SteadyOptions is an options trading forum where you can find solutions from top options traders. TRY IT FREE!

We’ve all been there… researching options strategies and unable to find the answers we’re looking for. SteadyOptions has your solution.

Kim

Welcome to Steady Futures

438 posts in this topic

Recommended Posts

Thanks @Kim

 

So as I said on the commodity board, this particular system is the brainchild of my business partner (and brother) John.  I will post his overview below and then give a schedule of what to expect in the next few days.  John will be joining up and chiming in shortly.  The formatting from the copy and paste is a little odd, sorry about that, wont let me edit.

 

Commodity Trend Following System Using Options

 

Introduction

Last year I developed a 100 day breakout system for trading commodity futures. The system worked well and surprisingly replicated the performance of some well known CTAs very closely. I quickly ran into difficulties regarding risk. The first issue was simply contract size. In my system I would often have to open a new trade with a minimum open risk of $15K. Considering the system had around a 40% win rate this was problematic for anyone trading less than a million dollars. The other issue I ran into was the fact that open risk varied throughout the life of the trade. For example, if a trade ran in a favorable direction the total open risk would often increase. As I got closer to a stop my open risk would decrease. This meant my day to day portfolio variation was very hard to model. I decided to develop a new trend following system that would allow me to capture asymmetric trend following type returns but with a process for maintaining constant risk and that was safe to execute for smaller portfolio sizes.

It was important to me to have the system only be long volatility (debit trades) on options. I know there is a ton of confusion around selling theta driven by people with a vested interest in getting people to trade a lot. I don't want to get into it here but selling far OTM options is a really dumb idea. My system would only be long options. I'm happy to walk anyone through (in a non calculus way) the math behind why selling options is dumb and theta is empirically and theoretically not an edge if there's interest but for this post let's just say, I'm only interested in long volatility.

After quite a bit of research I opted for creating a stop and reverse type system utilizing commodity options. This means that the system is always long or short every commodity that is included in the portfolio. I also wanted the system to be relatively easy to manage (I work full time, co-founded a startup in health care analytics and am a PhD candidate in financial economics...life is busy). To build this system I borrowed some ideas for time series analysis from econometrics.

System Schedule

The system runs every Thursday night and generates a signal before market open on Friday. I close all open trades and open new trades on Friday morning. This allows me to reduce risk on commodities that have been trending favorably as needed (in some cases add risk back on) and add risk for those commodities that lost money over the previous week. This means that the risk control for the portfolio is under far more control than a blind trend following strategy. While I check the trades through the week (I get curious) I do not modify open trades. No adjustments, no taking profits. The trade runs until the following Friday.

The options that are currently in the system are:

  • Crude Oil

  • Gold

  • Corn

  • Lean Hogs

  • Sugar

  • Coffee

  • Wheat

  • Soybeans

 

Signal Generation

We generate our signal using an ARIMA model. An ARIMA model sounds complicated so I'll break it down. Generally it is a tool used by econometricians to forecast time series and is particularly helpful when a time series shows evidence of non-stationarity (geek speak for trending). There are three parts to the model. The first is the auto-regression piece (AR). This piece forecasts a future value based on a regression against lagged values in the time series. The second piece is "integrated" (I). This refers to a technique used to make a time series stationary through differencing. (You can forecast stationary time series...non stationarity causes problems so the "I" term converts a trending series to stationary for the purposes of the forecast). The final piece is a moving average (MA). This is not a standard moving average you'll see in technical analysis books. The MA says that the error terms are a linear combination of previous error terms. Each of these items are a parameter in the model. For example, I could run the model on gold with parameters (3,0,1) which will give me a different forecast than (1,1,1).

Every week we run this model on every possible combination parameters for each futures contract and then select the parameters that return the lowest Akaike Information Criteria. This can be thought of as a relative "goodness of fit" measure. Technically this means we are data mining a bit. While I wouldn't try and get that past a peer review process for a journal I'm happy to use it for trading. Once we have the best forecast for the training time window (empirically that window needs to be between several months to several years) we forecast next week's return.

While we track traditional forecast accuracy what we're really shooting for is getting the direction right enough to make money. Using long options already gets us some asymmetry in our results and then the fact that we can ride some really big moves with leverage at defined risk improves our results.

Risk

We think about risk in two ways. The first is total possible loss on the trade. This is determined by the amount of the debit in each option trade. For example, if I'm trading all four commodities and I'm risking $1K per contract my top risk is $4K for the week. While it should be rare for each trade to lose 100% we expect it to happen at some point. This is a mathematical certainty if you trade long enough (one of the reasons leveraged mechanical option selling is stupid).

The second way we look at risk is delta targeting based on the last 4 weeks range in weekly returns. If based on the last 4 weeks the commodity moved 5% on average we will size the deltas in the trade to make sure the average loss is around 1/2 a percent. I'm still writing the code to automate this piece but expect it to be complete in the next week or so.

We take the lower of the two positions recommended by the criteria above.

Results so Far

First off, these are real results based on real trading in a real money account. That being said, we are not making any promises to anyone about anything. All the normal disclaimers.

We began trading the system in August with oil and gold. Since then we have 34 closed trades (and currently 4 open trades). If you're doing the math, yes, we missed a couple of weeks for various reasons. We haven't had any major trends which we think will help our win percentage over time but we will always lose far more than some popular options strategies. Our average win is 69 percent and our average loss is 40 percent. Our largest win was the week of November 16th when we made 242 percent on oil (over a three week period we've made almost 400 percent in oil). We've had multiple wins over 100 percent. Our largest loss was 66 percent in oil on October 26th.

I think a couple of notes are important. We never know where the win will come. So as mentioned above, we trade every contract every week. If you were to run a monte carlo with our results you would see that losing streaks are expected and happen. This is a long term strategy. One of the nice things about it is that it is scalable so as your account grows you should be able to keep increasing size.

-John and Rich A

Edited by RapperT
added new contracts

Share this post


Link to post
Share on other sites

What to expect over the next couple of week:

  1. We expect live trading to begin in about 2 weeks.  Sizing the trades is slightly more complex than equity options so we want to see how the current vol of the system would impact accounts of different sizes.  Quantifying this piece will allow us \to take the guesswork out using the signals to execute actual trades.
  2. We will provide a general overview on trendfollowing in general as it is very different than the other strategies employed here at SO.  Jesse also does a great job in his threads on the lorintine pages.  I highly recommend checking those out.  It's important that expectations are set properly as some of the keys to success in TF are slightly different than in our earnings strategies.
  3. We will also provide an intro to options on futures.
  4. We will explain why we go long vol
  5. Most importantly, we will post a discussion on the importance of long term commitment in trendfollowing. 

 

Down the road I'm hoping John will post his paper on forecast accuracy so all you geeks out there can get down into the nitty gritty.

 

As I said in the other thread, John and I are very excited for the opportunity to be part of the team here.  We both respect SteadyOptions a lot and look forward to interacting with its members.

Edited by RapperT

Share this post


Link to post
Share on other sites

Hi everyone,

First off, I'm excited to be here and get a chance to work with everyone.

A couple of notes:

We actually wrote that intro about a week ago and a few questions were asked by some people we sent it to.  One was, "Why are you worried about cycling through every possible parameter for the model and selecting the best?"  At the time I felt like I was data mining a bit doing that but the question prompted me to go back and dig a bit deeper on the time series literature.  Apparently cycling through every parameter is accepted for time series forecasting because you are not trying to explain causality with the model, just the best fit.  So who ever asked us that thanks; it led to me digging a bit deeper on some best practices for time series forecasting.

We also added sugar to our system last week.  We like sugar because it trends well and isn't too correlated with the other contracts.

I'll post some system info on forecast accuracy over the next few days.  Backtests are difficult with options for a whole slew of reasons but I do have trade history at this point.  We will need to take the results with a grain of salt because of the small data set but I'll run some simulations bootstrapping the data I have to get some hypothetical risk parameters for the system using options.

 

Thanks!

John

Share this post


Link to post
Share on other sites
9 minutes ago, Jjapp said:

Hi everyone,

First off, I'm excited to be here and get a chance to work with everyone.

A couple of notes:

We actually wrote that intro about a week ago and a few questions were asked by some people we sent it to.  One was, "Why are you worried about cycling through every possible parameter for the model and selecting the best?"  At the time I felt like I was data mining a bit doing that but the question prompted me to go back and dig a bit deeper on the time series literature.  Apparently cycling through every parameter is accepted for time series forecasting because you are not trying to explain causality with the model, just the best fit.  So who ever asked us that thanks; it led to me digging a bit deeper on some best practices for time series forecasting.

We also added sugar to our system last week.  We like sugar because it trends well and isn't too correlated with the other contracts.

I'll post some system info on forecast accuracy over the next few days.  Backtests are difficult with options for a whole slew of reasons but I do have trade history at this point.  We will need to take the results with a grain of salt because of the small data set but I'll run some simulations bootstrapping the data I have to get some hypothetical risk parameters for the system using options.

 

Thanks!

John

pretty sure that was @Christof+  tagging him here in case he wants to chime in

Edited by RapperT

Share this post


Link to post
Share on other sites

I am assuming you will be giving us suggestions on how to structure a portfolio, as Kim has done with SO. Do I take $25k and place $5k on each of the 5 options for a given week? Do we want to have funds in this portfolio over and above the actual trades? Do I take funds off the table every once on a while in case of the inevitable complete loss on all options in a given week?

Very much looking forward to your new product. Good job on putting this together.

Share this post


Link to post
Share on other sites

We'll definitely be putting together some suggestions on position sizing and portfolio construction.  In general though, yes, you want to have funds over and above the actual trades.  We're trading options (leveraged) on a leveraged underlying (commodity futures) so you really don't want to be too big. I've also had 4 losing weeks in a row.  That wasn't a 100% loss...but 4 weeks where the system was negative for the week.  So on 25K you might only have 2K in risk on but we're doing some work to to better quantify that.

 

I may have overstated the complete loss thing a bit.   That's dependent a bit on what expiration you use and how far OTM you go but we're trying to be hyper conservative especially in the beginning.  If you're 50 to 60 days out on an option the potential for a complete loss is lower than an option 2 weeks out (your capital in the trade for 50 days out is also higher).  I wrote that right after the optionsellers guy blew up and so it was front and center in my mind at the time.  I've been known to short vol in some cases but never naked and in this system it doesn't make sense to be short vol (the system only works because the average win is much larger than the average loss). 

 

We're also trying to balance how many positions we trade with account size etc.  More positions means less volatility in the returns stream. 

 

In general though with a trend following strategy there will be drawdowns.  You have to play the long game which means relatively small position sizes. 

Share this post


Link to post
Share on other sites

 

@Jjapp, @RapperT Curiously looking forward. You definately have a case for using ARIMA on commodities.

 

One more idea:
In fact, in one paper (direct download http://tiny.ph/PR90 ) in my dissertation ( http://tiny.ph/3kAf  amazon) I dealt a bit with autoregressive behaviour in commodities empirically. Not only there is evidence but for a number of markes I also find a positive relationship between level and strength of autoregressive behaviour and the level of non-commerial activitiy by fitting an AR model which smoothly transitions between states. (This is intuitive since it is mainly the non-commercials like CTAs and guys like you who engage in trendfollowing - which again reinforces autoregressive behaviour).

 

If that still holds it could potentially mean that a model like yours has a better edge in markets and times with high degree of non-commercial participation.

 

 

Share this post


Link to post
Share on other sites

@RapperT  I worked nearly a decade as a researcher and then equities quant portfolio manager for an (at least in European terms) not-so-small insitutional quant shop (quoniam.com). But then got fed up. Now I run a couple of own projects. 

 

The non-commercial effect was clearly there and it is nice since it is very intuitive. But it was not super strong and not in all commodities. So probably nothing to focus on by itself. But in case you plan to add some filtering for the model I could imagine it could qualify well for one of the factors to consider..

Edited by Christof+

Share this post


Link to post
Share on other sites

@Christof+Thanks for the paper.  That is really helpful.  I need to dig into some of the references a bit.  I'm not familiar with some of the models you used.  We are considering adding some filtering.  Honestly, the big problem now is prioritizing where we dig into next.  There are a lot of ways we could improve the model we're using.

Share this post


Link to post
Share on other sites

 

3 hours ago, Jjapp said:

I need to dig into some of the references a bit.  I'm not familiar with some of the models you used.

STAR models are not standard of course. Thought it could be interesting for you because estimating on Thursday hints that you already employ CoT data anyway. I used it because I needed something to allow for a smooth movement between states (representing the varying degrees of speculators' positions in a market), markov switching could only do 'hard either or'. And I liked the idea and flexibility of the model (I remember however that at some point I ran into quite some difficulties estimating it and no one was up for help. So I finally wrote to Teräsvirta himself. And he actually bothered to answer which I found quite cool for a little guy like me).

3 hours ago, Jjapp said:

Honestly, the big problem now is prioritizing where we dig into next.  There are a lot of ways we could improve the model we're using. 

And that will probably be the much longer part than coming up with the model first hand (the unavoidable subjectivness in an objective model).
 

Concerning backtest, I do not know a cheap source for commodities options data and I also do not have any unfortunately. Simply running some rudimentary backtest with commodity futures prices alone for a start could not give some insights?

 

Really looking forward what you guys will be coming up with :)

Share this post


Link to post
Share on other sites

I can post the forecast stuff and some backtest info for what the system would look like for straight futures in the next couple of days (traveling to a client's location right now and the model we're running on the server is slightly different than what we ran the initial tests on).  The options stuff is a bit more difficult for a bunch of reasons including not having a good historical data source and the time it will take to code all the rules we're using to select the actual contract.  My plan for a first pass at that was to take the results I have so far and use it to build a simulation bootstrapping the returns we have.  That should be relatively easy to rerun as we get more trades also (right now we have 15 weeks/47 trades of data which I don't think is enough to really trust the results.)  @Christof+ there might be a better way to do it then that.  Open to any thoughts.  I thought that would get us some rough idea of the stats without necessarily having to try and model all the options pieces along with position sizing.

Longer-term we will figure out how to get the options data because I would rather use the IV of the option for risk management than historical returns which is what we're working on now.

Share this post


Link to post
Share on other sites
2 hours ago, Christof+ said:

STAR models are not standard of course. Thought it could be interesting for you because estimating on Thursday hints that you already employ CoT data anyway. I used it because I needed something to allow for a smooth movement between states (representing the varying degrees of speculators' positions in a market), markov switching could only do 'hard either or'. And I liked the idea and flexibility of the model (I remember however that at some point I ran into quite some difficulties estimating it and no one was up for help. So I finally wrote to Teräsvirta himself. And he actually bothered to answer which I found quite cool for a little guy like me).

I thought the smooth transition between states was cool and could maybe be used to help with position sizing.  For example, maybe you lend the forecast a higher weight when the speculator's relative position in the market is high.  Also very cool that Terasvirta was willing to help out! 

Share this post


Link to post
Share on other sites
12 hours ago, Jjapp said:

@Christof+ there might be a better way to do it then that. 

Yes, imho bootstrapping with 15 weeks of data is just not worth the trouble at all. Better not even look at it. Two of many reasons to consider:
- Amplifying a microscopic small part of recent history you are 100% sure to not get anything representative for history (look at crude's behaviour over the last 15 weeks for example).
- You build on ARIMA so you need to be careful not to destroy time dependent/autoregressive characteristics while bootstrapping. Some of the block bootstrapping techniques might be a good choice (i.e. bootstrap not from single returns but from chunks of consecutive data). But that further increases your need for data.

 

 @Kim 's suggestion of a simple 'manual' backtest is a shot, despite the manual work? I would urge you to look at least at the difficult times that way. I learnt that you can know a model only (if you can know it at all) when you understood what is happening during the most adverse market conditions. Imho this is crucial, bootstrapping so short will not give you anything even close to useful here either.

Take classical options writing as an obvious example (although returns of your model will look very different). It does work nicely in the long run. But returns have that nasty negative skew, these rare but hefty large negative outliers. Say you built an options writing model and test it with bootstrapped data from the wrong sample (one that does not allow for such a steam roller to be produced in the data) even hundreds of years of simulated data will not give you an adequate picture, it is practically worthless. And this is very dangerous. You are extremely likely to draw deadly wrong conclusions but at the same time are sure that you did your homework well.

 

 

Share this post


Link to post
Share on other sites

@Christof+ you’re suggesting manually placing trades in ONE  to backtest?

 

im not opposed but I’m also not sure how much that would tell us unless we did it for an extended period which would be exceptionally inefficient.

 

October for example would qualify as a difficult time for most classic TF.. we are much shorter based time frame wise ( which also can have  its downsides).  Most big shops got crushed.  

 

...just want to be sure I understand you

Edited by RapperT

Share this post


Link to post
Share on other sites

@RapperT That's completely up to you of course. Maybe I got you wrong concerning the actual work required. Just thought since you trade only each Friday morning that it was not more than manually collecting a few prices each Friday, plug them in excel and then use them to calculate weekly P/L figures over a longer time frame?

Share this post


Link to post
Share on other sites
1 hour ago, Christof+ said:

@RapperT That's completely up to you of course. Maybe I got you wrong concerning the actual work required. Just thought since you trade only each Friday morning that it was not more than manually collecting a few prices each Friday, plug them in excel and then use them to calculate weekly P/L figures over a longer time frame?

well its not quite that straightforward.   we would need the forecasts to run for each friday in the back test period which requires recalculating the parameters for each contract.   If we go back several years that would take a lot of time.  Obviously it would be beneficial and that is the kind of stuff we will devote more time to when we are doing this full time.  We both travel a ton for work.

It will be interesting after the the 6 month trial period to compare our system to some of the CTA indices.  We will have a year of real trading at that point

Share this post


Link to post
Share on other sites
4 hours ago, EmilyF2 said:

Curious what basic knowledge and capital requirement are needed to participate in this service? Any basic course/book recommendation for futures? Thank you. 

Hi Emily,

We will give you all the basic knowledge you need to start trading the system as well as posting the actual trades we take each week.  We will also be adding educational topics on trendfollowing, our approach, and trading with futures options.  You should have most of what you need.  In addition we will be linking relevant articles and podcast from some of the great trendfollowers.

We are recommending a min account size of 25k and will be posting a topic on risk/allocation per trade before going live.

Share this post


Link to post
Share on other sites

Just to clarify, 25k account size is recommended in order to be properly diversified and still keep at least 50% of the account in cash. Average position size could be anywhere from $500 to $2,000. So theoretically, you can trade with smaller account, but it will be much riskier. 

Share this post


Link to post
Share on other sites

I know there are currently 4 commodities that are being traded.

I guess my question is what drives the selection of the underlying commodities and, 

is this a situation where risk is reduced by being in a greater number of underlyings?

In other words, can you start with a subset of the 4 commodities, and add the additional commodities as one gets comfortable with the system and amount at risk?

Share this post


Link to post
Share on other sites
29 minutes ago, Kim said:

Just to clarify, 25k account size is recommended in order to be properly diversified and still keep at least 50% of the account in cash. Average position size could be anywhere from $500 to $2,000. So theoretically, you can trade with smaller account, but it will be much riskier. 

Thanks Kim.  yes we dont want to exclude anyone but in small accounts, the positions would be larger relative to account size (hence increasing risk)

Share this post


Link to post
Share on other sites
24 minutes ago, NJ_KenRob said:

I know there are currently 4 commodities that are being traded.

I guess my question is what drives the selection of the underlying commodities and, 

is this a situation where risk is reduced by being in a greater number of underlyings?

In other words, can you start with a subset of the 4 commodities, and add the additional commodities as one gets comfortable with the system and amount at risk?

a:  yes more commodities reduces risk, we will be going up to ten over time.

b:  if you trade the system, you need to take every trade every week.  We dont know when/where the winners will come.  Obviously we willadjust risk exposure per position as we add new tickers. Jerry Parker (one of all time greats) has a story about returning something like 35% one year in one of his systems and it was all in one month on one underlying.  Im paraphrasing but you get the picture, i will  find the actual podcast.  We have a post forthcoming on this but if you have larger accounts, my suggestion is trade smaller at first but commit to a long period of time (several months at least...the longer the better). 

 

Share this post


Link to post
Share on other sites
13 minutes ago, RapperT said:

a:  yes more commodities reduces risk, we will be going up to ten over time.

b:  if you trade the system, you need to take every trade every week.  We dont know when/where the winners will come.  Obviously we willadjust risk exposure per position as we add new tickers. Jerry Parker (one of all time greats) has a story about returning something like 35% one year in one of his systems and it was all in one month on one underlying.  Im paraphrasing but you get the picture, i will  find the actual podcast.  We have a post forthcoming on this but if you have larger accounts, my suggestion is trade smaller at first but commit to a long period of time (several months at least...the longer the better). 

 

@RapperT You might have already mentioned this, but based on your backtesting over several years what was the average annual return on account? Also, what percent of the account was invested to come up with the return? I realize that what you observed in backtesting might not project to what will occur going forward or what will occur with our own trades.

Actually, one other question: how scalable is this method of investing?

Edited by Alan

Share this post


Link to post
Share on other sites
15 minutes ago, Alan said:

@RapperT You might have already mentioned this, but based on your backtesting over several years what was the average annual return on account? Also, what percent of the account was invested to come up with the return? I realize that what you observed in backtesting might not project to what will occur going forward or what will occur with our own trades.

Actually, one other question: how scalable is this method of investing?

I just noticed a December 10 post that seemed to indicate that due to the nature of the forecasting a prolonged backtest was not done.

Share this post


Link to post
Share on other sites
25 minutes ago, Alan said:

I just noticed a December 10 post that seemed to indicate that due to the nature of the forecasting a prolonged backtest was not done.

We're working on a prolonged backtest that will reflect what this would look like trading the underlying futures versus the options.  That should give you an idea.  Computationally doing that for any meaningful time period will lock up my laptop so I'm rewriting some of the code and getting ready to run it on a server.  We're still a couple of weeks out from having that completed. 

This method of investing (Trend Following) is highly scalable.  You won't have issues with scaling this up.

Share this post


Link to post
Share on other sites
2 hours ago, Alan said:

I just noticed a December 10 post that seemed to indicate that due to the nature of the forecasting a prolonged backtest was not done.

edit:  i missed John's response

Edited by RapperT

Share this post


Link to post
Share on other sites
2 hours ago, Jjapp said:

 

This method of investing (Trend Following) is highly scalable.  You won't have issues with scaling this up.

@Alan Hell you could always trade the  futures if you can handle the underlying notional value.

 

 

Also, only about 5-6% of commodities trading involves TF strategies so you're even diversified within that particular market.

Edited by RapperT

Share this post


Link to post
Share on other sites
On 12/9/2018 at 5:58 AM, RapperT said:

It was important to me to have the system only be long volatility (debit trades) on options. I know there is a ton of confusion around selling theta driven by people with a vested interest in getting people to trade a lot. I don't want to get into it here but selling far OTM options is a really dumb idea. My system would only be long options. I'm happy to walk anyone through (in a non calculus way) the math behind why selling options is dumb and theta is empirically and theoretically not an edge if there's interest but for this post let's just say, I'm only interested in long volatility.

Hi,

 

Can you please briefly explain why selling options is not a good idea ? I am really intrigued by your statement because lot of theory and trading out there emphasizes on option selling.

 

Do you mean option selling is not a good idea even when IV is high ??

Share this post


Link to post
Share on other sites
48 minutes ago, Manish71 said:

Hi,

 

Can you please briefly explain why selling options is not a good idea ? I am really intrigued by your statement because lot of theory and trading out there emphasizes on option selling.

 

Do you mean option selling is not a good idea even when IV is high ?? 

Hi,

I do sell options on occasion when IV is high.  There are a couple of caveats though.  First is I have to have a reason for thinking IV is high.  Usually that reason is that is high in relation to other options in the chain or high based on some model I have.  If every option in the chain has roughly the same IV I'll probably assume that is a pretty good forecast of volatility unless I have a model that tells me otherwise.  When I sell options they are always close to at the money and hedged.  They're hedged because I don't like the risk payoff of a naked short option.  They're at the money because the impact of gamma on your position when a short far out of the money trade goes against you is catastrophic.  I realize a lot of people don't agree with that.  I have no problem with "income" trades in general (butterflies, Iron Condors, etc).  I do think people that suggest selling naked far OTM options is some sort of edge are just wrong.

 

 

From an asset pricing standpoint all asset pricing is based on the "fair game" assumption.  I think that this is usually a good assumption.  Under the classic BS model this means that the expected returns from theta will be offset by the expected returns from movement in the underlying. This means that if selling theta mechanically was an edge than everyone would sell theta until the price of the option dropped and the edge went away. 

 

 

In the case of this system we are not modelling the underlying volatility of the future to price the options better than the market.  We're using options to take a directional bet on the underlying future.  In that case we want to be long vol because it gets us the risk/return profile we need to make the system's P/L attractive.  This is because we're counting on fat tails in the return stream of the futures to continue to occur.  A long vol trade lets us capture that.

 

 

-John

 

 

 

Edited by Jjapp

Share this post


Link to post
Share on other sites
2 hours ago, Manish71 said:

Hi,

 

Approximately what kind of returns can you expect ? Lets say you allocate 30% of your $25k futures portfolio to commodities trading and the rest is in cash. Can you beat S&P return ?

 

Thanks. 

 

We're working on a post that covers position sizing, allocation and leverage.  We're trading options on futures both of which are highly leveraged so staying small is important.  The good news with that is you should have a lot of cash left in your portfolio.

@RapperT  covered some historical performance of trend followers here: 

 

That's probably a good place to start.

We're working on some overall system stats we will share.

 

 

Share this post


Link to post
Share on other sites
On 12/9/2018 at 5:58 AM, RapperT said:

This allows me to reduce risk on commodities that have been trending favorably as needed (in some cases add risk back on) and add risk for those commodities that lost money over the previous week.

Pardon my ignorance or understanding, but why does risk increase when the trade trends favorably and risk decreases when it is in red ? Should it not be the other way around ?

Share this post


Link to post
Share on other sites
9 hours ago, Manish71 said:

Hi,

 

Can you please briefly explain why selling options is not a good idea ? I am really intrigued by your statement because lot of theory and trading out there emphasizes on option selling.

 

Do you mean option selling is not a good idea even when IV is high ??

Site like TastyTrade became very popular in the recent years. They are using those highly catchy reasons like "80% of the options expire worthless, so selling them gives you an edge.", "Options Selling has high probability of success (80-90% winning ratio)." etc.

But try to ask them for their track record, and they will use all kinds of excuses why they cannot provide it. Those who followed those guys can see what happens when you are short vega short gamma and volatility spikes. The last few weeks are the best example.

As @Jjapp mentioned, it doesn't mean you should not sell options at all. But having onlyoptions selling trades in your options portfolio is a certain path to ruin.

I know many tastytrade style followers will not agree with us, but the track record is the best proof. Numbers don't lie. Tastytrade has been trying to prove for years that buying premium before earnings doesn't work - and we are debunking their "research" time after time.

More details: 

Selling Options Premium: Myths Vs. Reality

 

Share this post


Link to post
Share on other sites
9 hours ago, Jjapp said:

Hi,

I do sell options on occasion when IV is high.  There are a couple of caveats though.  First is I have to have a reason for thinking IV is high.  Usually that reason is that is high in relation to other options in the chain or high based on some model I have.  If every option in the chain has roughly the same IV I'll probably assume that is a pretty good forecast of volatility unless I have a model that tells me otherwise.  When I sell options they are always close to at the money and hedged.  They're hedged because I don't like the risk payoff of a naked short option.  They're at the money because the impact of gamma on your position when a short far out of the money trade goes against you is catastrophic.  I realize a lot of people don't agree with that.  I have no problem with "income" trades in general (butterflies, Iron Condors, etc).  I do think people that suggest selling naked far OTM options is some sort of edge are just wrong.

 

 

From an asset pricing standpoint all asset pricing is based on the "fair game" assumption.  I think that this is usually a good assumption.  Under the classic BS model this means that the expected returns from theta will be offset by the expected returns from movement in the underlying. This means that if selling theta mechanically was an edge than everyone would sell theta until the price of the option dropped and the edge went away. 

 

 

In the case of this system we are not modelling the underlying volatility of the future to price the options better than the market.  We're using options to take a directional bet on the underlying future.  In that case we want to be long vol because it gets us the risk/return profile we need to make the system's P/L attractive.  This is because we're counting on fat tails in the return stream of the futures to continue to occur.  A long vol trade lets us capture that.

 

 

-John

 

 

 

I agree. The real edge is in being open minded and flexible. Knowing that there is a time and reason where buying options is preferable, and vice versa.

I think one is setting themselves up for failure to ONLY sell options because the reasoning is not sound.

I also very strongly agree that, when you are a net seller, you HAVE to be hedged against any form of open ended risk (Cordier as the most extreme example)..

 

,I also think that the concept of buying options has been generalized by those in that camp to only mean buying OTM options and holding until expiration,  "hoping" for an outsized, unrealistic move in the underlying which will mostly never happen.

 

Nobody ever points out all of the other ways one can buy options.

One of the basis of their arguments is that options are a "wasting asset", and because of time decay, statistically you WILL lose over the long run.

 

They never mention a situation like buying a 30 day, 50 delta option, for leverage,and holding it for 1-2 days, to catch a 2-3% swing in the underlying.

Are these part of the xxx% of options that expire worthless?

How much time decay is there on a 4 week option when held for 1-2 days....or a 50/25 (delta) vertical held for 1 week?

 

The reasoning is totally flawed.

 

I have done well both as a net buyer, and seller , of options by having some sense of when the time is right for each method.

But, over the long run, I definitely have earned more on positive gamma, long option trades, when held for "reasonable " amounts of time.

 

The Tasty Trade people are screaming and shouting about how great the current environment is because now that IV is high, they can just sell sell sell, with abandon.

I would really like to know exactly HOW they are going about their selling.

 

My guess is that, over the long run, it is not going to be a pretty picture.

And pushing this on their "followers" is totally irresponsible, and they will have a lot to answer for when these people blow up.

Edited by cuegis

Share this post


Link to post
Share on other sites
6 minutes ago, cuegis said:

The Tasty Trade people are screaming and shouting about how great the current environment is because now that IV is high, they can just sell sell sell, with abandon.

I would really like to know exactly HOW they are going about their selling.

 

My guess is that, over the long run, it is not going to be a pretty picture.

And pushing this on their "followers" is totally irresponsible, and they will have a lot to answer for when these people blow up.

And the problem is that they hide behind all kinds of excuses in order not to provide their track record. And if they do, they talk in dollar terms which is meaningless - the only thing that matters is percentage gains.

Share this post


Link to post
Share on other sites
12 hours ago, Manish71 said:

Hi,

 

Approximately what kind of returns can you expect ? Lets say you allocate 30% of your $25k futures portfolio to commodities trading and the rest is in cash. Can you beat S&P return ?

 

Thanks.

 

 

I would suggest reading the AQR article in the thread JJapp linked.  Trend systems have a long history of outperforming standard benchmarks over the long run.

The point is this strategy is less about beating a particular index and more about exposure to a market with basically no correlation to equities with significant upside relative to risk.  That said, yes our system has done quite well relative to us equities during our short period of live trading and we will be posting these results.

 

Share this post


Link to post
Share on other sites
7 minutes ago, Kim said:

And the problem is that they hide behind all kinds of excuses in order not to provide their track record. And if they do, they talk in dollar terms which is meaningless - the only thing that matters is percentage gains.

Just on it's face .."we NEVER, EVER buy options under any conditions ever!" is SO dumb,and irresponsible.

I remember watching one of their extremely curve fitted studies showing that it was ok to allow yourself to sell naked straddles/strangles because the kind of "blow up" type of move only has occurred less than once a year going back 20 years.

Their point was to make people feel comfortable, and ok, with selling an open ended, non defined , position, because the ("their") statistics prove that it almost never happens.

Until it does (they left that part out)

Edited by cuegis

Share this post


Link to post
Share on other sites
6 hours ago, Manish71 said:

Pardon my ignorance or understanding, but why does risk increase when the trade trends favorably and risk decreases when it is in red ? Should it not be the other way around ?

you have to take this quote in context of what John was saying about the weekly nature of adjusting positions.  All he was saying is that we factor historical vol when deciding what our allocation to a particular option should be be in a new week.  The deltas of the trades will often change from week to week to fit our model portfolio risk profile.

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.