Ok so maybe I missed it, but the last thing I remember was that your data don't identify combination orders, but that you use heuristic methods (timestamp) to identify them. Now I'm a bit confused by your response: When you say "one trade" and "two trades", is that identification made (implied) by the same-millisecond assumption, or do you have a way to actually identify them directly from the data? Because in the latter case, you could obviously 100% identify combination orders and by implication box spreads. If you could please clarify, thanks!No way for me to tell that apart unfortunately. All spreads on the site are matched on two rules, one trade with all four legs with the same timestamp with lowest possible granularity (iirc it's like one thousand of with a second or ms) or two trades of two legs each within 1 second because many of my spreads get broken out into separate threads and auctioned individually. I've limited it to two separate trades though to avoid issues with 4 single legsI see for example boxes Sep 2024 5255-5260 @ 3.49% and Sep 2024 5260-5280 @ 7.45% on 08/09. That makes no sense. The strike prices of those "boxes" seem to be close to the index value.
@adamhg: Can you possibly verify that these trades are not mistakenly interpreted as box trades, when in reality they were just individual legs traded independently at slightly different times while the market fluctuated?
I guess that might also be a possible explanation for the 4000-5000 "boxes" off the trend line; but it's hard to tell without having the details of the algo used to identify box trades.
How do you know if your spread orders get broken out and auctioned individually? I only ever saw fills with an implied cost basis (based on an algo defined by the CBOE) for each individual leg at IB and at Schwab, with no information on how it was auctioned or filled. Also, I thought the entire benefit of a combination order is that it is auctioned in its entirety in a separate order book (unless there are standing limit orders for each leg, or perhaps standing combo orders, that match the limit of the box order, which I think would be highly unlikely because the bid/ask difference of legs are much wider), especially for box spreads which have no delta exposure and therefore little volatility and therefore presumably narrower effective spreads.
Statistics: Posted by comeinvest — Mon Aug 12, 2024 6:45 pm — Replies 620 — Views 88116