Some maths required...
- Thread starter earth_imperator
- Start date
I looked for a degree in a few Pez dispensers but I quickly gave up on that approach!!
lolIt's true. I do some statistical consulting and it's fascinating how many different areas can benefit from properly conducted statistical analysis and/or modelling. It's also scary how often statistics seems to be abused out there "in the wild."
...and how much money that can cost a company who relies on those jackleg numbers. And why they pay well for a good consultant it would seem.
Sometime if it suits, love to hear a story or two of what you've seen "serious businessmen" do and how you break it to them gently, lol. "Gentlemen, what do you say we take a ten minute break and stretch em out. Mr Jones, I have a question if you have a minute... muted tones would you know who could fill me in on where these numbers come from?"
Don't think he was trying to find implied volatility as he defined "IV" in his post as: IV (ie. a stddev). I think he was trying to find future volatility. Anyway he answered his own question. Another of his frivolous threads can be closed.
TD, I dig your avatar picture. I too am a box man.
Some remarks on the above used work-around approximate method for finding -1SD and +1SD :
Imagine you have usual options chain data consisting of the table data for Call and Put options including the underlying ticker and underlying spot (ie. just the "last traded" stock price), nothing more.
You get these options data from your data provider (requires a costly subscription) and each such data "page" counts as a "request", ie. your payment is defined by the number of requests you make. So, the more you download, the more you have to pay. Ie. I cannot afford to download also the data of the underlying stock itself (ie. historical data of it), b/c it would take much more time to download (currently it already takes > 1h), and generating the analysis then would take even more time due to the additional data to process. And also not to forget the additional cost it requires. And it's really unnecessary in my case.
You have thousands of such data records (ie. more than 4000 tickers with options, each having on average 4+ Expiration dates, makes 16,000 Call+Put chain data tables, or pages. Your list-generating program (or scanner program) goes thru all these 16,000 tables and creates a list according to set filters, and of course also from some new computations it makes. Among the new information that would be useful on the generated list is to know also the -1SD and +1SD from the current underlying spot.
Since we don't have any further data of the underlying itself (ie. historical data), then it makes sense to find a work-around approximate solution, as was shown.
And: the resulting lists are not intended for presentation or so, but just for the trader (human and/or program) to find good trades to make... So, no need to be 100% exact with the -1SD and +1SD.
That's the whole story & idea behind it.
For those who don't know such options data, here's such an options Call & Put table page as an example: https://finance.yahoo.com/quote/BBBY/options?p=BBBY&date=1663286400
During regular market hours these tables fill with additional data, especially the Bid, Ask, and Implied Volatility (IV) then get filled up. IV is calculated from Bid/Ask, but when the market is closed then all non GTC orders, ie. Day orders, get removed from the order book, so then the IV no longer is representative; meaning IV becomes real only during regular market hours when the order book fills up with all the Bids and Asks...
See also this posting https://www.elitetrader.com/et/threads/options-first-paper-trade.369091/
Imagine you have usual options chain data consisting of the table data for Call and Put options including the underlying ticker and underlying spot (ie. just the "last traded" stock price), nothing more.
You get these options data from your data provider (requires a costly subscription) and each such data "page" counts as a "request", ie. your payment is defined by the number of requests you make. So, the more you download, the more you have to pay. Ie. I cannot afford to download also the data of the underlying stock itself (ie. historical data of it), b/c it would take much more time to download (currently it already takes > 1h), and generating the analysis then would take even more time due to the additional data to process. And also not to forget the additional cost it requires. And it's really unnecessary in my case.
You have thousands of such data records (ie. more than 4000 tickers with options, each having on average 4+ Expiration dates, makes 16,000 Call+Put chain data tables, or pages. Your list-generating program (or scanner program) goes thru all these 16,000 tables and creates a list according to set filters, and of course also from some new computations it makes. Among the new information that would be useful on the generated list is to know also the -1SD and +1SD from the current underlying spot.
Since we don't have any further data of the underlying itself (ie. historical data), then it makes sense to find a work-around approximate solution, as was shown.
And: the resulting lists are not intended for presentation or so, but just for the trader (human and/or program) to find good trades to make... So, no need to be 100% exact with the -1SD and +1SD.
That's the whole story & idea behind it.
For those who don't know such options data, here's such an options Call & Put table page as an example: https://finance.yahoo.com/quote/BBBY/options?p=BBBY&date=1663286400
During regular market hours these tables fill with additional data, especially the Bid, Ask, and Implied Volatility (IV) then get filled up. IV is calculated from Bid/Ask, but when the market is closed then all non GTC orders, ie. Day orders, get removed from the order book, so then the IV no longer is representative; meaning IV becomes real only during regular market hours when the order book fills up with all the Bids and Asks...
See also this posting https://www.elitetrader.com/et/threads/options-first-paper-trade.369091/
Last edited:
Hey @TheDawn! You are in my Ignore List. You should not see my postings. So then why do you still post in my thread?...
You have, as usual, no clue of the topic, so your comment is as deplaced as wrong in all aspects, again, and as is usual with you. You should take some courses, man!
You have, as usual, no clue of the topic, so your comment is as deplaced as wrong in all aspects, again, and as is usual with you. You should take some courses, man!
Last edited:
I'm glad you feel you've solved your problem well enough for your purposes. You say "case closed" and so I'll respect that and try not to say too much more.
I would, if I may, just like to say that I am intrigued by your assertion that you only have one data point (or would like a solution that can be applied to a single data point) and I am curious about your use case. It is very often the case that we use as much data as we can reasonably collect/afford/measure/etc in order to estimate the parameters for a distribution that we want to use; in the case of the normal distribution, for example, we need to estimate two quantities using data in order to identify the normal distribution that best fits our data and is therefore most likely to be of use in making inferences or drawing conclusions. (There are infinitely many normal distributions, of course!)
If you must rely on a single data point then you are forced to "bake in" a lot of assumptions... but at the end of the day it's all about practical utility and for all I know you have some use case where you just need a quick and dirty, very rough estimate... or perhaps you have some theoretical knowledge or prior research that already validates some of your assumptions; without actually knowing the problem I can't possibly know.
Above I've posted the background story with additional info. Do you think this approximate method taken over from the ND method (as given in the initial posting) to LogND is acceptable or not?
What about the "pure" method used in the original ND case? Do you think it's a wrong method, so then everything else building on that is wrong too?
Just asking a professional statistician, as I assume you are according to what you write and your profile page indicates.
Thx for your opinion.
Or asked differently: how would you have solved this problem (getting a -1SD and +1SD for the underlying stock price) under the given constraint of having just the last stock price?
Last edited:
For the underlying problem of calculating the -1SD and +1SD, one only needs the mean (not the underlying data points that formed this mean), and I treat this single data point that I have as the last stock price as the needed mean
, and continue from that assumption...So, IMO in this case just only 1 data point ought to be enough, the rest lies in the magic of the Normal Distribution, ie. the said famous -34.1% and +34.1% around the mean for the -1SD and +1SD, respectively...
I just extended this idea, so that it can be used also in the LogNormal Distribution (LogND) case, by a rough simple approximation (cf. the avgIV in my above posting). For my use case this solution is sufficient enough.
Last edited: