This forum was established to help traders (especially futures traders) by openly sharing indicators, strategies, methods, trading journals and discussing the psychology of trading.
We are fundamentally different than most other trading forums:
We work extremely hard to keep things positive on our forums.
We do not tolerate rude behavior, trolling, or vendor advertising in posts.
We firmly believe in openness and encourage sharing. The holy grail is within you, it is not something tangible you can download.
We expect our members to participate and become a part of the community. Help yourself by helping others.
You'll need to register in order to view the content of the threads and start contributing to our community. It's free and simple, and we will never resell your private information.
I didn't think you were dismissing the ideas at all. As I said earlier if I can contribute I am more than happy to do so. Had a look at the backtesting code and can see where you are going. I have not tried to run any of it yet. I think I understand where you are coming from in terms of new trading ideas.
Selection of strikes within the tool has a defined approach. Deep OTM strikes with deltas lower than 5. I tend to lean more on probability statistics. Therefore the dashboard. Probability is important for me both at the start of a trade and through the trade as a management tool. The missing piece for me has been getting a better understanding of time decay and rate of change in time decay. The ability to get time series option chains provides the data required to do a number of time series charts that might help with understanding the characteristics of time decay. My approach to this is to exposure myself to the experience rather than the theory or conceptualization of time decay ideas. The time series data I can access through your file will allow me to view and render what actually happens in difference commodities. With coding I want to do this on the fly and on demand. I want to look at these perspectives in as short a time as possible. But I want to be able to do it often to increase the magnitude of the experience. To me the power of the tool is not so much finding new trading ideas but better understanding the animal I am seeking to tame. I greatly appreciate the role futures.io (formerly BMT) has had in expanding my experience and exposure.
From the screen shots I wish to pursue further the time series nature of the data. For example I want to see the last 90DTE of last months expiring contract for each of the commodities I am considering as candidates. Then over time this experience will build as a sample and there may be new insights discovered.
For example, one of the other coding tasks I have been working on is to look at price range into expiry for different periods and reference probability to historical and seasonal price range results.
What you see in the screen shots is not complex. The dash board is simply excel formulas. The inputs are entered on the fly directly from an interactive option chain generated from your scanner. The full option chain is then used as the basis to build the chart. From the Scanner to the dashboard to the chart takes less than a few minutes at most. So now I can visually see the vol smile, the fat tails and what strikes are in and out of probability.
If I sent you the code right now you would curse me because it still needs to be consolidated, tested and tweaked. And you would spend to much time trying to get it to work. But I am happy to email it to you to browse. But it is in no state to post here. If you PM me your email address I will send you the code. But maybe it would be better to wait until I get the times series module working to my satisfaction.
The following 2 users say Thank You to BlueRoo for this post:
BlueRoo - You've done some very nice work; I've seen commercial packages that are not that well designed. Bonus points for putting a visual representation of the StdDev ranges and position break-even points on the chart.
Couple of comments/suggestions.
- I noticed you have the Hoadley Options tools and it looks like you've calculated the probability of an option expiring worthless (perhaps you used HoadleyProbAtEnd). If you're so inclined, you might also use his formula for the probability of touching the strike at any time during the life of the option (HoadleyProbAnyTime1). As you probably know, the latter is often 2x the former, but it can vary beyond that rule of thumb with a higher IV and greater time to expiration.
- I believe the lower section of your dashboard (the probability section) was calculated using the standard normal (Gaussian) distribution. Depending on the input parameters, that may be close enough, at least to calculate one StdDev, but I'd strongly encourage you to use the lognormal distribution. That's the underlying distribution assumption for the models and programs the pros use, and whether or not it's the best one, it's a fact that prices cannot go below zero. You can adjust for skew, kurtosis, jumps and so on in more sophisticated models (and Hoadley has some examples), but you'll have a more accurate picture of probable price ranges if you switch to lognormal.
I hope I'm not sticking my nose where it doesn't belong. These are just friendly suggestions and there's no way I could replicate what you've done. My only expertise, such as it is, is that I started with the normal distribution, too, realized it wasn't the right one, and then spent way too long trying to get the correct lognormal calculations into Excel (it's been a long time since I sat in a statistics classroom).
Last edited by CafeGrande; May 11th, 2014 at 10:44 AM.
The following user says Thank You to CafeGrande for this post:
CafeGrande, I appreciate the post and your insight. Without your post I would not be pushed to review and learn. Thanks for the suggestions. As a result I have done the following...
See the new screenshot below. Includes normal, lognormal and hoadley probabilities. The previous screenshot only uses formulas with not hoadley. I didn't want to include them because possibly sharing the file and it not working for others. Anyway I have now because the discussion is about the goodness of fit of the option chain to normal distribution. Clearly from the chart it is not normally distributed. I have added descriptive statistics to further see this discrepancy. This is done for each variable option price, volatility and delta. This is exploratory. I am trying to better understand the behaviour of the asset. Observations: LogNormal Probability of expiring in the range is much larger. However hoadley probabilities are similar to normal. This is interesting to me and thankyou for pushing me to explore it further. In the case in question the strikes of the spread a pretty much where I would place the trade give everything else is constant. I like you realize lognormal is a better fit for the option chain. But I have not had enough experience looking a real world results to say if the difference provides a significant enough result to change strikes. In this case the hoadley probabilities may be the umpire. I would be interested in understanding more from your experience if you have seen a significant difference.
Please register on futures.io to view futures trading content such as post attachment(s), image(s), and screenshot(s).
First, I don't know how you added all those elements since I posted my message; because of the time difference, you should have been sleeping most of the time. You must know this stuff like the back of your hand if you can do it that fast!
I am attaching an image that shows At Expiration, At Any Time, Prob of expiring in range, and Standard Deviations.
The Excel snip (shaded in light blue) is something I put together. It looks like there are some differences with your program. As background here's where my numbers came from:
The At Expiration and At Any Time are from Hoadley formulas and you can see the input parameters that I borrowed from your image. My At Expiration matches to two decimal places with a "test" source I use - a probability calculator from ivolatility.com (same calc they license to OptionsHouse, TradeKing, Scottrate and probably others). My At Any Time doesn't match precisely, which puzzled me for a while until I figured out that the Hoadley default is set to "continuous" price observations and that makes a difference. If you ratchet it back a little you can get it to almost match the ivolatility test calc (I don't know what ivolatility's settings are).
The standard deviations in my calc also match the test calc but they came from an el-cheapo ebook I bought online. I could not find them in Hoadley.
When I was putting the simple spreadsheet calc together I also tested it with the online Hoadley calcs but they restrict you to x number of calculations per day (it's fairly small).
I'll address your other comments in a separate post.
The hoadley formulas in my sheet have expected return as zero. My Hoadley is over 10 years old as well. Might be the reason for the slight differences?!?!
My formulas for lognormal are definitely off! I need to have a play with them. Can you post the formula's in your cells for your log normal standard deviations. It might give me a clue to what I have done wrong.
Pretty fast in most software stuff, but the back of my hand is a little shaky at times. Need some trading buddies to help me keep on focus. Looking forward to your further response.
BlueRoo - I've attached the formulas for 1 to 3 standard deviations using the lognormal distribution. Rows 7 and 8 (I think) are used as a shortcut so the long formula in Row 7 doesn't need to be repeatedly typed in if you want to make 1.5, 2.5 etc SD calculations. In my little Excel calculator I keep it down at the bottom (not visible) but with a little bit of typing it looks like it could be eliminated.
Some notes on the input box (rows 2 through 5). You probably already know these, but someone else might want to use the formulas and this will save them some head-scratching:
1. "Mean" is Expected Return/Risk Free Interest Rate
2. "Time" is in fractions of a year. I changed it to DTE and adjusted the formula slightly (/365)
3. "St Dev" is expected volatility, I renamed it to avoid confusion with the output SDs
Last edited by CafeGrande; May 12th, 2014 at 08:37 PM.
The following user says Thank You to CafeGrande for this post:
Update to Normal and LogNormal Standard Deviations. See screen shot below.
For Normal my formula is - =EXP(($E$8)*SQRT($E$6/365)*-1*NORMSINV(C25))*$E$5
For LogNormal my formula is - =EXP(($E$8)*SQRT($E$6/365)*-1*LOGINV(C25,($H$30/100),($J$30/100)))*$E$5
Removing the excel *INV function and reverting to what you posted you can see the slight differences.
Please register on futures.io to view futures trading content such as post attachment(s), image(s), and screenshot(s).
That's valuable! I don't know if you plan on archiving it, but in my experience it's useful to have that type of history, although it's expensive to procure.
It's easy and cheap enough to buy a time series for ATM IV, but everything changes when you want a particular option-delta-IV. Nice work!
I'm going to have to start drinking that Oil Can beer (Foster's). Must be something in it because you and Peter Hoadley are pretty darn smart.
Not sure what you mean by archiving the chart. Is that something you do on futures.io (formerly BMT)?
Yes it is the standard theory time decay curve that got me frustrated once because it did not seem to apply to OTM options. And then I discovered in was solely basted on ATM.
I am interested in pin-pointing the greatest rate of change. I am also interested in taking a sample of say 90 different option series 90DTE and observing what variance there is in decay. Then I start thinking does decay have different rates of change in different seasons?
Anyway without the work Dudetooth and others had done on the code I would not have been able to do this. I hope this contributes something new to the table.