Average Standard Deviation Move

[tjlocke99 18]

Knowing the 1 and 2 Standard Deviation moves of the underlying is quite useful. I know Chris often brings these up in his post, but I think he calculates them himself based on his own database.

Does anyone know sources of this data or if other metrics can be used like Average True Range (ATR)?

Also, does the standard deviation go off high/low or open/close?

Thanks!

Richard

[cwelsh 1,549]

I always calculate based off the close -- because that's the data I get for free.

Just for everyone's information, I have a VERY COMPLEX way of getting the data --

I pull it from Google finance for free. Excel has a nice function that will do that for you -- I just enter the ticker symbol into cell A1, and have it pull from web (See the data tab) the historical prices from the last year into spreadsheet number two. Spreadsheet one then pulls the closing price column back into spreadsheet one, and VIOLA all my calculations are displayed for me

[tjlocke99 18]

Hah! Chris I can't tell if your "VERY COMPLEX" is meant to be a joke? Sorry!

I think I am going to replicate this type of spreadsheet, BUT I think maybe Yahoo has high/low which I like better.

Are there any good tidbits like this in the Augen Excel book?

R

I always calculate based off the close -- because that's the data I get for free.

Just for everyone's information, I have a VERY COMPLEX way of getting the data --

I pull it from Google finance for free. Excel has a nice function that will do that for you -- I just enter the ticker symbol into cell A1, and have it pull from web (See the data tab) the historical prices from the last year into spreadsheet number two. Spreadsheet one then pulls the closing price column back into spreadsheet one, and VIOLA all my calculations are displayed for me

[Marco 223]

I use attached spreadsheet to import data from yahoo finance. It get the closing prices for any ticker for 3 years. I use it to compute historical/realised vols but I just stuck a standard deviation box on it (just set the number of days for which you want to compute it)

Can be easily changed to everyone's needs.

Disclaimer: I didn't spend much more than an hour or so on this and haven't used it much yet, so didn't check it for any errors yet....

also may not work (properly) with older versions of Excel.

ah - cant upload the file (Error You aren't permitted to upload this kind of file)

I'll try with dropbox file share - never used that before - let me know if that works

https://www.dropbox.com/sh/e37hbseycgbwxqf/biNvEe9lDq/historicalVol_last.xlsm

[cwelsh 1,549]

Thanks for doing that, very helpful

[tjlocke99 18]

I use attached spreadsheet to import data from yahoo finance. It get the closing prices for any ticker for 3 years. I use it to compute historical/realised vols but I just stuck a standard deviation box on it (just set the number of days for which you want to compute it)

Can be easily changed to everyone's needs.

Disclaimer: I didn't spend much more than an hour or so on this and haven't used it much yet, so didn't check it for any errors yet....

also may not work (properly) with older versions of Excel.

ah - cant upload the file (Error You aren't permitted to upload this kind of file)

I'll try with dropbox file share - never used that before - let me know if that works

https://www.dropbox....alVol_last.xlsm

Thank you Marco. This is pretty impressive. I didn't double check your std deviation calc yet, but I will let you know when I do!

[Marco 223]

Thank you Marco. This is pretty impressive. I didn't double check your std deviation calc yet, but I will let you know when I do!

actually there was an error in the sheet. It was using the first 60 (or which ever number you chose) data sets on the sheet to calculate STDEV but they are the first 60 days from like 3 years ago rather then the most recent 60 days - which is what you want really. I changed the file so if you use the link gain you get the new one or just change the formula in E3 to

=STDEV(OFFSET(B767,0,0,-D3,1))

cheers,

m.

[tjlocke99 18]

Thank you very much Marco!

[tjlocke99 18]

actually there was an error in the sheet. It was using the first 60 (or which ever number you chose) data sets on the sheet to calculate STDEV but they are the first 60 days from like 3 years ago rather then the most recent 60 days - which is what you want really. I changed the file so if you use the link gain you get the new one or just change the formula in E3 to

=STDEV(OFFSET(B767,0,0,-D3,1))

cheers,

m.

Hi Marco. When I download and attempt to open the workbook I get an error. Could you try and post it again? Maybe as a new file with a different name?

Thank you!

[tjlocke99 18]

Hi Marco. When I download and attempt to open the workbook I get an error. Could you try and post it again? Maybe as a new file with a different name?

Thank you!

Also Marco this maybe a result of the file being corrupted, but the Px Change column has a formula using the natural log function (LN). Is that correct? If Px is price change than should it be simple subtraction of the current close from the previous close?

Next the Volatility 30d, etc. fields all calculate the STDDEV and then multiple that by the square root of the calendar year days field? If that is correct then could you explain why this is done?

Thank you!

[Marco 223]

Also Marco this maybe a result of the file being corrupted, but the Px Change column has a formula using the natural log function (LN). Is that correct? If Px is price change than should it be simple subtraction of the current close from the previous close?

Next the Volatility 30d, etc. fields all calculate the STDDEV and then multiple that by the square root of the calendar year days field? If that is correct then could you explain why this is done?

Thank you!

mhm link works for me, but I uploaded a new file and here's a new link

https://www.dropbox.com/sh/e37hbseycgbwxqf/ltdlBfnLZ6/historicalVol_new.xlsm

maybe its a bit off an odd way to express it but LN(price2/price1) gives you the same result as (price2/price1)-1 basically the % change on the day.

STDEV*sq.root (time) is the formula for (in this case annualised) volatility. Its commonly done with 252 trading days for the year. As sq.root of 252 is ~ 16 you get to a rough approximation of multiplying the daily move by 16 to get to the annualised volatility. So a stock that moves 1% a day has aprox. a yearly realised vol of 16% (2% ~ 32% etc.) Or the other way around: for a Stock with a 32% implied Vol the market expects in AVERAGE ~2% move a day. Over earnings that can mean a low daily move combined with a big 1 day move on earnings day.

here's a bit more about the concept of Volatility and a few more formulas if you are interested.

http://en.wikipedia.org/wiki/Volatility_%28finance%29

[tjlocke99 18]

Marco,

This new link works thank you!

I am primarily using this data and statistics to decide how to price Iron Condors (long ICs - since that term seems to have different meanings depending on the book I read, I mean where I short an OTM and long a further OTM on both the put and call side).

I want to enter the IC around the time the next month monthly options come out and hold for maybe 30 days. I am actually thinking that simply the historic price range over similar periods of time may be a better way to price the IC.

Does anyone in this group have any thoughts on this? I don't like the technical analysis where people look for trend lines or similar technical analysis. Its a bit too much like reading tea leaves to me.

Does anyone know if Open Interest data can be downloaded anywhere on a specific option? I think CBOE has OI on the entire market.

Thanks!

Richard

[Marco 223]

http://www.ivolatility.com/custom/pbc/

this also gives you STDEV for a future date and you can put in a number of historical and implied vols to calculate a statistical probability of certain levels being reached. I would look at it just as that though - a statistical probability based on the input you make. I don't think you'll find many 'free lunches' (options that are priced below their probability of ending in the money - if they are then you are using a different input that the rest of the market is and you need to be confident that you know better than everyone else ) If you trade IC look at them as what they are - you are selling insurance to someone else and if things go wrong they will come to collect their money. Doesn't mean you cant make money with them but you are getting paid to assume a risk.

m.

[tjlocke99 18]

Great points Marco. Thank you.

[cwelsh 1,549]

Ahhh but Marco I do know things that the market doesn't -- solid material facts actually. I'm working on a piece I hope to eventually get published about the inehrent inaccuracies in option pricing.

At its heart, the option pricing models are based upon a few flawed assumptions. Do I have the perfect solution, absolutely not. However, do I have tweaks to the models that have given me better than average returns -- absolutely.

For instance, both the Black Scholes and Binomial Tree model assumes the geometric Brownian motion theory of stock price behavior is accurate. This is complete crap. Any historical analysis over almost any time frame will prove so. It also assumes a risk-neutral valuation -- which is also crap. I can cite 20 studies that say people are more apt to skew on the negative side than on the positive side (e.g. pay more for to protect against a downturn than they will take to protect against an upwared movement).

Both models also create, what I consider, a fundamental error in calculating implied volatility and historic volatility -- namely they don't attempt to normalize for events, such as earnings. This is the entire reason earnings volatility trades can make money. When you are calcuating historic volatility -- the days before and after earnings are weighted equally as any two days in trading. That is stupid, yet for some reason everyone seems to subscribe to it.

So what do I do? I recalculate, weigthing those periods differntly. Do I have a perfect process, of course not -- but I firmly believe its better than the market as a whole.

That said, as Marco noted, we are still pricing risk. I can be 100% correct that an option is overpriced or underpriced for any current level of risk. But so what? Let's say I think a call option is worth $4.00 and the market says $2.00 -- great investment -- unless the market still moves way against me. In that case I lost money, just less than the market as a whole (hope that statement makes sense).

That's why position sizing is so important. If I am right, more than the market makers, I can make money over the long term -- but if I price an event with 10:1 odds that the market is pricing at 5:1 -- I still lose, on average, 10% of the time. I just need to be sure to capute that other 90%.

[lee 3]

Chris, this sounds just like the kind of stuff that made John Bender so successful when he was trading/managing a hedge fund:

http://uamconsult.co..._dogovorov.html

I was so intrigued with his work that I even tried to track him down in Costa Rica a few years ago. He was a reclusive guy then - and I couldn't get to him, and unfortunately he was murdered in 2010.

I'd very much love to read your writings on this subject. - Lee

[tjlocke99 18]

Ahhh but Marco I do know things that the market doesn't -- solid material facts actually. I'm working on a piece I hope to eventually get published about the inehrent inaccuracies in option pricing.

. . .

That's why position sizing is so important. If I am right, more than the market makers, I can make money over the long term -- but if I price an event with 10:1 odds that the market is pricing at 5:1 -- I still lose, on average, 10% of the time. I just need to be sure to capute that other 90%.

Chris,

What books or web resources would you recommend for trading ICs (not RICs)? If we don't want to get as complex as you are to start then what do you recommend?

I have looked at ICs on stocks with high prices, and you when you look for monthly options about 30-40 days out, the strikes that are 2 standard deviations away are almost worthless and with the bid/ask you lose alot of money as soon as you enter the trade!

Thanks!

Richard

[Marco 223]

I have looked at ICs on stocks with high prices, and you when you look for monthly options about 30-40 days out, the strikes that are 2 standard deviations away are almost worthless and with the bid/ask you lose alot of money as soon as you enter the trade!

Thanks!

Richard

that shouldn't be too surprising - 2 STDEV away by definition means that the statistical probability is ~95.5% that the options will be worthless at expiry - so you wouldn't pay a lot for that.

You have to pick closer strikes and manage your risk (just as Kim describes it in his IC post)

[cwelsh 1,549]

You would be suprised how often a 2SD trade ends up making money and makes sense to enter. That said, as I alluded to earlier, I don't use a "pure" two SD trade -- I seperate out non-earnings and earnings periods and then normalize the results.

For instance, if I was calcultaing the 7 day SD move in AAPL over the last thirty days for a non-earnings period trade, I would find out the 7 day movement each day, and then take out the moves immediately after earnings (typically 2-3 days worth, sometimes more, have to look at the data fairly closely and trade volume). This has the obvious effect of, almost always, lowering the SD. That said, it gives you a MUCH more accurate picutre of what should be expected to happen in non-earnigns periods.

To me, to do anything else simply doesn't make sense. When using volatility to calcuate something, why include a KNOWN increased volatility when that event won't happen to your trade?

So using a "pure" SD calculation, I might end up with a 2SD of around 80, but when extracting out the earnigs data, it drops to 60. So, I'll use 60, and if I can get my 10% return, it's a possible trade.

I then do that over a universe of about 100 stocks. Sometimes I end up with 25 possible trades (rarely) and sometimes none (also rarely). Most often I end up with 10-12 canidates.

[Marco 223]

You would be suprised how often a 2SD trade ends up making money and makes sense to enter. That said, as I alluded to earlier, I don't use a "pure" two SD trade -- I seperate out non-earnings and earnings periods and then normalize the results.

For instance, if I was calcultaing the 7 day SD move in AAPL over the last thirty days for a non-earnings period trade, I would find out the 7 day movement each day, and then take out the moves immediately after earnings (typically 2-3 days worth, sometimes more, have to look at the data fairly closely and trade volume). This has the obvious effect of, almost always, lowering the SD. That said, it gives you a MUCH more accurate picutre of what should be expected to happen in non-earnigns periods.

To me, to do anything else simply doesn't make sense. When using volatility to calcuate something, why include a KNOWN increased volatility when that event won't happen to your trade?

So using a "pure" SD calculation, I might end up with a 2SD of around 80, but when extracting out the earnigs data, it drops to 60. So, I'll use 60, and if I can get my 10% return, it's a possible trade.

I then do that over a universe of about 100 stocks. Sometimes I end up with 25 possible trades (rarely) and sometimes none (also rarely). Most often I end up with 10-12 candidates.

agreed, I'd rather buy the 2SDTEV move than selling it. It hardly ever happens but if it does you usually can pay for MANY of these trades that lost money (buzz words 'fat tails', 'black swan event' etc)

and yes comparing realised vol or STDEV of a period with earnings to a period without earnings (or any other big event) doesn't make much sense.

If that's your edge over the market though that you describe earlier - I don't think its there. Nobody prices future IV comparing it to realised Vol with an earnings (or other) event. That's the reason IV collapses after earnings - people price that event out of the option.

Is Black scholes flawed? Yes, everyone who does use Black Scholes (or a derivative of it) does or at least should know its limitations.

Is the market 100% efficient? No

But do MM's consistently buy/sell options at the wrong price e.g. giving you an edge that you just need to harvest? I doubt it - MM's, banks and other users of options spend a lot of money and employ Quants that would look down on rocket scientists to improve models. I've seen quite a few in my time but while they describe the reality better than Black scholes I never felt they'd give me an edge over the market in trading volatility. I also think the misprice is more in the wings or how the market prices skew and not so much in the atm option. As long as all your trades still depend that your view on the market being right, that edge - if it exists - doesn't help you THAT much either. I doubt that the misprice is as much as you describe above (the market sells you a 4$ option for 2$) I'd be shocked if its as much as 5% so 0.20$ on a 4$ option and you easily make or lose that 5% by just entering or exiting that trade a little bit earlier or later.

I still need to read the story that lee posted. I only got to the bit where it says the guy made $7mio out of $80k. If you are on the brink of that ignore my scepticism and take $50k seed money from me please!

(and I'd still like to read what your findings are when you publish them)

Marco.

[Kim 7,943]

Guys, thank you very much for an amazing discussion. It reaffirms again the tremendous value of our community for those wishing to learn.

I found out over time that entering 6-7 weeks before expiration and using deltas of the 15-25 range for the short strikes has worked the best for me. If you go with smaller deltas, you simply don't get enough premium to be able to close before expiration, which for me is a major feature. I prefer not hold beyond the last 2 weeks due to the large negative gamma.

[cwelsh 1,549]

$7m from $50k? No, not happened (I wouldn't be here daily if it had ). And the $2 -- $4 is just an example, I've never seen one thta big.

But what I do see, somewhat regularly, is I price a 2SD $5 spread at $.55 when the market might have it at $.48 -- this would move the trade from something I wouldn't consider into something I would.

As to whether I'm long or short a position -- I do both. I regularly sell the out 2SD spread and let it expire worthless. The goal obviously being, on a 10% return trade, to be right on right about eight five percent of them (I average about a 50% loss on the trades that go against me). Anything above that point is gravy. Also, since I typically actually have profits in the 11-12% range on the trades, that helps as well. (As does position sizing, trade management, and different types of trades).

Think about it, theortically, I "should" be correct 95.45% of the time -- but I make money just by being right 85% of the time. A volatility edge model (historic v. implied) doesn't have to be perfect -- it just has to be better than average with good position size management.

[tjlocke99 18]

Obviously Kim, Marco, and Lee - you are way more advanced then I am. That is why I am trading a very, very small % of my portfolio. Simulating trading has not worked for me, because at least with my broker, with simulated trading they will only execute the order at the bid for a short and the ask for a long. In many of these trades that completely distorts the P&L.

All this being said, I thought, that the general statement that I have read is that the market tends to overprice risk and thus options are generally more expensive to purchase then they "should" be. The same as insurance, its likely something you need, but the people selling you it are making money off you.

Chris - the analysis you describe is pretty wild, but I not clear how from a software perspective you do it all. Is it really all done in Excel? Where are you getting 1 day option bar?

Marco - your point on a 2d move is well taken. How do you decide what ICs to enter?

Also what terminology are we using? Are we saying longing an IC getting a debit and shorting an IC is getting a credit (shorting also what Kim has been calling RIC)?

Thanks!

R

[lee 3]

Chris, you describe your process for culling trading candidates (stock). This is obviously more complex than simply focusing on an index option (e.g. Kim's Russell IC trade). Are you trading both individual stocks and indices, or only focusing on stocks, and if so is this because your analysis shows more "mis-pricing" opportunities on individual stocks? - Lee

[cwelsh 1,549]

I trade quite a few things, for instance right now I have open positions on:

Three different calendars (one expiring in June and two in July)

A short IC that expires Friday (PCLN);

A long IC (on the VXX)

and 3 IV plays (straddles)

and I closed on Wed my June RUT short IC (for a nice return I might add).

I do NOT classify a IV RIC in the same category as my VXX type ICs.

As for what I trade:

- On IV trades I trade stocks around earnings -- any stock with options with sufficient volume and reasonable spreads is fair game

-- On short ICs I trade both indexes, high valued stocks (I have a universe of about 100-150 stocks there)

-- I SOMETIMES trade hold through earnings stocks on RICs or long ICs, but that's rare, but again includes all stocks with sufficient volume and reasonable spreads

-- I trade long ICs on indexes and my universe of stocks

-- On calendars I trade anything. I use the screenign process I described to cover all indexes, stocks, and anything else that moves with WEEKLY options. In fact, I regularly pull the CBOE weekly list for the starting point of that analysis.

Hope that helps.

[lee 3]

Chris, I so appreciate your willingness to share and for responding so quickly! Bear with me for one more multi-part question.

I'm wondering why so many strategies? It sort of makes my head spin to think about such a diverse trading plan as yours - though I'm sure after years of doing it it probably feels like second nature to you. But why so many strategies going on simultaneously (in other words, how and why did you "build" your trading plan/style to the way it exists now), and why not just focus on a few specific trades/set-ups with a bit larger size so there's less to manage and research?

I know you get what I mean, but to beat the question like a dead horse, another example is that one of my recent yoga clients, who was a hedge fund founder and CEO who sold to Blackrock a few years back for a few billion dollars, shared his trading advice. He suggested getting to know one specific niche (the example he gave was "banks" in the Pacific Northwest), and know it better than nearly everyone else in the business. It's overwhelmingly clear to me that you know your stuff; I'm just wondering how and why you wove your net so wide? Best, Lee

[cwelsh 1,549]

Well you're right in one thing -- I don't think my net is that wide. I used to try to also trade using fundamental analysis (ala Buffett/Graham), technical analsyis, trend following (e.g. turtle trading), forex market timing --- read and modeled it all, and found this is what I seemed "best" at. I stick to primarily shorter term option trades.

As to how I evolved to where I was now, at one point I traded almost exclusively far OTM ICs. (short and long, but primarily short), and was quite good at it. However, about once every six to nine months, I would have several trades go against me. I can handle 5-10% trades going wrong, but if 30% do, well it wipes out a year of profits. The thoght process was that is stupid, how do I deal with that.

So I evolved into trading ATM IC's AND OTM ICs in a given month. That way if a market event happened, that screwed me, as opposed to just underlying stock movements, I could still at least break even. Except it still wasn't perfect because ATM IC's are limited profit, so if a move was big enough, and sustained across a market, I could still theortically lose -- not as bad as before, but could happen.

So, I moved into also IV earnings plays. These trades worked great, and if something unexpected happens -- well they just profit more. So, for about a 9 month to year period, I traded three instruments.

But then volatility collapsed for a sustained period. That made getting premium on far OTM ICs more difficult, ATM ICs more risky, and IV earnings trades harded to profit on --- but made calendars home runs.

So I modeled how do these all interelate in different markets. (High volatility, low volality, earnings, non-earnings, high volume, etc) and have learned that if you include ALL of them, you are greatly hedging your losses. It is VERY unlikely to get burned on all in anyone period.

So each week I make a weigting model. Sketch out your proposed trades, pull out an option calculator and figure out what will happen if:

1. market IV drops a ton

2. market IV increases a lot

3. market IV does nothing

By running through a combination of calendars, ICs, and IV plays, I can normally develop a scenario where I, theortically, should never have a losing month, regardless of market condtions. That's the ideal situation.

Of course, you can have the scenario of where you're covered for the market moves, but you get hosed on all individual trades. For instance, let's say AAPL drops 6% in a week on a calendar play, I don't get an IV move on SINA that historically happened 7 out of the last 8 earnings cycles, I don't get an IV move on stock XYZ that was expected, and my RUT condor, due to increasing market volatility, did not make money.

That is very unusual -- typically if you're seeing AAPL drop and market volatility rise, your IV trades would have been profitable. And in fact, 95% of IV trades for the month probably were -- I just chose the wrong ones. That type of thing can still happen, and until I get a $1b portfolio that I can trade all things, can't deal for that situation. (Believe it or not, I actually think, with enough money, you can develop an option portfolio that makes money EVERY month -- for instance if you lose on EVERY IV trade, well that means market IV dropped and my ICs and calendars will be profitable and vice versa).

Anyways, hope that provides some insight.

[fieldydwb 31]

Marco,

Any chance you can upload that file again?

Would much appreciate it.

Thanks in advance

[Marco 223]

Marco,

Any chance you can upload that file again?

Would much appreciate it.

Thanks in advance

I put the file back again so above link should work again.

https://www.dropbox.com/sh/e37hbseycgbwxqf/ltdlBfnLZ6/historicalVol_new.xlsm

just to be clear this calculates the STDEV of daily returns over the no of days chosen. If you are looking to create some sort of confidence interval for a weekly or monthly RIC it would make more sense to look at STDEV of weekly or monthly returns respectively. You would have to modify the sheet for that.

m.