[FRIAM] github for databases

[Jon Zingale]

Ok, maybe not complete trash, but I am finding that I have to be more
careful about what data is there and what questions I can ask of it.

For instance, the spot dates don't always correspond to an option date in a
nice way, which adds a lot of noise when trying to cobble together
meaningful time series. While the lags aren't awful:

Lag (days) -> count of observations:
  0: 217
  1: 95
  2: 16
  3: 3
  4: 2
  8: 1
 11: 1

 They are bad enough to watch.

In the chart I offered, for instance, the last midpoint call price
corresponds to a date far enough away from yesterday to be misleading.

I am sure there are some other nasty holes like that. In the meantime, I am
trying to find better questions to ask. One question is how well delta acts
as a proxy to probability, in practice. So far the evidence is that it is
not very good at all for this data set until one gets out to about 75 DTE:

~0.5-delta calls empirical ITM: 0.22580645161290322
tenor=75: delta≈0.25: empirical ITM=0.500
tenor=75: delta≈0.35: empirical ITM=0.333
tenor=75: delta≈0.45: empirical ITM=0.500
tenor=75: delta≈0.55: empirical ITM=0.667
tenor=75: delta≈0.65: empirical ITM=0.500

Then it gets ok for a bit before getting bad again.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20260121/5a962644/attachment.html>