vincent wrote: ↑Fri Apr 26, 2019 11:37 am
Sierra Chart has recently introduced the capability to support up to 200 Levels of Market Depth with it's own Sierra Chart Exchange Data Feed.
Please allow me to comment on this regardless of the discussion on adapter. I'm surprised that such updates are presented as achievement 2 years after CME switched into full depth market-by-order MBO data. This means that at early stages someone made bad technological decisions based on wrong assumptions. Namely, using indexed arrays to store order book data instead of maps (e.g. hashmap or treemap). The actual number of price levels can be many billions (10^9). For example, this link gives you the latest order book of BTC/USD at Bitstamp
https://www.bitstamp.net/api/v2/order_b ... sd?group=2
When I clicked on it, the lowest bid price was 0.01 and the highest ask price was 1,000,000,000 (see the last entry at the end, it looks like this: ["1000000000.00", "0.00200000", "138268380"]). This means there are
One Hundred Billion price levels in steps of 0.01. This doesn't mean that each level is always occupied. And it doesn't require a billion times more traffic to handle than, for instance, a market with 100 price levels (50 on each side) because most of the activity happens on the close to market price levels. I's also easy to demonstrate that a regular laptop can process ~5 millions of data updates of unlimited depth order-by-order data faster than processing market-by-price data (where only the number of orders is known) in DTC or similar format. This is because it becomes very costly performance-wise to insert/remove elements in indexed arrays when they become large/long.
https://www.dtcprotocol.org/
I was looking to keep costs down for me and other traders