Corrupt data?

[Mark17]

I've started backtesting in 2007 with ONE and I have issues with the data. Please take a look at the "Capture" screenshot. This is roughly 1% ITM/week compounded geometrically and there's no extrinsic value showing. In fact, there are is a range of strikes there with no extrinsic value. Zero. That doesn't feel right to me--even DITM--with four weeks to expiration.

If you look OTM, in the "Capture3" screenshots, puts have extrinsic value well into the 3rd SD.

Now look at the "Capture2" screenshot from 2017. There's a much more reasonable distribution of extrinsic value across these options. That, to me, feels right.

I used to use OV and I'd sometimes have the same issue with their data. It didn't seem to be every day, certainly, but before 2008-9 or so, it did seem somewhat prevalent.

What might be going on here?

 

[gariki]

looks like bad data.

 

[Marcas]

Surely it is bad data.

All data I know do have issues. If what you said happens often I'd be worry about using ONE's results blindly. It also depends what you use it for.

In general,
if this is a rare issue (for what I remember ONE's data weren't that bad, but also I didn't use ONE extensively),
I'd just suck it up.

This is not a real problem unless you do need that particular data for your study.

I believe ONE do their own iv calcs, if so, they should be able fix your dates if you ask, but they may not, depending on the system they use.
You should ask ONE if they apply any algorithms to data anyway You should know that stuff if you want to use ONE for anything more serious than playing around.

 

[Shin]

Mark, you're looking at SPX EOD data. These quotes, definitely back in the days, used to be quite random. So this is not necessarily a ONE issue, but more likely the data that came from the exchange. The broader ramification is that you may want to re-consider backtesting based on EOD data. Unlike these quotes that caught your attention, there are likely many other quotes that look innocent, hence you can't realize they are off, sometimes way off. I typically backtest based on 5min before close, thereby reducing the sensitivity to irrelevant prices. ONE has intraday data starting on 10/1/2010.

 

[Mark17]

Marcas: it's not just a day or two. From Jan 2007 through May 2008, there's some corrupt pattern in the ITM/DITM puts for the front two months almost every day. I commonly see zero extrinsic in strike ranges where I would expect to see positive values (especially with weeks to expiration). Another thing I commonly see is identical extrinsic value over a range of strikes. Extrinsic should plot like an inverse V, bell curve, or something roughly approximating one of those symmetric about the ATM strike.

Shin: that's a very good point. Unfortunately, until some date, the only data provided is EOD. In OV, I would typically run my EOD backtests using 15:30. ONE doesn't offer that until the intraday data begins. This may mean I can't use that data for backtesting at all.

One other thing occurred to me. OI is thin to none across many of these DITM strikes. I wondered if many of these are stale quotes from inception because they are hardly traded?

In fact, I really have no idea how these DITM strikes trade. I haven't done a whole lot of classic butterflies to know better. Premiums can easily range from $40 to $140 on the options I've been looking at. Does anyone else have experience trading DITM options like these? What kind of slippage can be expected?

In ONE, I was taking the slider bar and moving it 60% of the distance from the mid toward the natural price (five hash marks are given between mid and natural. I was sliding over to the third). This was typically $0.60 with SPX around 1200-1600 for a single option.