It's a shitty trade, but it's my shitty trade

I finally got my modeling tool to a satisfactory stopping point. Using sticky delta to model the IV surface.

SPY Diagonal call debit spread.
S = $439.75 on Oct 11 @ 11:00 ET.
long call: Dec 17 444 strike. 45 delta, IV 15.6%, $9.59 debit
short call: Oct 29 439 strike. 52 delta, IV 15.3%, $6.29 credit
Net: -7 delta, +0.3% IV, $3.66 debit

I am not sure how people stress-test these kinds of positions, but currently I am just modeling "reasonable" upside and downside moves in SPY. Not modeling outlier moves, since it's defined-risk.



The P&L looks fairly unbiased to downside or upside moves. If I close with 10 DTE, looks like ~$0.75 profit for the upside or downside case, about $1.00 for the flat case. 20-25% return on the $3.66 debit. That seems OK(?)

Net delta starts negative and stays there except in the downside case. I'm OK with mild net delta, since my retirement is very long SPY delta!

I appear to be slightly net long IV. I'm OK with that, since market IV seems low and apt to rise.

Net gamma is negative. I guess that's a good thing(?).

I'm still not even close to fully appreciating (in a practical sense) these greek risks, nor understanding what net greek positioning I should strive for (except delta).

Now comes the part where you tell me why this is a shitty trade.

 

So here's my take on this:

First of all, negative gamma is a bad thing, period. When market goes up, you become short and when market goes down you get long.

Second, I don't quite understand why you would consider buying vega in the long maturities and shorting vega in the short maturity when OCT trades below DEC in IV terms for a loss in roll down of 0.3% IV.

I guess it's the long calendar idea that is sold to retail in order to generate profits via decay.

Your risk is actually quite simple to gauge: You are in a short call spread (short 439 calls and long 444 calls) and you potentially make the most money when the OCT expiry is pinned at 439 and you get a rally right after...give that you don't manage the trade to be delta neutral.

In terms of software development, I'd build a risk matrix:
Unterlying price at the x - axis, implied volatility at the y - axis. Do this for delta, vega and gamma. Next thing you do is you normalize all your vegas to 30 days by using the following formula:

sqrt(30/days to expiration)*vega

You do this because you have more vega in the longer terms than you have in the shorter terms and when you look at your current position, you might get the idea that your position is long vega.
However, short term vega moves a lot more than long term vega and you have to account for this by normalizing your vega exposure. You might find out, that your long diagonal is actually short vega

By using the risk matrix (also called slide), you know where your risk is and which greeks become prominent in a certain scenario.

 

Awesome, thanks mucho for the homework and concrete feedback.

I guess I kind of look at negative gamma as a "hedge" (Boy, did I get spanked the last time I used that word!). Imagine that the underlying is oscillating in a short-term trading range about a longer-term trend. When the underlying rises to the top of the trading range, wouldn't I want my net delta to decrease? And vice versa.

Vaguea. That's my new name for vega. Not because vega is vague, but because my understanding of vaguea is pretty darn vague.

I need to better understand this vega normalization business. As it turns out, if you normalize the vegas in my trade, they are both 0.5 today. Net normalized vega is flat for 10 days or so then turns positive.

 

Here's a plot of net normalized vega vs net vega. This is for the declining SPY case. norm-vega-neutral until Oct 19, then turns upward.

 

But...
Do you know where the price is going?

 

no, but the trade is profitable for both an up or down move (see blue dotted lines on first graphic) of about +/- $8 on SPY.

 

No matter where the price goes, it will always win. Where is the catch?

 

It loses when the move is large, up or down

 

What happens if back month vols fall to say 10% ?

 

I build my IV surface by cut and pasting option chain data into a spreadsheet. Very advanced technology. To simulate your case, I would have to manually tweak the data. I will try it.

Is it statistically “common” for the back month vol to fall so much in a vacuum, e.g., with the front month vol staying constant? I honestly have no sense of historical bounds on vol skew, since I don’t have historical options data