In it’s simplest form, back-testing is the process of attempting to verify how a given investment strategy would perform in real-world conditions, without risking real $.
Let’s say you have a hypothesis that a perfect time to buy is when the 20 day moving average (DMA) crosses the 50 day moving average to the upside. You could go out and trade this immediately, but a back-test will help you rule out obviously bad strategies.
I highlight those specific words because there isn’t a full-proof manner to ensure your trading strategy will end up in the black. However, a good back-test…
There are many NLP models floating around, and my favourite framework for dealing with NLP models is Spacy. The out of the box models are great for general well-written English which is great; but poses a problem when we start dealing with Reddit / Twitter :D
Firstly we have hashtags, and company symbols (e.g. $GME). Many frameworks do not have native handling of these. Generally the lowest common denominator is the ‘token’, generally characters grouped with spaces demarcating them.
One framework might pre-process the data so that $GME gets transformed to GME so they are equivalent tokens when training. Others…
There are various types of quantities in financial systems, and the blindness to its complexity can cause a tailspin as systems become more complex. The goal of this article is to give readers more perspective as to why such a simple idea can become a complex topic.
Cash equities (a.k.a common stocks) are a good place to start because they are quite simple.
Any investor can go onto a brokerage’s website and buy 1 share of AMZN at around $3300 as of Feb 10th. Simple.
Cash bonds are still quite simple. First, let’s try to re-use the concept from equities…
There can never be a singular ‘correct’ way to view financial products used by banks, asset managers, pension funds, etc, but at some point everybody has to code it somewhere into their system(s).
Here is a stab that may be useful. You can ignore the root node. Beyond that, the first row is the ‘asset class’, and the second row is the ‘product type’.
The bottom rows refer to whether something is referred to as a cash instrument or a derivative. Generally meaning:
Following on from:
In previous posts we talked more about getting data than analyzing data. I firmly believe the bedrock of any successful technology product is the ability to easily get access to the data you want in an easy, scalable and secure manner. This allows you to build automated solutions easily.
However, there are times when you need to explore the data more fluidly:
This follows on from:
After thinking through your API experience, you will naturally end up with a set of services that are generally aligned with key datasets. For each service you will need to pick the best technical architecture. This story talks through some guiding principles that will help you on that journey.
Firstly a disclaimer. There is no singular technology implementation across the industry for all these datasets (at least yet), so this can be subjective on the edges.
Try to have some organizational goals to help keep everyone on…
This follows on from the first post on business strategy ~= data strategy
Let’s assume your business strategy aligns with exposing data, building APIs, or building a platform.
Some business strategy examples for a trading exchange might be:
Regardless of which opportunities you want to pursue, there is…
There is an inherent link between your data strategy and your business strategy, in that to quickly execute your business strategy, your data needs to be in great shape. So therefore — ideally — you know your business strategy already and you can leverage that knowledge to prioritize which dataset to improve first.
Here are some example business strategies within the financial industry:
I really struggled to title this because I like to mix economics, markets and technology into a single article. The goal of this article is to discuss how correlations come and go, provide a couple concrete examples, and then discuss how to mold that into your technology strategy. Given FinTech is probably still more art than hard science, the last part becomes subjective!
Every time I read an artificial intelligence FinTech article I always look to see if the model has incorporated this bit of wisdom. The biggest challenge is that markets are driven by human behavior which evolves over…
Many successful companies are now built as an API-centric SaaS model, sometimes able to deliver an API that solves a problem and its so good people are willing to pay for it.
In finance, we are still in the beginning of the journey of this way of looking at the world. Where there has been changes in the last 10 years, this approach has been adopted where possible but it is far from prevalent. APIs exist for market data, regulatory reporting and in some cases for aspects like trade execution. But. It is far from a prevalent practise.
Spent career at various financial institutions across continents and across the buy/sell side. Content/thoughts are my own, not my employers’