system will sit side by side a human trader, supporting and augmenting his capabilities. AI will become a tool like an OV or TOS, only far more capable and powerful. When that happens, trading without an AI tool will be like fighting a modern warfare with bows and arrows.

For those interested in the technical details, here's the link to the paper - http://www.nature.com/nature/journal/v550/n7676/full/nature24270.html#access

Specifically I see three technologies that are promising to options trading - Re-inforcement Learning (RL), Deep Neural Network (DNN) and Generative Adversarial Network (GAN). As explained in the article above, AlphaGo started learning to play Go by learning from human. Now AlphaGo Zero is no longer learning from human but rather learning by playing against itself, hundreds of thousands of games a day. That's how it can improve so quickly and more important create new plays that human has never seen before. That to me is exciting.

I see the same potential in options trading, using the same techniques, not to create a automated trading agent (at least not in the beginning) but rather to develop a system that assist in creating trade plans. Imagine an "AlphaOptions" that continuously learns on its own by trading hundreds of thousands of times a day, across thousands of underlyings, testing/discovering all kinds of possible trades. Not only can it help us create trade plans that we could not have imagined ourselves, most important of all, the plans will continuously morph and adapt to market changes. This can resolve a few issues that are severely limiting us today.

Currently, we do manual back testing and manual creation of trade plans. This process is tedious, time consuming and hardly comprehensive. When market changes, our plans stay static for too long. We also rely heavily on our own bias and intuition, thus getting stuck with old ideas. AI, on the other hand, can discover trading ideas beyond our biases and comfort zones. It can correlate multiple underlyings and create trading ideas stringing together multiple symbols, a task that is extremely difficult even for highly experienced options traders. Bear in mind, this approach is not the same as algorithmic backtesting. While algo backtesting can help speed up the back testing process but it will not help discover new trading ideas and will not automatically morph the plans based on changing market conditions.

RL is a fast growing machine learning technology that is at the core of AlphaGo. RL can be used to help us develop a trade plan. Let's say - if we want to develop a butterfly based strategy - if a human does it, he'll have to test out different wing size, how much OTMness, what price to pay, find out what's the optimal greeks for adjustments etc.. If you add up all the permutations that one has to consider, it'll become humanly impossible to do it comprehensively. Human need to use intuition to contain the permutations, but our intuitions will become the limiting factor. On the other hand, RL will discover trading ideas on its own. They less we instruct it what to do, the more creative it will be. It will test out all permutations and then randomly create new ways to trade that might potentially change how we view trading. Too far fetched? I think not - AlphaGo has already created moves in GO that human experts have never imagined and changed forever how the game is played.

DNN - is the another machine learning technology. It is good at pattern recognition. DNN can help recognize patterns and correlations in the market which will take human analysts way too long to figure out. Combining RL and DNN is the one of the hottest pursuit in AI now. It's called deep reinforcement learning. That's what AlphaGo uses and that's the technology I'm eager to explore for options trading. Potentially we can use deep RL to figure out which type of trades are optimal in changing environment, ie think changing skew, changing IV, pricing, bid/ask spreads etc..

Lastly GAN - it's an exciting technology that can generate fake data that mimics real life data. Why ? One thing that ML needs is tons of data to train effectively. We may not have enough. For example - how many blackswans have we seen? Handful. That's not enough to train a network. We need 1000x more. GAN can create fictitious but realistic market data to train our network. GAN is now used actively to create things as well. For example - get AI to re-draw a picture in the same style as Van Gogh or Picasso or create music in the same style as (who ever your favourite musician).

Sorry for the long post, but you probably can sense I'm excited !
Can all these be done ? To be honest, I don't know. I'm just a beginner learner but I think I know enough to see a probable path. This path may lead to dead ends. I can't be sure, but I see this as a challenging but intellectually stimulating path that I'm willing to invest a few years of my time exploring. The worst case is to realize it doesn't work in trading but I will definitely pick up an exciting skill critical for the future.

The reason I'm writing this post is to share what I know but also to invite others who have similar beliefs to collaborate. The vision I'm seeing requires a herculean effort and collaborative efforts from like-minded individuals. I'm still in the learning and discovery process. I won't be ready to start developing the system until a year or two later. Join me in the discovery process if you're interested. For those who are completely new to AI and ML and are interested to get started, I'm happy to share what I know about how to get started. The good news is all the technologies I described above are open sourced and there is a huge amount of learning resources. The only limitation is our time and dedication.

Last but not least, one common question is - "Even if what you said is true, the big boys have 1000x more brains that us. They will wipe us out before we can get started". My response is if that logic is true, then the same must be true for options trading itself. Since we can still be profitable trading options, it shows that if we choose a niche carefully, there are opportunities despite the competitions.

OK - I think I should stop typing !
 
#2
@Kevin Lee, thanks for sharing your opinion on AI and Options trading. I don’t think what you are outlining is more than max 5 years away!

Your excitement clearly shines through your post and I am 100% with you. AI will not only change options trading, but all aspects of what we today consider “work” and how we live our lives in general. We are already dependent on AI (all the apps on our phones, search, navigation, medical applications, and so on) but the huge awakening will be once it starts displacing people from their work...

While Singularity is far away, if at all achievable, AI assisted applications will become more and more powerful and soon seem to many as “intelligent”! It’s super exciting times!
 

GH13

New member
#4
I'm just a beginner learner but I think I know enough to see a probable path.
The fact that you refer to yourself as a "beginner" shows me how far behind I really am. I am still afraid that my AI will realize that I am an "inferior" partner and take me out! Thanks for the info, I wonder how much of this technology will be used to find short term mispricings in the option market vs. longer term trades like what a lot of us do on a day to day basis.
 
#5
Kevin thanks for posting. I'm in the same position as Trader G. Saying truth, for me your posts are main source of info about AI at this moment. I understand your excitement, I'm not sure though that I'm fully sharing it with you.

First it does seem that you envision future trading-AIs in present market environments. I think, that may be the case at the beginning stages only, later there will be AI vs AI. The best wins, if no single winner is possible then cooperation will be probable solution. This process will change markets as we know it.
To go little farther. In that new market games obvious edge will be to connect to AIs outside markets - at factories and financial institutions etc. - to influence good or bad decisions that can be cashed in. To far stretched? Maybe.

Another point I have little different opinion is fate of human traders. They will be gone. At the beginning – yes, later – no. I don't think human can add value to AI competing/fighting another AI. Is there a human supervisor on GO-machine?
Small traders vs Big Boys. Yes, small traders have a chance because it is still human vs human even if helped by algos. With AI present there will be no too small profit, every option to gain a penny will be exploited. What is not worth time and effort for humans will be well worth for relentless AI.

Maybe I read too much si-fi or maybe not enough white papers but above scenario seems pretty natural. Regulations may slow down this path but I'm not convinced.

Kevin, despite my view of future AI markets I'd love to join your quest to understand more of the subject. Unfortunately I'm stuck on more basic, more archaic – to say, tasks.
I wonder about your progress on VolSkew-AI (although I'm in stage of doubting real value of models at all : ).
 
#6
I wonder how much of this technology will be used to find short term mispricings in the option market vs. longer term trades like what a lot of us do on a day to day basis.
Mispricing, if they exist, can be found using traditional non-learning data analysis techniques.


I think, that may be the case at the beginning stages only, later there will be AI vs AI. The best wins, if no single winner is possible then cooperation will be probable solution. This process will change markets as we know it.
This scenario may play out in the future but that's probably many years away. Although AI has done seemingly incredible stuff, the technology is still very rudimentary. In the foreseeable future, we can expect Augmented Intelligence, ie machine augmenting human capabilities for quite a while before Artificial Intelligence can fully take over. The irony, in my opinion, is when AI is ubiquitous, it will actually become more equitable and much easier. We can just spend money to buy what we need, like we don't need to know how a car works to be able to drive one. But before that happens, Augmented Intelligence requires human knowledge to partner with machines. That's when we can't just spend money to level the playing field. It requires an upgrade of our own skills.

on more basic, more archaic – to say, tasks.
I wonder about your progress on VolSkew-AI (although I'm in stage of doubting real value of models at all : ).
I agree with your doubt of models, but for me, my distrust is with theoretical mathematical models. The more I study, the more I realized how unrealistic the assumptions are in order to make the math work. The models look beautiful mathematically, but they don't work well enough in real life.

Well... people tell me that regardless of how bad the math models are, that's the best we have and everybody trades with it. My response is no, that's not true anymore. Empirical data driven models have proven to be more accurate and with new computing capabilities, these methodologies are affordable to individuals. Black Scholes for example attempts to model options for every kind of asset classes and every individual symbol with a single framework. That's a futile attempt. Empirical methods, on the other hand, can model each and every symbol individually.

For example - we know there are critical differences even between something as closely related as SPX and RUT. How can SPX and Gold or Oil for example be modelled the same way ? We end up having to patch the generic BS framework with our own understanding of the asset. Therefore, although we know the greeks shown on our software isn't entirely accurate, we have a way to mentally compensate for it based on experience. However this is severely limiting. We are stuck with the few symbols we are used to. But with empirical methods, we can have individual models for tens if not hundreds of underlyings, learned by machine learning techniques. Once we can do that, we can start to work on trades that are based on multiple underlying and expect reasonably accurate modelling. According to what I've read, even wall street is moving towards empirical modeling. But that's my opinion and I might be bias.

No, I haven't done anything on VolSkew with ML yet. Although I already have an idea of what can be done, I don't want to start on it until I have completed my studies. I want to use the right tools for the work but I don't know all of what I need to know yet. I have shared my thoughts with some folks who are ahead of me in terms of ML. They might have started something. I don't know.
 
#8
Kevin, I wont continue discussion about AI. I appreciate your update. My position is stoic mean I observe (not very closely) development in AI (I use this name but in my opinion it is misnomer - what was mentioned here before), so I watch it, without emotions and with hope that I will be able to recognize when time to step down comes.

As per VolSkew-AI (again: misnomer) - I wish yours vision of it will come to life ASAP. It'll level play field, making trading easier for folks like me.
 
#9
For a light-hearted view of artificial intelligence..... progress in human-like robots has been staggering in recent years. Though clearly still some ways to go but it's phenomenal what has already been achieved

 
#11
Kevin, sorry I can't resist. Last video you posted... it's creepy. I guess it is different when you watch with highly positive attitude.
From 14' when not everything goes as planned - you may see it as just case for improvement in software or as sign of future problems when we improve software.
After 20' it is hmmm... odd just about Dave.
 
#12
Kevin, sorry I can't resist. Last video you posted... it's creepy. I guess it is different when you watch with highly positive attitude.
From 14' when not everything goes as planned - you may see it as just case for improvement in software or as sign of future problems when we improve software.
After 20' it is hmmm... odd just about Dave.
Ha ha ... it looks creepy because it’s 90% human like. When a robot is only 50% human like, it won’t look creepy. It’ll just be like a toy but when it gets to 90%, it’ll becomes creepy because suddenly we start associating it as human. Whether we like it or not, robots like this will be everywhere in our society in 10 years.
 
#14
Hi everyone, I share Kevin's enthusiasm, and I also believe that we about to enter an era of AI-assisted options trading. Deep RL could indeed be used to figure out "which type of trades are optimal in changing environment, ie think changing skew, changing IV, pricing, bid/ask spreads etc..", while GANs could be used to create market data as deep learning networks need at least an order of 100K data points to be trained. At the moment, I am exploring LSTM deep neural networks that I think are suitable for skew prediction. Given a current skew for a certain expiration and a percentage change in underlying - can we accurately predict how a new skew would look like?
 
#15
The only times it looked creepy to me was when it tried to smile.
Absolutely. It is creepy double way.
Firstly - because behind this human-like smile that evokes all our human responses is nothing else than calling out function ".smile".
Secondly - because you may allow yourself think that maybe behind this smile there is something else, something not human. Then you may start wondering what electric sheep dream about?

Whether we like it or not, robots like this will be everywhere in our society in 10 years.
yep

I share Kevin's enthusiasm
many do. "Dave's case" might be more interesting that robot's. I agree that practically process is unstoppable (or is it?).
 
#17
Regardless of one's view of artificial intelligence, one thing is irrefutable - the rise of data is already impacting every profession and our lives. Today, traditional business like banking, retail, manufacturing, logistics etc... have already turned into data driven businesses. Companies that are not incorporating data analytics into their strategic plans are quickly becoming obsolete. In my opinion, there is no reason why options trading is so unique that won't eventually become a data science.

Up till now, when I develop a new technique to resolve an issue that market throws at me, I rely far more on my experience and intuition than data analytics. For example, if I know I need to use a spread to address an issue, question is why 20 pts and not 30 pts or 40 pts ? Why place it at these strikes and not others ? Why maintain my portfolio at X delta and not Y delta ? Why should (greek 1/greek 2) ratio be A% and not B% ? What evidence do I have that the spread will work and that is the best available choice ?

Well... again I always start with my intuition and then I go through a series of back testings to see how the spread reacts in different environment. The problem is that this process is manual, time consuming and error prone. Often due to time limitation, manual back testing can only be done only on a small set of scenarios. It's hardly comprehensive and most important of all, it is always limited by my own knowledge of options and skewed by my own biases. Nevertheless, this manual process has serve me well until now, but I think things will change when the market becomes more competitive and all the low-hanging edges are all gone sometime in the future.

So my point is regardless whether one incorporates machine learning and artificial intelligence into the process, migrating what I call the "options R&D" process from an intuition-based system to a data-driven and analytics-driven system is imperative. Bear in mind, I'm not just talking about automating the back testing process with something like algonet (or whatever you call it) but rather using proper data analysis techniques.

I know many people would disagree with these comments. That's fine. Different opinions makes a market work anyway !
:)
 
#18
I know many people would disagree with these comments.
I wont. On contrary, I agree in 101%. Just proper data analysis and AI are bit different things in my mind though. Creating AI to do analysis e path we are presently using and it seems that there is no alternative. Just AI may become problem itself, not necessary sc-fi style, but most likely because it significantly increase human dependence on technology.
 

EIN

New member
#19
Hi everyone, I share Kevin's enthusiasm, and I also believe that we about to enter an era of AI-assisted options trading. Deep RL could indeed be used to figure out "which type of trades are optimal in changing environment, ie think changing skew, changing IV, pricing, bid/ask spreads etc..", while GANs could be used to create market data as deep learning networks need at least an order of 100K data points to be trained. At the moment, I am exploring LSTM deep neural networks that I think are suitable for skew prediction. Given a current skew for a certain expiration and a percentage change in underlying - can we accurately predict how a new skew would look like?
How did this work out? I am doing some work on AI for optimal options portfolio hedging with/without alpha generation. I'm curious about the predictive models especially for more difficult single stocks
 
Top