I trade weekly and monthly options and often don't grasp how market prices earnings dates, prior to being officially announced. Obviously historical dates are very useful but still until it is confirmed, where does the confidence come from? For example, AMD IR website doesn't seem to have a confirmed date yet, Q2 earnings was 30 july. Last 3 years, AMD reported Q3 on 24 Oct, 24 Oct, 20 Oct (2nd last week of Oct) (http://quarterlyearnings.amd.com/financial-information/quarterly-results) Extrapolating, that falls into Oct 21-25. Yahoo Finance is estimating 22 Oct https://finance.yahoo.com/calendar/earnings?day=2019-10-11&symbol=AMD Earningswhisper 23 Oct https://www.earningswhispers.com/stocks/amd The options are saying earnings fall into week of Oct 28-1. If I were to model options without calibrating to implied earnings, things would have been a disaster and my arse stuffed by gamma. But then, where does the Oct 28-1 earnings confirmation come from? What I see is historically Q2 earnings were announced 1 week earlier, so Q3 also 1 week earlier. This year it is a week late (last week of July vs 2nd last week), so market is expecting last week of Oct (vs 2nd last in previous years). That looks reasonable and fine as the estimate is calibrated to this year earnings dates vs Yahoo! etc using previous years dates. But still, a few days early or late is possible, how can the market be so confident it will indeed be end of Oct without official announcement. My point is, FY19Q1 reported in the last week (30 apr) when previously in 2nd last (25 apr FY2018-2017), so if the market were pricing using the historical week the Q1 earnings vol must have been messed up by a week. I don't have historical option data to check.
On my screen I see OCT 18 - 44% OCT 25 - 48% NOV 1 - 64% NOV 8 - 61% The market is not certain. Some suckers are still bidding on the OCT 25 expiration. It's pretty easy to figure out the earnings date and the risk premium for betting on it with good data, a bit of programming and some discretionary work.
You answered your own question here: “How can the *market* be so confident” So it’s the “market”. And this doesn’t apply to earnings but to all options pricing. So market makers may just be responding to market demand, but since they may have a stake here, they may also make own projections - that’s still the market. Though whoever plays the main role here, I’d assume that it wouldn’t hurt “them” to sell earlier options for as much as they can, with the possibility of IV imploding if the earnings date gets moved to a later date - quick & easy profit if this happens. So no one selling options really needs to project very specific earnings date, just shoot for an earlier possibility. While only people buying options should wait until confirmation.