Napoleon Hernandez’s Post

Most people should stop writing RESTful APIs for financial data. I think this trend is driven by three reasons: 1. Many newer fintech firms were inspired by the crypto exchange wave and just copied these crypto exchanges' REST APIs blindly, without having seen best practices in mature asset classes. 2. There's a lot of tooling, standards, and support around RESTful APIs, with OpenAPI/Swagger, Redoc, etc. so it's an easy default. 3. There are many high quality, openly-documented web APIs based on REST, like Stripe and Twilio SMS, so it's easy to copy them blindly too. Folks on Databento's API team, myself included, have worked with many internal APIs at top trading firms and proprietary vendor APIs, and we've rarely seen RESTful APIs in use aside from powering simple dashboards or click-trading interfaces. REST at its core is about manipulating entities. For example, a use case that's mostly CRUD on elementary resources, like a dashboard backend API for adding users, deleting them etc., is a very good fit for REST. Idiomatic REST APIs avoid verbs in their endpoint paths as HTTP verbs like GET are sufficiently expressive for basic CRUD operations. But more abstract mutations like subsampling, space, merging—which are essential in electronic trading—are hard to express in this manner. In the words of my colleague Renan Gemignani: once you start doing things involving multiple resources (e.g. datasets, venues, symbology type, schemas, formats, symbols) at once and across multiple domains (e.g. resolve symbols, fetch this range and merge these CUSIPs and then submit this batch job with output symbols remapped to native exchange IDs), your REST endpoints become either unnecessarily complicated, too rigid, or unexpressive. It's awkward to force entities like schemas (top-of-book, MBO, OHLCV, etc.), symbols, dates, etc. to become resource-oriented. Neither do they really behave as simple sorts and filters that typically get treated as query parameters under RESTful convention. What are some indirect symptoms of your vendor fighting REST conventions? I've seen several, e.g. - Their API forces you to query 1 symbol or 1 date at a time. - Their API has a much larger API surface than a raw feed's wire protocol. - Their API introduces nearly duplicate interfaces to deal with different asset classes or multi-symbol and full-venue scenarios. - Their API has a disorderly, deeply-nested structure of URL parameters. - It's unpredictable how more abstract mutations like subsampling, space, merging are expressed. In our case, we opted for a more flexible, RPC-style HTTP API. A good HTTP API doesn't have to be RESTful. Slack's API is a great example of a HTTP API that isn't—it's based on a JSON RPC style instead. gRPC has a fallback protocol based on protobufs over HTTP. The end result is an API with a small surface of orthogonal methods and very little overlap between them. #restapi #http #apis #marketdata #algotrading #crypto

easy peasy

Key Performance Indicators KPIs are the critical (key) quantifiable indicators of progress toward an intended result. KPIs provide a focus for strategic and operational improvement, create an analytical basis for decision making and help focus attention on what matters #BigData #Analytics #innovation #tech #KPI #Cloud #KPIs #DigitalTransformation Credits: Dr. Joerg Storm

This one took a while to put together but here it is: A list of top investors in generative AI today. A few things to call out: -- Obviously, this is just a select list. There are many other great investors worth mentioning but there's simply not enough room. Note that I have also left out accelerators/incubators. -- This is not meant to be 100% accurate or comprehensive. My apologies if I have inadvertently left out any companies or made a mistake. -- The sources for this list include Pitchbook, Crunchbase, and investor websites. Lastly, let me share a little bit about my firm Translink Capital and our investment focus: -- While we aren’t featured on this list today, we are actively exploring investment opportunities in this space. We prefer to invest at the Series A stage, although we will occasionally participate in seed or later-stage rounds. Our firm is backed by 30+ Asian multinationals, and we have a proven track record of connecting our portfolio companies to Asia, among other things. If you think there might be a mutual fit, feel free to get in touch. [UPDATE] If you would like the PDF, I have attached it to a new post. Please feel free to download directly.   #artificialintelligence #generativeai #startups #venturecapital #VC #AI #ML #machinelearning

An updated must-read version of Michael Lones' amazing paper "How to Avoid the Pitfalls of Machine Learning: A Guide for Academic Researchers" "This document is a concise outline of some of the common mistakes that occur when using machine learning, and what can be done to avoid them. Whilst it should be accessible to anyone with a basic understanding of machine learning techniques, it was originally written for research students, and focuses on issues that are of particular concern within academic research, such as the need to do rigorous comparisons and reach valid conclusions. It covers five stages of the machine learning process: what to do before model building, how to reliably build models, how to robustly evaluate models, how to compare models fairly, and how to report results." Paper: https://lnkd.in/dyp7iEZJ #machinelearning

A Brief Introduction to Machine Learning for Engineers Are you looking for a LinkedIn post based on this text? If so, here's a suggestion: Are you looking to learn more about the key concepts and algorithms in machine learning? Check out this monograph, which provides a concise yet comprehensive introduction to the field. With a focus on probabilistic models for supervised and unsupervised learning, this resource offers a unified notation and mathematical framework for building on first principles and exploring more advanced topics. Whether you're an engineer with a probability and linear algebra background or simply interested in learning more about machine learning, this monograph is a great starting point. #MachineLearning #DataScience #Engineering #AI