#1

Market Data API Choice

Starting off with a free API, irrespective of whether its worldwide (should cover the US) for now

Choices:

API	API Calls	Historical Data (yrs)	Delay
Polygon	5 / min	2	Real time
Finnhub	60 / min	1	Real time
Alpha Vantage	25 / day	20	15 mins
Alpaca	200 / min	From 2016 so ~9	15 mins

The API choice doesn’t matter that much, super interchangeable, lets go with Alpaca for now as it’s only downside is its delay which doesn’t affect anything much right now as I’m only paper trading

Languages

Data retrieval/handling pipeline is going to be coded in Python

Core execution engine: I could use either C++, Go, or Python I’m going to go with C++ as I would like to build experience with it, and wayy more importantly, its super super low latency.

#2

Integrating Alpaca’s APIs

Market Data API: To retrieve market information
Trading API: To place orders

A couple of things I’ve just discovered:

Alpaca’s free tier only provides real-time market coverage (This refers to the source of your live, streaming price data) from IEX

<aside>

”IEX (Investors Exchange) is just one of many U.S. stock exchanges., you are only seeing the trades and quotes that happen on IEX. While this data is real-time, it represents only a small fraction of the total U.S. market volume (around 2-3%). For many stocks, this is perfectly fine for testing. However, you will not see the full picture of all buy and sell orders across the entire market.”

</aside>

The free tier only allows for 30 ticker symbols (ie, a max of 30 diff stocks/ETFs)

Limitations

Just looked a bit more into the downsides of having only data from IEX on my algo trader and a big one is “Do not assume a profitable backtest on IEX data will translate to a profitable live algorithm.”

It might cause the following issues (according to Perplexity (Gemini)):