Volume and OI: some help needed

nonprophet · Jul 26, 2006

Two scenarios that play out regularly re. option volume and open interest data.

I draw my conclusions and would like to know if these are correct. It all seems pretty straightforward but perhaps I am overlooking some triviality. Anyone willing to take a look and comment: thanks in advance.

All numbers refer to strictly one option series (one particular contract, put or call, that doesnât matter).

One note:
Open interest of course is reported the next day. So âDay 1 OIâ means open interest as reported on Day 1, which represents open interest of Day 1 BEFORE market open. âDay 2 OIâ represents open interest as reported the next day, which is equivalent to open interest of Day 1 AFTER market close. Just to avoid confusion.

Scenario 1:

Day 1:
Volume = 200
OI = 1000

Day 2:
OI = 1300

Conclusion:
This is impossible. Volume 200 cannot create an OI change of +300. These data MUST be wrong.

Scenario 2:

Day 1:
Volume = 200
OI = 1000

Day 2:
OI = 700

Conclusion:
This is possible, assuming at least 100 contracts were exercised after market close on Day 1. These would represent the âmissingâ -100 OI change that volume does not account for. (btw in that case all 200 Volume must have been CLOSing positions).

Of course, these cases refer to any day in an option life except expiration day. Also, we donât assume contract adjustments due to stock splits, dividends etc.

Basic stuff huh? Willing to take a beating if Iâm missing the obvious. Thanks!

Option Trader · Jul 27, 2006

Quote from nonprophet:

Scenario 1:

Day 1:
Volume = 200
OI = 1000

Day 2:
OI = 1300

Conclusion:
This is impossible. Volume 200 cannot create an OI change of +300. These data MUST be wrong.

Perhaps OI is expressed as net difference for position of specialist/MM. If this is correct, then these are my thoughts: If one retail trader bought 100 options and another retail trader sold 100 options, and the specialist is in the middle, I would think the net affect on OI is zero. But the moment the specialist exercises the ones he bought against a retail trader, then it would seem to me the net difference would be an increase of 100 on OI.

nonprophet · Jul 29, 2006

Quote from Option Trader:

Perhaps OI is expressed as net difference for position of specialist/MM.

Thanks OTrader for sharing your thoughts on this. But I think the premise of your argument is not correct: In calculating open interest (and volume) they donât distinguish between whichever party is involved.

It took me awhile to find some confirmation on the web. Oh, lots about it from Bernies & Larrys, like how âdecreasing open interest with increasing volume invalidate the trendâ or something like that.

Finally I found some lesser-known guys who actually seem to understand the concept:

âWhen the change in volume is less than the change in open interest, we classify it as an exchange error. They're classified as errors because the volume can't be less than the change in open interest. The official response from the Exchange is that accounting/clerical errors certainly accounts for a portion of this anomaly, but since no other explanation has been offered by the exchange, apparently all of these differences are due to errors. It is disturbing to see that the exchange's error rate can be as high as 16% of the daily traded volume of calls or puts. That means 1 in 6 trades was errant.â

1 in 6 ???? Ah well, that's within the margin of error for reliable charting...

Still think they overlook the possibility of exercise. I emailed âm about that.

Thanks again for your input, it apparently takes some courage to comment on a simple issue like open interest.

Cache Landing · Jul 29, 2006

Quote from nonprophet:

Thanks OTrader for sharing your thoughts on this. But I think the premise of your argument is not correct: In calculating open interest (and volume) they donât distinguish between whichever party is involved.

It took me awhile to find some confirmation on the web. Oh, lots about it from Bernies & Larrys, like how âdecreasing open interest with increasing volume invalidate the trendâ or something like that.

Finally I found some lesser-known guys who actually seem to understand the concept:

âWhen the change in volume is less than the change in open interest, we classify it as an exchange error. They're classified as errors because the volume can't be less than the change in open interest. The official response from the Exchange is that accounting/clerical errors certainly accounts for a portion of this anomaly, but since no other explanation has been offered by the exchange, apparently all of these differences are due to errors. It is disturbing to see that the exchange's error rate can be as high as 16% of the daily traded volume of calls or puts. That means 1 in 6 trades was errant.â

1 in 6 ???? Ah well, that's within the margin of error for reliable charting...

Still think they overlook the possibility of exercise. I emailed âm about that.

Thanks again for your input, it apparently takes some courage to comment on a simple issue like open interest.

I didn't comment because I figured someone who could explain things more eloquently would.

I believe you are right. If someone writes a call/put (i.e. doesn't currently own one but sells one to someone else) the OI will increase by 1, as will the volume. If the person who bought it then exercises it the same day, the OI will decrease by 1, but the volume will remain at 1. In this case the volume would only be 1 because the person exercising is not counted in the volume. He did not buy/sell a contract, he merely redeemed his contract at current value. So the change in OI is 0 and the volume is 1.

OTOH, if there are 100 buy/sell transactions on a given day, and in addition to that, 200 contracts are exercised which were purchased at an earlier date, the change in OI is 200 while the volume is only 100. This situation could potentially go unnoticed for the most part as one of the only reasons to exercise early is to claim a dividend, and even that reason is debatable. So you might see this occurring very close to expiry or in conjunction with a dividend. Most of the time traders will merely sell the contract to someone else and keep the additional time value.

nonprophet · Jul 29, 2006

Quote from Cache Landing:

I didn't comment because I figured someone who could explain things more eloquently would.

Cache Landing, both eloquence AND a thorough understanding of the subject may be hard to find

As far as Iâm concerned you are eloquently hitting it right on the nose by describing exactly my line of reasoning as well.

Meanwhile I also received e-mail confirmation from the website that indeed early exercise may account for OI/volume discrepancies, but obviously only in case of OI DE-crease (when this is larger than volume). OI-INcreases surpassing volume remain clear-cut errors so far.

BTW the site Iâm referring to has a gigantic amount of data crunching under the index option section, definitely worth a look:

http://www.smallinvestors.com/SP500/indexoptionstoc.htm

Having pretty much covered both scenarios I will do some research on bulk option data (both index and stock options) to get a handle on the amount of errors. I will try to incorporate dividends to see if early exercise is plausible at all (and exclude Euro-contracts from early-exercise. Do I forget something??).

Of course we can only detect such flagrant errors where Volume does not account for the OI change; safe to assume there are more errors that we are unable to detect. Perhaps exchanges might as well refrain from reporting option volume altogether!

Probably ainât that bad. Iâll post some results.

I could use some help with these issues:

- Am I correct that (early) exercise data is NOT made public, so this information is not available anywhere?
- am I correct that the ONLY U.S. Euro-style options are ALL index options except OEX?

Thanks for thinking along!

Option Trader · Jul 30, 2006

Quote from nonprophet:

Thanks OTrader for sharing your thoughts on this. But I think the premise of your argument is not correct: In calculating open interest (and volume) they donât distinguish between whichever party is involved.

Thank you much. Here's my difficulty with this: e.g. if you sell me 50 options, and I unload them a week later to another retail trader, the open interest would change by 100, even though there is only a net of 50 out there!!

Option Trader · Jul 30, 2006

Meanwhile, this is what I found from Investopedia, quoted elsewhere:

Investopedia Says: 1. A common misconception is that open interest is the same thing as volume of options and futures trades. This is not correct as demonstrated in the following example:

On Jan 1, A buys an option, which leaves an open interest and also creates trading volume of 1.
On Jan 2, C and D create trading volume of 5 and there are also 5 more options left open.
On Jan 3, A takes an offsetting position and therefore open interest is reduced by 1, and trading volume is 1.
On Jan 4, E simply replaces C and therefore open interest does not change, trading volume increases by 5.

nonprophet · Jul 30, 2006

Quote from Option Trader:

Thank you much. Here's my difficulty with this: e.g. if you sell me 50 options, and I unload them a week later to another retail trader, the open interest would change by 100, even though there is only a net of 50 out there!!

Not so, this operation would not change OI at all, it would only create Volume of 50. It would be an example of what happens on Jan 4 in your Investopedia example.

Say we start the day with OI = 100 ( = # of longs = # of shorts).
Volume for the day is 50.
On the one âextremeâ it could be that those 50 were all closing transactions of already existing contracts, so by the end of the day OI will have dropped to 100 â 50 = 50.
On the other extreme it could be that those 50 are all new contracts independent of the 100 that were already there, then by the end of the day OI is 100 + 50 = 150.
In the middle it could be that one side of existing 50 were transferred to other parties but the contracts basically stayed âin placeâ, then weâd end the day with still OI = 100.

In all cases Volume for the day is 50.

Any kind of combination of the above transactions (within Volume 50) would result in OI between 50 and 150. What is NOT possible is OI > 150 or < 50 (except the latter in case options were exercised). In other words: the CHANGE in OI cannot be greater than Volume (except ....)

Without a doubt OI is the reliable figure and volume is the culprit, see

http://www.smallinvestors.com/SP500/ioExchangeErrors.htm

I would not be surprised if only one or two exchanges continuously mess up their volume data. I think we deserve correct data.

Any thoughts on these questions?:

- Am I correct that (early) exercise data is NOT made public, so this information is not available anywhere?
- am I correct that the ONLY U.S. Euro-style options are ALL index options except OEX?

freehouse · Jul 31, 2006

Quote from nonprophet:

Two scenarios that play out regularly re. option volume and open interest data.

I draw my conclusions and would like to know if these are correct. It all seems pretty straightforward but perhaps I am overlooking some triviality. Anyone willing to take a look and comment: thanks in advance.

All numbers refer to strictly one option series (one particular contract, put or call, that doesnât matter).

One note:
Open interest of course is reported the next day. So âDay 1 OIâ means open interest as reported on Day 1, which represents open interest of Day 1 BEFORE market open. âDay 2 OIâ represents open interest as reported the next day, which is equivalent to open interest of Day 1 AFTER market close. Just to avoid confusion.

Scenario 1:

Day 1:
Volume = 200
OI = 1000

Day 2:
OI = 1300

Conclusion:
This is impossible. Volume 200 cannot create an OI change of +300. These data MUST be wrong.

Scenario 2:

Day 1:
Volume = 200
OI = 1000

Day 2:
OI = 700

Conclusion:
This is possible, assuming at least 100 contracts were exercised after market close on Day 1. These would represent the âmissingâ -100 OI change that volume does not account for. (btw in that case all 200 Volume must have been CLOSing positions).

Of course, these cases refer to any day in an option life except expiration day. Also, we donât assume contract adjustments due to stock splits, dividends etc.

Basic stuff huh? Willing to take a beating if Iâm missing the obvious. Thanks!

How about this one. After trading day 1 the underlying stock had a 2 for 1 stock split. Open interest doubles. Share price is halved. There was also exercise activity at the end of day 1.

nonprophet · Jul 31, 2006

Quote from freehouse:

How about this one. After trading day 1 the underlying stock had a 2 for 1 stock split. Open interest doubles. Share price is halved. There was also exercise activity at the end of day 1.

Good point freehouse. If thereâs no symbol change like in 2-1 splits, OI will double. Additional exercise will in fact somewhat lessen the damage but the thing is messed up for sure. (If thereâs a symbol change like in 3-2 splits this wonât happen because these are essentially new contracts).

Contract adjustments are an absolute nightmare for bulk option data crunching. I use a very big hammer to remove any and all classes that have multiple root symbols or weird strike decimals like 0.375 or 0.625. Itâs just not worth looking into and âadjust for adjustmentsâ.

I looked up some recent 2-1 splits and found both MON and MRVL removed for any of the reasons above. However INFY (split on Jul 18) slipped through and indeed causes an OI/volume error. Now in my case I have a stock chart by default so I can right away confirm that there has been a split (I use non-split-adjusted charts exactly for this reason).

While I donât think that adjustments can account for the assumed huge amount of volume errors, I do appreciate you bringing this up. Thanks.