You are assuming a total cause and effect relationship where there is only a partial cause and effect relationship:
"There is a stream of orders and, as a result, quotes change. That is what I meant by "trades trigger quotes". So you are confirming that in the listing, it may be common (usual?) to see a quote first and then the trades (even though it was these trades that caused that quote)."
That's not true. Quotes change for two reasons:
1. Because an order has filled which satisfied the bid or offer. Note, though, that a given trade might or might not affect the quote. Depending on the structure and rules of a given market, trades may occur between parties who have not contributed to the quote - and thus, the quote will not change in any way. (For example, NYSE trades where neither side was against the order book.) And, these trades may or may not have occured at prices at, better than, or away from the quote.
2. Because traders (who may include, depending on the particular market's structure, public and institutional investors, market makers, and specialists) make decisions about the transactions they are willing to enter into. We don't know and can't determine why traders make a given decision. A trader could pull an existing bid or offer, could modify a bid or offer - in response to trades - or for any other reason, could rescind a bid or offer, could make a new bid or offer, etc. All for any one of various reasons which may or may not be responsive to order flow and price.
"10 years ago, the quote printed first and then up to 5 seconds later, the trades that caused this quote were listed, according to some research I've seen.
How much has this changed now?"
With caveat about the casual relationship - this has changed a lot. I don't think there is any significant lag today. Certainly not the several seconds to a minute or two it once was.
But quotes and trade reporting still are, generally speaking, seperate data streams, often handled by seperate computer systems. So, for many markets, quote data and trade data are recorded and/or reported asychronously to each other, and you cannot determine definately in which order they occured.
Perhaps you are no really so concerned with cause and effect as you are with sequence? That is, perhaps your need is really to know, given a particular quote and a particular trade, which came first - the chicke... er, the quote or the trade?
Note that data is often "time stamped", but there isn't a great deal accuracy or consistenty in this. I understand there are some rules on trade reporting, but there are several seconds of leeway allowed. The advantage of the timestamp is that, although real-time data may lag, at least it is possible to determine that it IS lagged, and then it is possibly to align data (such as when doing a study such as yours) to at least get a rough idea of adjacency (if not precise sequence).
FWIW, Island and Archipelago ECNs timestamp quotes with 1mSec precision (and unknown accuracy), other ECNs and exchanges typically timestamp quotes with 1 second precision (and unknown accuracy, but presumably with 1 second accuracy, as that is so easy to acheive). I am not familiar enough with timestamps on trade reports to comment on them. |
