Rappert's straddle forecasting charts

[RapperT 336]

 

Thanks for all the questions/feedback. It’s always helpful to see that. I thought it may help to break down how I approached the problem not because it is necessarily the best way but because it explains why I’m looking at it the way I am. Also, from a purely development standpoint I’m building all of the modules out in an object oriented way so we can quickly iterate and improve various components of the model over time to improve it.

 

Assumptions:

 

Implied Volatility is normally a good predictor of the future volatility of the underlying. If you take this idea to its logical conclusion it means that absent any fees etc options are usually priced correctly and the long term expectation value of buying or selling any option is zero.

 

Approaching earnings IV will sometimes rise due to speculation or additional hedging (or some other factor...not really important) but this rise in IV does not reflect any change in the expected volatility in the underlying instrument. This rise in IV means that the standard pricing model may underprice options.

 

This mispricing due to an expected rise in IV may lead to an opportunity to purchase options to either capture the increasing spread between near and far month IV or allow us to exploit the rise in IV directly through a long vol position.

 

Development Process:

 

Build each model component independently (OOP) to allow rapid iteration and improvement.

Build in model/forecasting metrics to capture model error and guide future improvements.

 

Components:

 

Current Market IV estimator

IV forecast.
Option pricing model (currently BS but that will change)
Data Model

 

I’m not going to explain all of these components in detail. It is important to know that each is “stubbed” so we can improve them over time. Suffice it to say while our components are currently rough, we should be able to improve them quickly and we’ve implemented some standard and necessary forecasting techniques already like normalization etc. Recent results show that our methodology has some merit (ORCL spike etc).

 

Source of Edge:

 

The avoided theta decay due to rising IV is in essence the edge in this trade if our assumptions hold. We've even seen some instances where theta is actually positive for a long straddle despite what the BS model would state.  While there are sources of model error, if we isolate times when the calculated edge is significantly greater than model error we should be able to trade it profitably. Also, doing this quantification upfront is necessary for us if we want to develop a empirically verified objective function that we can optimize.

 

One additional note, to do this really well we likely need to build out our own pricing model using Monte Carlo methods. I’m scoping this now as well as working on some shorter term improvements to the BS implementation we’re using now.

 

Regards,

 

Rich A and John A

 

this one looks good to me:

 

 

 

Explanation of Earnings Charts:

 

Definitions:

 

Normal IV: Implied vol prior to any influence from the upcoming earnings. Can be thought of as “fair” volatility.

 

Earnings IV: The implied volatility rise due to the upcoming earnings cycle.

 

Theoretical Delta Neutral Straddle: An artificial straddle that is perfectly delta neutral. This allows us to isolate price changes in the underlying options due solely to IV rise.

 

Normalized IV Curve: We look at historical implied volatilities and adjust them to take into account the current normal IV and then forecast an IV return curve over the next several days. This allows us to use historical IV from different periods without any bias due to higher or lower overall IV in the market. This curve shows how the IV changes over the course of the trade.

 

Normalized straddle prices: This is the blue line on the chart. It shows how the value of a theoretical delta neutral straddle will change over time based solely on the IV curve. The inputs to this chart are normal iv, earnings iv, normalized IV curve, the black scholes model and days to expiration etc.

 

Standard straddle prices: The orange curve on the chart. This shows how the same straddle would perform if there were no IV price spike.

 

Advantage: The difference in accumulated theta decay between the normalized straddle prices and standard straddle prices. This should be evident if you understand the relationship between theta and gamma. (Note: advantage can be used to calculate edge and fractional position sizing). Simple way to think about it...the difference in the two line shows you the amount of gamma you’re getting for free.

 

(Note: I get it is a little confusing to look at the return stream instead of actual option prices. We did this for a number of reasons related to future plans with a more automated trade system. If you want to convert it to an actual straddle price simply take the price of a delta neutral straddle on day t-25 and multiply it by the daily values of the blue line in the chart. This will give you an actual price forecast for that straddle. )

 

How to use the chart: 4 scenarios.

 

Scenario 1: No significant advantage. Downward sloping blue line.

 

 

 

 

 

This shows Costco. Notice the blue line not only slopes downward it never opens a large gap from the theoretical price. While there is a slight advantage the model does have some error in it. This is probably not a good trade.

 

Scenario 2: Advantage but downward sloping blue line.

 

 

Notice the time period from around Sep 3^rd to Sep 8^th. The blue line is sloping down showing that without any movement in the underlying the straddle will lose money. However, it still has a wide gap between the blue line and orange line. That means that if you got in around Sep 2^nd you probably got a lot of gamma for free. This trade (Sep 2 to Sep 8) should have a positive expectancy although a lower win rate than some other trades.

 

Scenario 3: Advantage and a flat blue line.

 

Look at Oracle between Sep 5^th and Sep 7^th. The line for those few days is flat. This implies that if you opened a straddle then you would pay zero theta for 2 to three days of gamma. This is a low risk trade with almost free gamma upside.

 

Scenario 4: Advantage and upward sloping blue line.

 

This is the best scenario. Look at Oracle between Sep 8^th and Sep 11^th. The blue line is upward sloping. This means that you could buy a straddle and actually get paid theta to gain on gamma. This is a great spot to be in.

 

Edited October 8, 2017 by RapperT
edited to add photos that will show in all browser

[RapperT 336]

2 minutes ago, jbthree said:

@RapperT Just so I understand, what about this chart looks good to you, that you don't see in COST?

blue line is forecast return into earnings using historical vol.  Orange line is return under normal conditions.  You can see that starting around 10/5, the delta between the lines grow significantly.  This would indicate a very low risk straddle (ie holds value well against theta even without any stock movement).  Cost is the opposite.  Theta destroys it.

[RapperT 336]

the other piece is that, by utilizing yowster's combo trade, we can enter earlier and increase probability of gamma gains.  Im in this one already.

[jbthree 6]

1 minute ago, RapperT said:

blue line is forecast return into earnings using historical vol.  Orange line is return under normal conditions.  You can see that starting around 10/5, the delta between the lines grow significantly.  This would indicate a very low risk straddle (ie holds value well against theta even without any stock movement).  Cost is the opposite.  Theta destroys it.

then why wouldn't you wait until closer to 10/5?

[RapperT 336]

Just now, jbthree said:

then why wouldn't you wait until closer to 10/5?

i would if I werent using Yowster's combo trade.  See above.

[Kim 7,943]

9 minutes ago, RapperT said:

blue line is forecast return into earnings using historical vol.  Orange line is return under normal conditions.  You can see that starting around 10/5, the delta between the lines grow significantly.  This would indicate a very low risk straddle (ie holds value well against theta even without any stock movement).  Cost is the opposite.  Theta destroys it.

Sorry, could you please explain more? What is the difference between the 2 lines? What "under normal conditions" means?

[RapperT 336]

@Kim The orange line shows the price of an artificially delta neutral straddle using IV from the date of the start of the chart (i.e "normal conditions")

The blue line forecasts return of an artificially delta neutral straddle using historical IV from previous earnings cycles.  We use anywhere from 4-10 cycles depending on the stock (still working on automating everything).

When the delta or difference between the two is significant , in most cases it will signal an edge in trading that particular straddle.

 

Does that make sense?

Edited September 18, 2017 by RapperT

[RapperT 336]

"delta" in the mathematical sense, not option greek.  I will stop using that language

[Kim 7,943]

7 minutes ago, RapperT said:

@Kim The orange line shows the return of an artificially delta neutral straddle using IV from the date of the start of the chart (i.e "normal conditions")

The blue line forecasts return of an artificially delta neutral straddle using historical IV from previous earnings cycles.  We use anywhere from 4-10 cycles depending on the stock (still working on automating everything).

When the delta or difference between the two is significant , in most cases it will signal an edge in trading that particular straddle.

 

Does that make sense?

So in case of IBM, yellow line forecasting 60% loss (from 1.0 to 0.40)?

[Stanislav 302]

8 minutes ago, RapperT said:

@Kim The orange line shows the return of an artificially delta neutral straddle using IV from the date of the start of the chart (i.e "normal conditions")

How do you calculate that? Take IV from the date of the start of the chart, fix it and all other parameters in Black and Scholes formula except time to expiration and strike price. And then change time to expiration and strice == underlying price, compute cost of the straddle using B/S?

[SBatch 3,123]

3 minutes ago, Kim said:

So in case of IBM, yellow line forecasting 60% loss (from 1.0 to 0.40)?

Only down to about .8 if closed before the release.

[jbthree 6]

1 minute ago, SBatch said:

Only down to about .8 if closed before the release.

would that mean that, if the stock does not move, one would need to make up 20% in hedged strangles?

[SBatch 3,123]

Just now, jbthree said:

would that mean that, if the stock does not move, one would need to make up 20% in hedged strangles?

Historically, yes.

[jbthree 6]

13 minutes ago, RapperT said:

"delta" in the mathematical sense, not option greek.  I will stop using that language

So, in the mathematical sense, delta neutral means the stock price stays the same?

[Javier 99]

Just now, jbthree said:

So, in the mathematical sense, delta neutral means the stock price stays the same?

No in this case delta is the greek

[Djtux 548]

3 minutes ago, Stanislav said:

How do you calculate that? Take IV from the date of the start of the chart, fix it and all other parameters in Black and Scholes formula except time to expiration and strike price. And then change time to expiration and strice == underlying price, compute cost of the straddle using B/S?

That's for the orange line. For the blue line, you also need to change the implied vol as well for each day. You need some sort of IV forecast model, could be as simple as taking the average IV at T-X for a certain number of cycles.

That's how i would do it.

 

[RapperT 336]

11 minutes ago, Kim said:

So in case of IBM, yellow line forecasting 60% loss (from 1.0 to 0.40)?

depending on when one closed, if there were no spike in IV, yes it could be that bad. 

 

But this is why we trade long straddles when there is an IV spike that gives us an edge

[Kim 7,943]

2 minutes ago, RapperT said:

depending on when one closed, if there were no spike in IV, yes it could be that bad. 

 

But this is why we trade long straddles when there is an IV spike that gives us an edge

Ok, so by "under normal conditions" you mean if there are no earnings? But why is it even relevant? The whole point is that there ARE earnings, and we know that IV WILL go up - we just not sure if it will go up enough to compensate for theta. 

We know that for IBM for example, straddle might lose up to 20% if entered now without short strangles. This is what matters, not how much it would lose if there are no earnings.

[RapperT 336]

16 minutes ago, Kim said:

So in case of IBM, yellow line forecasting 60% loss (from 1.0 to 0.40)?

really we are simply trying to quantify what we /you/augen already know to be true for many of the straddle candidates we trade into earnings.

[RapperT 336]

3 minutes ago, Kim said:

Ok, so by "under normal conditions" you mean if there are no earnings? But why is it even relevant? The whole point is that there ARE earnings, and we know that IV WILL go up - we just not sure if it will go up enough to compensate for theta. 

We know that for IBM for example, straddle might lose up to 20% if entered now without short strangles. This is what matters, not how much it would lose if there are no earnings.

you see the blue line right?  That is a price forecast using historical IV run up.  Including the orange line shows the effect vol has on the price of the straddle.  This is exactly what you are talking about.

 

Without a baseline you are simply guessing on any edge.  We are attempting to quantify it

Edited September 18, 2017 by RapperT

[Kim 7,943]

2 minutes ago, RapperT said:

you see the blue line right?  That is a price forecast using historical IV run up.  Including the orange line shows the effect vol has on the price of the straddle.  This is exactly what you are talking about.

 

Without a baseline you are simply guessing on any edge.  We are attempting to quantify it

So when the blue line goes below the yellow line (like around Sep.26-29), that means that straddle not only losing value, but it is losing value even more than "normal" straddle without earnings?? That doesn't make sense. We know that there are periods when straddle would lose value, so IV increase does not compensate for theta losses, but it is very uncommon for s straddle to lose value even more than straddle without upcoming earnings would. 

[Stanislav 302]

12 minutes ago, RapperT said:

you see the blue line right?  That is a price forecast using historical IV run up.  Including the orange line shows the effect vol has on the price of the straddle.  This is exactly what you are talking about

Well, if you calculate the orange line as I described above, the calculation is flawed. Since we have an earnings event that occurs before the expiration, to proper model the option price we need to take into account that IV consist of two components: normal option volatility and earnings option volatility. If we just take IV at some point and fix it, and put this into the B/S formula, we are going to receive completely incorrect theta decay, i.e. B/S will tell us that option price will decay as quickly as if we had no earnings event. But we do have a binary event before the earnings.

 

More details on Earnings Volatility calculation can be found in  Brian Johnson's book "Exploiting Earnings Volatility: An Innovative New Approach to Evaluating, Optimizing, and Trading Option Strategies to Profit from Earnings Announcements".

 

I understand what you are trying to achieve - assess a theta decay in option price before the earnings, but as well as @Kim I do not see how you calculation is relevant here. Proper earnings volatility modelling is quite complicated thing - one needs to analyze whole option chain and calibrate a model to be able to split IV to Normal IV and Earnings IV. And so on (some time ago I put a lot of efforts into studying Brian Johnson's work, and can tell it is not as simple as take B/S formula and put some values into it).

Edited September 18, 2017 by Stanislav

[RapperT 336]

@Kim

BS assumes static IV, real life IV isnt static. Our blue line is based on historical IV.  There is a lot of noise on this particular chart due to the look back we used. but regardless, there will always be variation.This is why the lines can cross when we go further out.

Any forecast will have some level of error.  This is simply how it goes.  This is why we open the trade when the signal is big enough to override forecasting errors.

If you prefer I not share this resource on your board, Im happy to refrain from doing so.

Edited September 18, 2017 by RapperT

[RapperT 336]

7 minutes ago, Stanislav said:

Well, if you calculate the orange line as I described above, the calculation is flawed. Since we have an earnings event that occurs before the expiration, to proper model the option price we need to take into account that IV consist of two components: normal option volatility and earnings option volatility. If we just take IV at some point and fix it, and put this into the B/S formula, we are going to receive completely incorrect theta decay, i.e. B/S will tell us that option price will decay as quickly as if we had no earnings event. But we do have a binary event before the earnings.

 

More details on Earnings Volatility calculation can be found in  Brian Johnson's book "Exploiting Earnings Volatility: An Innovative New Approach to Evaluating, Optimizing, and Trading Option Strategies to Profit from Earnings Announcements".

 

I understand what you are trying to achieve - assess a theta decay in option price before the earnings, but as well as @Kim I do not see how you calculation is relevant here. Proper earnings volatility modelling is quite complicated thing - one needs to analyze whole option chain and calibrate a model to be able to split IV to Normal IV and Earnings IV. And so on (some time ago I put a lot of efforts into studying Brian Johnson's work, and can tell it is not as simple as take B/S formula and put some values into it).

You misunderstood me. Im not going to argue about it though. 

Edited September 18, 2017 by RapperT

[Kim 7,943]

1 minute ago, RapperT said:

@Kim

BS assumes static IV, real life IV isnt static. Our blue line is based on historical IV.  There is a lot of noise on this particular chart due to the look back we used. but regardless, there will always be variation.This is why the lines can cross when we go further out.

Any forecast will have some level of error.  This is simply how it goes.  This is why we open the trade when the signal is big enough to override forecasting errors.

If you prefer I not share this recourse on your board, Im happy to refrain from doing so.

No, please do. Personally I'm not very clear how to use it and how beneficial it is (for the reasons I explained), but additional data points never hurt. 

[jbthree 6]

I agree. I love the discussion.

[Stanislav 302]

5 minutes ago, RapperT said:

@Kim

BS assumes static IV, real life IV isnt static. Our blue line is based on historical IV. [...]

If you prefer I not share this recourse on your board, Im happy to refrain from doing so.

The problem is not that BS assumes that IV is static, but rather that BS formula does not know that a binary event (which is a volatility generator) is going to take a place before the expiration. To proper model this we need a special volatility model or some other tricks.

 

Please, keep sharing your charts and research. We a happy to have a good topic to discuss, do not perceive this as a criticism. We are just trying to figure out what these charts do really show and how to properly use them and possibly how to improve them.

 

Edited September 18, 2017 by Stanislav

[RapperT 336]

3 minutes ago, Kim said:

No, please do. Personally I'm not very clear how to use it and how beneficial it is (for the reasons I explained), but additional data points never hurt. 

in my humble opinion,these charts show how a stock holds up to theta unto earnings with less noise than something like an RV chart.  I see the benefit of both.

[RapperT 336]

also, we're using ORATs to handle IV skew.

 

I have to jump on a call but will respond in more detail later on.  Thanks all

Edited September 18, 2017 by RapperT

[Stanislav 302]

6 minutes ago, RapperT said:

You misunderstood me. Im not going to argue about it though. 

Can you, please, at least, explain how do you calculate the orange line? The blue line seems to be an average volatility from the past cycles. But there were no clear description for the orange line calculation. This does not make it easy to properly interpret the charts.

[Javier 99]

2 minutes ago, RapperT said:

in my humble opinion,these charts show how a stock holds up to theta unto earnings with less noise than something like an RV chart.  I see the benefit of both.

I am with You, I see the value of your chart and appreciate, but I miss the Multi-cycle RV charts too.

[Javier 99]

2 minutes ago, Stanislav said:

Can you, please, at least, explain how do you calculate the orange line? The blue line seems to be an average volatility from the past cycles. But there were no clear description for the orange line calculation. This does not make it easy to properly interpret the charts.

As You said, take initial IV into BS and Sustract one to the DTE day after day. All rest the same.

[SBatch 3,123]

6 minutes ago, Stanislav said:

Can you, please, at least, explain how do you calculate the orange line? The blue line seems to be an average volatility from the past cycles. But there were no clear description for the orange line calculation. This does not make it easy to properly interpret the charts.

I believe he had mentioned previously that he uses an IV number that is approximately 25 days before earnings and holds it static.

[RapperT 336]

13 minutes ago, SBatch said:

I believe he had mentioned previously that he uses an IV number that is approximately 25 days before earnings and holds it static.

or whenever the chart starts...typically around 25ish days

[Stanislav 302]

39 minutes ago, Javier said:

As You said, take initial IV into BS and Sustract one to the DTE day after day. All rest the same.

 

22 minutes ago, RapperT said:

or whenever the chart starts...typically around 25ish days

@RapperT, Well, then I am not sure in what part I misunderstood you as you indicated above. My description of how the orange line is calculated and why this calculation is not correct in a presence of a binary event seems to be accurate enough.

 

That said, I found ORATS service does have some proprietary method to calculate what they call Fair Volatility (levels to which implied volatilities are projected to fall after earnings, plus an adjustment for the forecast earnings move - https://orats.com/data/earnings). If you are using this service as a data provider, it seems this data can be used to properly model earnings volatility without creating sophisticated volatility models manually. I just do not see any indication you are using some kind of differentiation between earnings volatility and normal volatility in your calculations. You mentioned ORATS IV skew - I am not sure what exactly this references to (IV30, horizontal skew?).

Edited September 18, 2017 by Stanislav

[RapperT 336]

@Stanislav maybe it's me who is misunderstanding you, not sure.

 

we are intentionally not including the earnings event in the calculation of straddle price on the orange line.  The point is to use that as a baseline to compare to our forecast which does include the earnings event (the blue line).  Orange line shows what would happen to price if we didnt have benefit of the binary event (earnings) you identify. Without the orange line there is no way to quantify your edge.  We prefer not to guess but obviously some are ok with that.

 

[Stanislav 302]

54 minutes ago, RapperT said:

@Stanislav maybe it's me who is misunderstanding you, not sure.

 

we are intentionally not including the earnings event in the calculation of straddle price on the orange line.  The point is to use that as a baseline to compare to our forecast which does include the earnings event (the blue line).  Orange line shows what would happen to price if we didnt have benefit of the binary event (earnings) you identify. Without the orange line there is no way to quantify your edge.  We prefer not to guess but obviously some are ok with that.

 

Ok, thanks! Now I believe I understand the idea behind, and the implementation.

 

One more question. When you calculate OrangeLine and take IV of the straddle from T-25, what are you using: is this an IV of the current cycle, or an average IV from the past cycles?

 

I can see some value in taking IV of the straddle from T-25 from an average IV of past cycles (this will be close approximation to a Normal IV of the past cycles on average, since it is around 25 days to earnings and IV does not start to rise much). But if OrangeLine is derived from current cycle prices, it becomes really difficult to tell what the difference from BlueLine and OrangeLine really means in a mathematical sense.

Edited September 18, 2017 by Stanislav

[RapperT 336]

We've looked at using some sort of average but I'm not convinced that really helps that much.  I'm open to hearing why you believe it is optimal.

 

 

[Kim 7,943]

Yes, I think I understand what you are trying to achieve.

So the blue line tells you how much on average earnings straddle would lose in xxx days. This is basically similar information that we have in RV charts. Lets say RV shows average decline from 5% to 4%, your chart will show decline from 1.00 to 0.80. Correct?

This is good - don't have to calculate it manually. But what does comparison with "normal" (non-earnings) line tell you? 

For example:
Stock xxx shows blue line declines from 1.0 to 0.8 and yellow line declines from 1.0 to 0.5.
Stock yyy shows blue line declines from 1.0 to 0.7 and yellow line declines from 1.0 to 0.6.

Which stock is better? The "delta" is more favorable in stock yyy, but I think stock xxx still provides better value. Also, with stocks reporting closer to end of the week (Thursday), the yellow line will always lose much more percentage wise because it's all theta (no earnings IV), and theta 1-2 days to expiration is much higher. So again, it might be a bit misleading to compare blue and yellow lines between stocks that typically report on Mondays and stocks that typically report on Thursdays.  

[RapperT 336]

20 minutes ago, Kim said:

Yes, I think I understand what you are trying to achieve.

So the blue line tells you how much on average earnings straddle would lose in xxx days. This is basically similar information that we have in RV charts. Lets say RV shows average decline from 5% to 4%, your chart will show decline from 1.00 to 0.80. Correct?
 

Yes correct, both charts are similar

One  difference is that you are less likely to miss trades that have a high probability of success that may be expensive comparable to previous cycles (see: orcl last cycle).  That being said, both charts are helpful.  Im a paying subscriber of DJ's and use his charts often.

28 minutes ago, Kim said:

For example:
Stock xxx shows blue line declines from 1.0 to 0.8 and yellow line declines from 1.0 to 0.5.
Stock yyy shows blue line declines from 1.0 to 0.7 and yellow line declines from 1.0 to 0.6.

Which stock is better? The "delta" is more favorable in stock yyy, but I think stock xxx still provides better value. Also, with stocks reporting closer to end of the week (Thursday), the yellow line will always lose much more percentage wise because it's all theta (no earnings IV), and theta 1-2 days to expiration is much higher. So again, it might be a bit misleading to compare blue and yellow lines between stocks that typically report on Mondays and stocks that typically report on Thursdays.  

Im not totally following you here.  We are using calendar days in our forecast.  It's possible there are dynamic changes based on day of the week but I think our chart captures that.  I think the slope (especially day to day) is very important.

However, the discussion of what's MOST important is an interesting one.  I think I agree with you in that, if I can enter a straddle 25 days out from earnings and only risk a decrease of 1.0 to .8, I would take that trade every day and most likely before one that moves from 1.0 to .7

Again I am trying to identify edge more so than edge relative to other trades.  If I were selling strangles I would probably take both the trades you identified...

[SBatch 3,123]

35 minutes ago, RapperT said:

We've looked at using some sort of average but I'm not convinced that really helps that much.  I'm open to hearing why you believe it is optimal.

 

 

But can it hurt?  I think a 30 day average using data taken from 60 days to 30 days prior to the earnings release over a two year period may prove valuable.

[Stanislav 302]

.

Edited September 20, 2017 by Stanislav

[RapperT 336]

19 minutes ago, SBatch said:

But can it hurt?  I think a 30 day average using data taken from 60 days to 30 days prior to the earnings release over a two year period may prove valuable.

true..stand by.  Most recent 30 days probably makes sense

[jbthree 6]

I believe @Stanislav is correct.

[RapperT 336]

I havent read the book you keep referencing but my partner has..you dont need to simplify things for us   I just dont think what you're outlining is really necessary.

a:  we're using ORATs for the orange line (these calculations arent proprietary...they are published)

b:  we're normalizing the data which takes care of this cycle versus past cycles and combines the two types of IV (earnings plus market)

c: Normalization is standard practice in forecasting and takes care of the "apples to oranges"

d:  if you're making decisions based on RV, than the issues you're outlining are far more significant

 

 

Edited September 18, 2017 by RapperT

[Stanislav 302]

46 minutes ago, RapperT said:

I havent read the book you keep referencing but my partner has..you dont need to simplify things for us   I just dont think what you're outlining is really necessary.

a:  we're using ORATs for the orange line (these calculations arent proprietary...they are published)

b:  we're normalizing the data which takes care of this cycle versus past cycles and combines the two types of IV (earnings plus market)

c: Normalization is standard practice in forecasting and takes care of the "apples to oranges"

d:  if you're making decisions based on RV, than the issues you're outlining are far more significant

 

...Curious what methodology you suggest that is better than this?

Do not take my words on simplification personally. I actually simplified things for me. Since I realized that it is not easy to write a post which explained my point in details and at the same time not to quote half of the book.

 

I updated my post to include the original aggregate volatility formula from the book, found that formula not only contains coefficients but is also non-linear SQRT(A^2+B^2). This invalidates all my attempts to find some mathematical validity in subtracting blue and orange lines assuming they represent option prices or IVs. So my previous post can be safely ignored. Sorry for this.

Then, I was assuming you are plotting option prices (and not IV's). Since y axis did seems to be like option price. But now you are talking about normalization.

 

Thus, it turned out I still do not understand what and how you exactly calculate and plot. And how normalize the data. So I think there is no point to further discuss your implementation and the charts. I'll just assume you know what you are doing.

 

Quote

d:  if you're making decisions based on RV, than the issues you're outlining are far more significant

I believe RV charts have no such issues, since they just show straddle prices from different expirations expressed as percentages. It is very clear what they show, and what their limitations are. There are no attempts to subtract some BS based calculated volatilities e.t.c. All straddle prices have the same source (market) the dimentions match. Surely, these charts can be significantly improved if one implements a volatility model designed specifically for earnings events.

 

Quote

...Curious what methodology you suggest that is better than this?

I can not suggest a better that this methodology, since I do not completely understand the rationale behind you methodology as well as some implementation details (for example what is the normalization procedure).

 

But, if I was going to implement "better RV charts methodology", I would take Aggregate Volatility Formula and Earnings Volatility Model from Brian's book. Then implement a methodology to split market IV into EarningsIV and NormalIV (I am using terminology strictly as in book). Then I would think what our edge really is? We are trying to find situations when IV rise is more or equal to the theta decay.

 

So the first idea will be to plot RV charts using the EarningsIV component only, ignoring NormalIV component.

 

Second idea, take the Earnings Volatility Model and calculate the TrueGreeks, which take into account earnings events. Then we can plot charts like TrueVega/TrueTheta. That can possibly allow us to find situations where Vega/Theta ratio is good enough to enter the trade.

 

And third idea is having calibrated earnings volatility model run a MonteCarlo simulation on a range of potential price movements till earnings, average the results and check where average NetProfit/MaxDD ratio good enough to trade the strategy (this is exactly what Brian's EEVIntegrated2 Excel spreadsheet does).

Edited September 18, 2017 by Stanislav

[RapperT 336]

thanks for the response @Stanislav  This kind of discourse is good for everyone.

 

I dont really know how to make the rationale for our methodology much clearer.  Perhaps my post on the general board will help.  Much of the confusion is on me for just randomly throwing charts in threads without telling anyone what I was doing. 

 

We are just quantifying an identifiable edge. Kim has written about this edge, so have Jeff Augen and others...we just took the idea and quantified it using a well known pricing model.  We will be improving our model in the coming weeks (including a monte carlo)  as there is much room for this and we have some good ideas..most of which were spurred on by feedback from SO members.

Edited September 18, 2017 by RapperT

[RapperT 336]

example:

 

[IgorK 111]

7 minutes ago, RapperT said:

example:

 

@RapperT Can you please explain what exactly this chart shows?

[Stanislav 302]

@RapperT, thanks for a great discussion!

 

Quote

I dont really know how to make the rationale for our methodology much clearer.

I think all members can benefit if you provide complete description of what and how is calculated in such a way and in such details that other member could potentially take the data and recreate the implementation from scratch. This is the way RV charts were discussed and many members recreated the technology (including myself) and made some improvements (@Djtux's chart design is a big step forward!). Possibly your new Forecasting Charts thread can be a good place for this. Some pseudo-code could help to clarify charts implementation.

 

Surely, if you wish to leave the technology proprietary, that is ok too. That just won't make it easy for others to suggest how to improve these charts.

 

Quote

We are just quantifying an identifiable edge. Kim has written about this edge, so have Jeff Augen and others

As for the edge, yes I, as well as other SO members, understand the rationale behind pre-earnings straddle strategy. The issue is implementation specific. Me and some other members (@Kim), asked a reasonable question:

 

- What exactly subtracting blue line and orange line means, why this subtraction is legitimate, how can we mathematically prove that this really extracts or helps to identify an edge, what is physical sense behind these calculations?

 

@Kim formulated it in a different way, but I believe his line of thought is the same.

 

These are a kind of questions any good system developer asks himself when designing a trading system.

 

So precise, complete and clear description of the chart calculation can help to answer these questions.

 

Anyway, thanks for the discussion again!

I am going to monitor your Forecasting Charts and keep an eye on further charts development.

 

Edited September 18, 2017 by Stanislav