Hola,
Long time lurker, you might've encountered my posting about this somewhere. Or you may be on something like the Unusual Whales discord and seen it. Anyway. I've been investigating a potential statistical anomaly in behavior of day-before options flow predicting the directionality of earnings results, specifically in the after-hour/pre-market session that the earnings occurs. I detail the proposed effect here, just to spare rehashing:
It turns out that at least since Mar 2019 on SPY tickers, this allows us to predict directionality of earnings roughly 70-80% of the time, versus 50.88% for the mean null hypothesis (that is, assuming stock price increases always).
I still am extremely skeptical about our results, but right now our biggest issue honestly seems to be data integrity.
—
On our website (thenope.info) we use Omnieq for EOD historical data, and honestly for the price ($40/mo) they do a great job. It only goes up to Mar 2019 which sucks, but I'm a student so it's not like I can afford big boy money to test this to get the data we need.
That said, for live data, we end up using a brokerage API, and here's where it breaks:
While Omnieq and our brokerage both return roughly the same option prices, volume, and share volume, the net delta is dramatically different. How different? For INTC yesterday, this discrepancy led to an incorrect bullish prediction yesterday for ER, while Omnieq's data end of day reflected a bearish prediction (and it dropped 10%, of course). All in all, among all option chains we measured, between our broker and Omnieq it was a 5 million delta discrepancy, which goes from 'acceptable error' to 'holy shit this is bad'.
We investigated today on some data dumps, and it turns out Omnieq and the broker reflect different implied volatilities, which likely is the reason the deltas (and therefore the net options pricing effect, which is net option delta/share volume) are so ridiculously different.
—
So, we're trying to figure out a way to do this without having to resort to $$$, especially since we envisioned this as a mostly free tool for traders to use (maybe freemium eventually if we have to pay for data and things). The major issue seems to be that the deltas between our live and end of day data are different, which is likely reflected or just caused by the implied volatility used by the providers being different as well.
We have:
– Option price (they agree)
– Option volume (they agree)
Is there any standardized method we could use to get a unified delta between both of them? Or, is there any service with decent data integrity that isn't terribly unaffordable (<$1,000/mo) that has live and historical data (ideally past 2019)?
Thanks!
Submitted October 23, 2020 at 10:04PM by the_lilypad
via https://ift.tt/2HloS5b