Handling a Multi-Asset Model
Handling a Multi-Asset Model
Building a multi-strategy portfolio from scratch - Day 3
👋 Hey, Pedma here! Welcome to this✨ free edition ✨ of Trading Research Hub’s Newsletter. Each week, I release a new research article with trading strategies, code, and much more.
Subscribe to get access to these posts, and every post.
Hi everyone!
After almost 2 weeks working on the business, and mostly on stuff unrelated to what we’re doing here, I am back.
Let’s continue with developing our multi-strategy manager and launch it as fast as possible.
I want to stack a few iterations of the model, before the last quarter of the year, when I expect good odds that the market will give us a new up leg again. By then, I want the model to be fully deployed and tested, for any potential errors and problems.
Last time, we ended our article on the backtesting tab, and we were able to have some basic research plotted in that dashboard.
Here’s the full article if you haven’t read it:
After our last article, as I was turning on the dashboard, it had stopped working, for whatever reason. So that will also be fixed today.
A lot of times we have to deal with these infrastructure errors that are really annoying. This is the reality of launching automated models. It’s not all “passive income” like many larps in the business claim.
It’s actually quite technical, and requires constant monitoring, to understand if our execution models are working as they were designed to.
If you’re going to be a true practitioner of the game, you’ll have to deal with a lot of these issues, that are exogenous to the actual signal research work.
Things break with our code, with accounting, with legal, with compliance, etc.
I’d much rather do research on signal, making my strategies better, my research better, but I have to deal with these issues too.
I could have skipped this part so that you only see the finished product, but this is also a personal log of what I am achieving with my projects. So let’s keep it real, log both the successes, and failures that come along.
This is what we’ve achieved today:
Let’s get into it!
Index
Introduction
Index
Multi-asset Testing
Adding Benchmarks to the Plot
Adding Performance Metrics to the Dashboard
Want to have my own custom assistance, with your systematic trading business?
We help individuals and enterprises, implement and deploy strategy portfolios, tailored to their specific needs.
If you want to discuss the possibility of working with me:
Multi-asset Testing
The first functionality we want to add, is to allow our portfolio to allocate across multiple signals that it gets, from a basket of assets.
We want to be trading as many assets as we can, so that we can get more efficient, and diversified exposure, to the asset class we’re trading. Sure, trading a bunch of cryptos is not a large diversifier, I know that, but it diversifies from individual risk of each coin. One example that comes to mind is LUNA.
So let’s begin with that task.
After fighting with the code on the backend, to be able to take in multiple assets, I got it done.
Now another problem comes up. The “asset” drop down list is only showing the first asset that shows up, regardless if we chose to trade all assets available on the list. I also want to be able to test assets individually, and see their performance for this particular strategy.
So I’ve went ahead and added to our dropdown the option to either pick “All Assets” or individual assets.
Perfect!
We can now run tests for the entire dataset.
Let’s now fix the tables.
After looking into the code, I noticed that somehow I was not passing the data for the tables, so they were not being filled.
This is a problem from ChatGPT. Once the code starts getting a bit more complex, it starts struggling with basic tasks.
What I am using mostly LLM’s for in this project by this point, is to make the code “prettier”. It starts becoming a bit useless, and sometimes adding negative value, after your projects grows. I am still working to find out better ways, so that it can continues to add value, even after a certain threshold of complexity, but it’s not clear to me yet.
But here it is, we have our tables.
Our tables now display the closed trades, currently open trades and our cumulative returns. This is a very important section that we will devote a lot more time to in the future.
Why?
One of the most important things you want to do when you’ve deployed a model, is that you want to maintain a constant vigilance, to how your system is performing, versus how it was modeled. If it deviates too much from what was expected, you need to adjust your model to that new information, and analyze if the strategy continues to offer the risk adjusted returns you thought it had.
So many times traders create these amazing strategies and models, only to find out, that they are not operationally viable once they put them out in the market. This is a big problem of our industry. Strategies are easy to find, but hard to implement.
That’s why you see me say so many times, that all I care about is live performance. Anything else is mostly irrelevant.
Adding Benchmarks to the Plot
Let’s add some benchmarks to our plots.
My initial thought is to use BTCUSDT and the current top 3 coins by market cap. This way we get a broader idea of the performance of the best performing assets thus far. In the future, we might build an index of the crypto market for comparison, but right now, let’s keep it simple.
Beautiful, innit?
Let’s maintain our goal in mind of deploying this as fast as possible.
Many traders, including myself sometimes, get stuck on making things pretty instead of focusing on what makes them money today. Again, it’s all rainbows and sun shines in a simulated environment, but all this work only matters if it produces cash in the bank for us.
Ok but we’re doing some progress.
But we’re still missing something on the dashboard…
Performance metrics!
Adding Performance Metrics to the Dashboard
The way I see it is having a table on the right, with a nice, summarized version of the statistics of the model. This way we can quickly see what the historical performance has been and make simple evaluations.
Just like the video below shows.
And voilá!
We have a few simple performance metrics, plotted on our chart, and I’ve even added the functionality of dragging that around.
We are now getting some shape into our project.
I am going to stop it here for today, and next time, we will start thinking on how to deploy it.
That is where the fun games will begin. Deploying a model is a beast to deal with. We have to deeply think about stuff like:
Inventory management
Risk management
Risk controls
All will be done live on these articles.
I hope you’re here for the ride.
Also if you’re looking to build your own systematic trading business, and don’t know where to start, or if you’re a company struggling to get a decent model performing, feel free to book a call with me, and let’s see how we can work together. I work with both individual and institutional clients, helping them with their systematic trading business.
If you want to discuss the possibility of working with me:
Disclaimer: The content and information provided by the Trading Research Hub, including all other materials, are for educational and informational purposes only and should not be considered financial advice or a recommendation to buy or sell any type of security or investment. Always conduct your own research and consult with a licensed financial professional before making any investment decisions. Trading and investing can involve significant risk of loss, and you should understand these risks before making any financial decisions.