Intelligent swing trading algorithm

[digitalnomad]

Ok, so I’ve engineered a model that swing trades around ten sector ETF’s, with statistically significant sample size trading both sides of the market, all with good liquidity, 20-30% annual returns.

I can automate all the signals successfully, but the exposure and correlation risk is a big deal breaker. Looking at the historical data on the underlying ATM options shows good liquidity and profit potential.

I’m considering investing time and money into the development of an intelligent algo/app that can use my signals as a basis for buying long ATM (back month) options that can get the deltas and quantities inline with the signals. Vega is of no concern, as the position needs be entered regardless. The algo is also intended to exit the option position based on its underlying stop signal, so that means a CBOE market order (bend over baby).

As an example, let’s say my signal generates a BUY on 1000 shares of QQQ. The algo would then hunt down the ideal ATM strike according to underlying price (let’s say .75 delta), and then buy 15 contracts at that price to equalize the deltas to 1.00. I do realize that I can buy a DITM option at 1.00 delta paying less commissions/slippage, but liquidity is of the essence here (god damn derivatives).

So, what do you fellows think? I need some option gurus to tell me I’m living a pipe dream, before I throw some money in the wind. Thank you in advance

 

[Option_Attack]

You can definitely trade long options on ETFs with swing systems. That's all I trade. What I mostly look at is what are the average days in the trade and average percent gain for each system, with each ETF. I only look to be in a trade at most 20 market days. Just too impatient for months-long swing trades.

I also don't care much about delta. 50-60 is fine for me. ATM or a strike ITM works best for me. . Gamma will take care of the rest. Liquidity/spreads and IV are much more important to me. I try to balance the option IV with the expected move. I like to make at least 50%. For example for a next month option, if the IV is 25 ish then about a 2.5 - 3% stock move will make about 50%. And 5-6% move will make about 100%, if I get the stock move in a week or two (my "normal" trade time). Higher IV of course means I need a larger move in the ETF.

Anyway, sounds like your system should work to me. Paper trade it, or better yet, try it with very small positions until you are sure it is really hitting. This time of year kind of scares me, especially after a big runup.

 

[TheBigShort]

I’m living a pipe dream, before I throw some money in the wind.

Is this solely for personal/firm use or is it going to be available to the general public? If the former, why does it differ from any other strategy implementation? You have an idea, back tested it, made sure it fit your risk/liquidity parameters and now are ready to deploy. If the latter, as long as you believe in the product and can make people see the value in it, fire it up. That being said if a lot of people start using your signals, you will either have to create a new strategy/software or become a better marketer.

 

[digitalnomad]

You have an idea, back tested it, made sure it fit your risk/liquidity parameters and now are ready to deploy.

It's for personal use. Not sure if you read my original post thoroughly, but the ATM option trades need to be automated based on underlying signals. Not easy to deploy. The algo needs to find optimal (liquidity, spreads) option strike prices as signal is received.

 

[tommcginnis]

I can automate all the signals successfully, but the exposure and correlation risk is a big deal breaker.

Options aren't your only choice.{Isn't that a neat sentence? I know -- "It's late." But WTH. }

While your system probably treats exit conditions with more 'umph' than entry conditions, consider those exit conditions to also be an entry for a 2x or 3x short ETF in the same field, only holding for 26-48 hours. Just as something to "take the sting out" of your primary's (temporary) fail. Small, specifically short term, but with SL and PT brackets, too.

 

[digitalnomad]

To reduce exposure, there is no other choice. I either pony up 500k to swing trade the 10 instruments, or put up 50k to trade their ATM options. Even if I gave up 1/2 the ROI the underlying offers, i'd still be happy. The uncertainty with the options lies in entry and exit slippage. It's very hard to quantify.

 

[fan27]

To reduce exposure, there is no other choice. I either pony up 500k to swing trade the 10 instruments, or put up 50k to trade their ATM options. Even if I gave up 1/2 the ROI the underlying offers, i'd still be happy. The uncertainty with the options lies in entry and exit slippage. It's very hard to quantify.

Do you have access to historical options data? If so, I don't think it would be too much effort to create a prototype to test your idea. Really, all you need are backtested trades from the underlying and then line up the entry datetime with a corresponding option datetime and have some logic built in to determine which option to buy. And then you would exit the option trade when you would have exited the underlying trade.

 

[fordewind]

The uncertainty with the options lies in entry and exit slippage. It's very hard to quantify.

spray the limit grid.

 

[lindq]

Ok, so I’ve engineered a model that swing trades around ten sector ETF’s, with statistically significant sample size trading both sides of the market, all with good liquidity, 20-30% annual returns.

I can automate all the signals successfully, but the exposure and correlation risk is a big deal breaker. Looking at the historical data on the underlying ATM options shows good liquidity and profit potential.

I’m considering investing time and money into the development of an intelligent algo/app that can use my signals as a basis for buying long ATM (back month) options that can get the deltas and quantities inline with the signals. Vega is of no concern, as the position needs be entered regardless. The algo is also intended to exit the option position based on its underlying stop signal, so that means a CBOE market order (bend over baby).

As an example, let’s say my signal generates a BUY on 1000 shares of QQQ. The algo would then hunt down the ideal ATM strike according to underlying price (let’s say .75 delta), and then buy 15 contracts at that price to equalize the deltas to 1.00. I do realize that I can buy a DITM option at 1.00 delta paying less commissions/slippage, but liquidity is of the essence here (god damn derivatives).

So, what do you fellows think? I need some option gurus to tell me I’m living a pipe dream, before I throw some money in the wind. Thank you in advance

When you strip away the technical challenges, you are left here with a basic strategy of long options on ETFs, entered at the market.

That's a very questionable strategy, IMO.

Before you sink thousands of hours into coding, you should first be certain that your backtesting includes realistic expectations of spreads and decay. And the best way to determine what you'll be facing is to place a few trades, even with a single contract, and track them carefully.

 

[digitalnomad]

When you strip away the technical challenges, you are left here with a basic strategy of long options on ETFs, entered at the market. That's a very questionable strategy, IMO.

Pretty much my assumption too. There is no getting cute with CBOE limit orders in fast moving markets. The logic will have to revolve around market orders. Thanks for that assessment.

Before you sink thousands of hours into coding, you should first be certain that your backtesting includes realistic expectations of spreads and decay. And the best way to determine what you'll be facing is to place a few trades, even with a single contract, and track them carefully.

We're definitely on the same page with that