Hi @fan27, For backtesting I just evaluate a selected trading algo for each candle. It's a simple loop that iterates through the candles and calls the algo interface. I got a bit of generic C++ template code there to make sure it's all compile-time optimized (i.e. no virtual calls), which is important for parameter optimization performance where this code gets executed a lot. Also the code is thread-safe to kick off tons of backtests to all cores on my PC and of course it's designed so that the same algo code works for parameter optimization, backtesting & live trading. I'm adding OpenCL support though to be able to run parameter optimization on GPUs to cut down the optimization time further. For candles I store only 5sec candles on disk but can create different aggregates in run-time to use with the algos. I also recently added custom candle compression which cuts down the data size on disk to ~4.5% of uncompressed Polygon JSON data or 5.5bits/candle (OHLC, VWAP, volume & trades data), which is good for storage and when I need to move the data around.
He is likely trading futures on the CME, so his VPS in Chicago gives him ultra low latency to execute his orders with no delay. If you are running any type of automated strategy, it's a must... We offer services in both Chicago and New York (depending on who you are connected to, and what you are trading futures, equities, forex, or crypto). If you aren't sure what the best location for you is, ping us via live chat on our website so we can better assist you.