Short Bear Call Spread Entry Criteria: General Options Strategy & VIX Term Structure

An important theme in options selling strategy is to buy intrinsic value and sell extrinsic value. I’ve been thinking of how this relates to the entry points for short Bear call spreads in general and my current strategy as applied to the underlying $VXX in particular.
(Note: In this particular example for simplicity, I’m assuming extrinsic value is only composed of time value, even though extrinsic value is also composed of implied volatility and interest rates. I’m assuming negligible volatility skew between strike prices & interest rates are negligible to calculation.)

The question is which is better?
– Larger credit from a (deeper) in the money Bear credit spread?
– Smaller credit from a (further) out of the money Bear credit spread?

Assuming that the underlying is currently trading at S, there are four base cases for short Bear call spreads of Y/Z where Y and Z are the short and long call strike prices, respectively.

1.  In the money: Y < Z < S
The most premium, lowest probability of success, and highest risk.

2. Underlying at Long Strike Price: Y < Z = S
The underlying is trading at the long call strike price.
– Long more Time Value than Delta
– Short more Delta than Time Value

3. Underlying at Short Strike Price: S = Y < Z
The underlying is trading at the short call strike price.
– Short more Time Value than Delta
– Long more Delta than Time Value

4. Out of the money S < Y < Z
The least premium, highest probability of success, and lowest risk.

So to answer the question, it would appear that to match the overall options selling philosophy of short time value and long intrinsic value, case #3 is optimal, where call spreads are shorted right at the strike price of the lower short call. More time value is shorted and more Delta is bought.

VIX Term Structure
This is the default of where I’ll be choosing my entry points with Contango less than 5%, with the further thoughts and modifications for larger values of Contango:
– the higher the contango in the VIX futures term structure
– the more likely that $VXX will fall and fall faster in price
– the more absolute profit potential outweighs considerations between intrinsic value (delta) and extrinsic value (time value)
– the more the trade should be considered an outright short of the underlying realized through options, rather than trading on the inefficiencies of options realized through the underlying

Roadmap: Portfolio Growth & Automation

Roadmap: Portfolio growth and development of trading program interfacing with Interactive Brokers API

0. High Level
A. Timeframe
– Current: Maintain minimum short 100x call spreads ($50k Reg T margin)
– End of Year: Maintain minimum short 200 call spreads ($100k Reg T margin)
– 18 Months: Maintain 500-1k call spreads ($250k-$500k Portfolio margin)
B. Highest priority is assistance in trade execution to automate submission of numerous limit orders for entry and exit, incrementally adjusting near the midpoint towards bid or ask.
C. Scanning is to start research component & descriptive high-level feel for numbers to augment manual selection, which in turn is aided by automated limit order placement.
– Automatic entry and exit based on scans running on cloud server based for later iterations.
D. Brainstorm how to incorporate Machine Learning / Deep Learning to optimize even more

1. Trade Execution: Scan Bid-Ask for Entry & Exit
Bid < (midpoint – X) < midpoint < (midpoint + X) < Ask
A. If Ask is desirable, start at (midpoint – X), then move up towards ask in 1 cent increments per time t
B. If Bid is desirable, start at (midpoint + X), then move down towards Bid in 1 cent increments per time t
C. Time: t = 1 to 5 seconds, X = $0.10
D. “Desirable” depends on combination of entry or exit and definition of Bid vs. Ask
E. Enhancements
– Scan different exchanges: CBOE, AMEX, BOX, NYSE/ARCA, etc.
– Factor in exchange fees & cost savings via liquidity rebate
F. Update: Looks like IB has this and other algos available through their Trader Workstation Software desktop and through their API, so the task is to leverage the pre-built libraries as much as possible

2. Scan $VXX Options Chain: General
A. For Expiration Months: 2 – 6
B. Download options daily IV and calculate IV rank
– IV Rank = (IV – lo)/(hi – lo)
– Consider removing outliers and/or N highest or lowest (N=1-3)
C. Open interest
D. Volume ATM
E. Bid-Ask spread ATM
F. Volatility skew
– Horizontal: Scan for any outsized skews horizontally across expiration months to focus on ideal expiration 2-6 months out
– Vertical: Scan for any large skews vertically to tweak entry for specific month
G. Enhancement
– Scan other liquid underlyings
– Research the long $SPX cheap puts trade used by Ibex Investors & Empirca Capital
– Calculate IV percentile (by trading day)

Important Principles of Options Trading

The following points are a quick reminder of the important issues with respect to options selling.

1. Net Selling

You always want to be selling options when they’re overpriced due to the high implied volatility priced into the market-traded options.

2. Capital Efficiency

You win not simply from high implied volatility rank reverting back down but also the rapidity in which it does so. In particular, if you hit your profit target in 2-3 weeks instead of 4 months, you can then close out your position, start another one, and recycle capital faster and therefore be more capital efficient.

3. Theta Decay
Sellers benefit from Theta time decay so you’re always winning even if nothing happens.

4. Movement of Underlying
You can have the underlying move in your intended direction, stay the same, or heck even slightly against you to win, as opposed to requiring it move strongly in your favor.

5. Spreads vs. Naked
It’s usually better to go with spreads unless you have a super high conviction and high probability trade, which you might have in this case. Usually want to enter into spreads since you have some offsetting position in all cases and aren’t exposed to nor dependent on a huge move in the underlying.

Profit Target Considerations

Profit target assessment for short options spreads strategy
* Standard Profit Target: PT = ~50%
* Higher Profit Target: PT ~75%

A. Profit curve
1. smoother for Standard PT due to less variance
2. better reward/risk ratio due to lower downside risk

B. Total return on capital
1. greater for Higher PT
2. better reward/risk ratio due to higher upside expectation

C. Prefer Higher PT with greater margin of safety from
1. greater distance from short strike
2. greater days to expiration
3. smaller width between short and long strikes
4. higher IV premium from higher IV rank upon entry
5. higher contango
6. higher delta1 = M1 – Spot
7. higher delta2 = M2 – Spot

D. Psychological pressure lower for Standard PT, so prefer Higher PT
1. as specific spread is lower portion of total account
2. as account size and experience grows
3. as quantity and diversity of spreads increas
4. as winning momentum grows (purely psychological, not mathematical)

Calculation: Risk and Trade Management

Calculation to provide some precision and numbers to various risk and reward aspects of my strategy to hold my views accountable to specific refutation and/or discussion. Let’s take a typical example with numbers tied to a specific spread I entered. Of course, all these specific numbers can change, but I’m tying this to specific trades I entered that were posted publicly on my feed.

0. Preliminaries
– Spreads: Short 50x $VXX Call spreads of $A/$B, where $5 = $B – $A
($5 strike width is completely arbitrary, will tweak as function of underlying price when I scale up to $100s of thousands and need to be more precise. At the moment, there’s no sense in going crazy to imbue this calculation or the trade with excessive precision, that’s usually “false: precision.)
– Margin Collateral: $25k
($25k = $5 spread width/spread x 100 shares multiplier per spread x 50 spreads total )
– Premium: $7k
(This premium has gone up all the way to $7.6k on spikes like my previous 50/55 spread, but let’s round down to something more attainable and realistic.)
– Max Risk: $18k = $25k – $7k
(Credit from premium serves to reduce risk)

1. Assuming profit target of 50%, each completed cycle brings nets $3.5k
2. Number of cycles completed to equal max risk
$18k / $3.5k = 5.14
Let’s round up to 6 cycles to account for transaction costs and add a bit of conservatism.

3. So that means that after 6 cycles occur where I don’t experience a max $18k loss, then I start to win on an expectation basis.
Very loosely, if I don’t suffer a max loss on a spread 1 in 7 times, or about 15%, I should be good. Since each spread takes 1-2 months to mature, this is 7-14 months. Can’t really go more precise than this, just some round estimates.

4. This is a first order calculation based on real numbers but with some contrived restrictions, such as assuming no trade management and assuming helplessly sitting to expiration if the underlying goes beyond my far strike.
- In reality, there are many ways to manage the trade, the best way is to roll the trade to the next N months. Obviously rolling incurs a cost for the specific roll, but is mitigated by additional premium credited that mitigates the costs of roll, and furthermore buys additional time for the trade to move favorably with the Bearish bias of $VXX.
– In reality, the reason for a large DTE is not to sit doing absolutely nothing if the trade goes against me, but to provide a positional/implicit risk managment in that we know $VXX is going to have a downward bias.
– In reality, I plan to close out at 50% profit target with a large amount of time before expiration, maybe 2 months on a 4 month DTE option. I realize I’m leaving a bit on the table here, but safety is my number one concern.
– In reality, max loss before expiration is only approached asymptotically since option value is composed of both an intrinsic component (difference in strike price and current underlying price) and extrinsic components (implied premium and time value).
– Much more, but this is where trading skill in practice and reality starts to matter more than theory or an infinite number of hypothetical counterfactuals. Being too afraid and paralyzed by potential risks is just as bad as being total too ignorant or complacent towards risk.

Calculation: Short Options vs. Short Underlying

Quick calculation: short call spreads vs. direct shorting for $VXX on drop from $40 to $32 based on recent options numbers

– Short 50x 40/45 bull call spread
$25k collateral, $7k premium, $18k max risk
Decay 50% in 2 months
$3.5k profit if $VXX drops from $40 to $32

– Outright short w/ 40% of $25k assuming 100% margin
Initial position of $10k = 250 $VXX
$250 per $1 $VXX decline
Max risk = infinite, effective max risk = $25k
$2k profit if $VXX drops from $40 to $32

– Outright short w/ 20% of $25k
Initial position of $5k = 125 $VXX
$125 per $1 $VXX decline
Max risk = infinite, effective max risk = $25k
$1k profit if $VXX drops from $40 to $32

– main 2nd order adjustments and limitations of model
1. time limit for short options vs. no time for limit short underlying
2. commissions for options vs short borrowing costs
3. for unchanged underlying, short options may gain value from theta decay if below strike vs. vice versa if above strike
4. calculation a function of underlying price ($10k gets 250 $VXX shares at $40/share vs 1k $VXX shares at $10/share, but risk profile changes due to spikes)
5. wider bid-ask spreads for options vs super liquid underlying
6. uncertainty: early option exercise vs. early recall of short shares
7. uncertainty: broker margin requirements

Calculation: Bid-Ask Spread in Entry and Exit

Quick calc of implications of 1 cent in option spread, single or multiple legs.

1 option = $1
100 options = $100
1,000 options = $1,000

Entry and exit requires 2 trades

1 option = $2
100 options = $200
1,000 options = $2,000

$VXX options spreads can easily be $0.20 wide, meaning that the difference between midpoint and bid or ask is $0.10. (This could easily 2x or more)

This means that for a roundtrip 1,000 options, it can set you back $20k = $2k * 10 simply from hitting the midpoint vs. hitting the bid or the ask. Very significant to get the bid-ask spread entry and exit process down correctly if your strategy depends on large volume. Imagine 10k options contracts? That’s $200k difference in returns! Over many years and 100k contracts, that’s $2M.

Implicit Assumptions of Predictive Models

What makes a good trading model in particular and models in general? Huge field of study in the philosophy of science with such words like epistemology, ontology, supervenience, counterfactual, ontic, lots of words with implications of $0.00, but I wanted to consider it in a very defined context to make money in trading or machine learning.

It’s interesting to consider the validity of models derived from empirical data and used to predict and act, in particular for trading the markets, and in general to minimize a cost function in supervised and/or maximize a utility function in supervised Machine Learning. There are three implicit assumptions that are usually never even considered.

Assumptions:

1. Theoretical Sufficiency 
There is sufficient data to create “any” model that can use empirical regularities in the past to predict and act on the future, at any level of precision, especially at a level of precision that can overcome trading costs and statistical noise.

2. Practical Sufficiency
Given the possibility of theoretical sufficiency, the assumption that the specific model you build or trained is good enough even if there is a theoretical mapping from past to future

3. Stationarity of Probability Distribution
The assumption that the probability distributions of event in the past will continue in the future may be incorrect, assuming #1 and #2 perfectly are satisfied.

I see traders use Technical Analysis to predict security movements and of course it’s sad to see them use it without much critical thinking to the range of validity of models, most especially in markets that may be illiquid, opaque, with high transaction costs, high information asymmetries, or even outright manipulation. The discussion above complements the entire discussion concerning backwards-looking, descriptive statistics vs. forward-looking, predictive probabilities, as well as pure data-driven empirical models vs. data-informed analytical models.