Any traders in here who specialize in volatility skew and probabilities-based strategies?

D

daytonabeach83

Hi,

I'm new to the forum, but I've been exploring the world of trading volatility skew for a short while now. This led me down the road of the Kelly criterion and traders such as Ed Thorpe, Blair Hull, Nassim Taleb, and William Ziemba. I started out trading modified vertical spreads in equities displaying positive (call) volatility skew. I discovered you could set up bullish defined risk positions with positive expected values, versus the neutral expected values one would expect with put-call parity. Originally, I discovered this through the use of hard collars, but then began modifying the collars. This eventually led me full circle back to vertical spreads.

However, in recent months, as the majority of equities displaying positive volatility skew have been on a relentless downtrend, I began to explore some of the more nuanced aspects of this trading style (nothing like a huge loss to spur a huge growth in learning), including the increased tail risk and violent portfolio swings that occur as a result when using the positive EV trading style and Kelly criterion. This led me to more advanced probabilistic calculations, where instead of just looking at overall projected P/L over time, I began looking at secondary and tertiary aspects of the strategy:

- risk of max loss vs risk of max profit (independent of net positive EV) and total area under the curve

- using kelly to modulate position sizing post-entry to directly reshape your portfolio's volatility curve against market direction (i.e., increasing positive skew on up moves and neutralizing it on down moves)

- skew trading as a mean reversion strategy

- employing synthetic ratio spreads to turn tail risk zones into potential profit zones

- ergodicity and non-ergodicity; specifically, the use of diagonalized spreads to create 3D probability curves by the addition of a time factor to avoid absorption risk

- using strategies such as synthetic put ladder spreads hedged against long stock position to increase your portfolio's positive volatility skew in equities with negative volatility skew

- basing strike selection on the specific shape of an expiration's volatility curve instead of delta exposure

- the concept of account pooling to increase exposure to number of occurrences and allow traders within the pool to increase individual position sizes accordingly

I have a ton of questions, as I still feel I'm in the learning phase with all this. I'm hoping I can find someone experienced in volatility trading, but more specifically, in trading volatility skew both in individual stocks and index ETFs. I was never a math guy and I've found myself trying to learn calculus now, lol.

I have a lot of ideas about how to mitigate the increased risk of ruin that occurs when shifting your account's volatility curve to the positive side, and ways to both mitigate and profit off that tail risk. But I have a lot of questions about position sizing per Kelly, things I "think" I'm understanding but might not be:

- If I understood Blair Hull correctly, Kelly states you should increase position sizing on green days and decrease it on red days, equivalent to your portfolio P/L? So, if my account is up 7% I should be placing 7% more bets, to keep my total exposure rising? And if I'm down 7% I should be downsizing positions by 7%? What I extrapolated from this when plotting out volatility curves is you are creating an exponentially rising growth curve as you expand positions with profits, but by contracting positions with losses you create an exponentially flattening descent curve? If you put on more and more bets on the way up your trajectory goes sky high, but by taking them off as you lose you gradually have less and less exposure so losses become smaller and the curve flattens out? Essentially, you create an account that crashes up and grinds down (positive volatility skew), versus the typical grind up and crash down (negative volatility skew)?

- Looking at Dr. Taleb's stuff on how positive skewed volatility curves increase tail risk, I think decreasing position size as things decline means you are basically "unskewing" your portfolio's volatility curve (less exposure to skew and more exposure to cash = your overall portfolio's volatility curve neutralizing), and when you fully "unskew" the curve the tail risk is no longer increased? So, essentially, as you start to move towards the tail risk zone you bring your portfolio's volatility curve back to neutral so as to negate the increased tail risk before you actually reach it? Am I understanding this correctly?

- And does this make ratio spreads (or synthetics with the same risk profile curve) the best strategy for trading volatility skew? As Dr. Taleb puts it, if the volatility skew is high that means most of the big influential moves are happening in the tail risk zones, which seems to imply that if you're trading skew the areas outside 1SD are the information you need and the stuff within 1SD is the "noise?" My thought is that a ratio spread risk profile basically turns both tail risk zones of the probability curve into potential profit zones. They seem to a) maximize the probability range in which you see a profit, even if one side is a super low probability, and b) turn the one weakness of probabilities-based Kelly trading - the tail risk - into a potential asset. Am I missing something here?

- Kelly makes sense if you're trading shares, but all the positive EV-probabilities trading stuff I've seen is centered around mispricings in options contracts due to volatility skew. If you're large enough to be trading 100 lots then it makes total sense cuz you can easily just throw on a few more contracts or take off a few as needed. But what if you're in a smaller account? Trading super narrow spreads makes everything very binary and increases the number of occurrences needed to see the probabilities play out. But while running wider spreads (like a 90/35 delta vertical) mean less occurrences are need over time to see the same probabilistic outcome they cost more money. My question here is how do you apply Kelly management tactics in an account where you're only running 1-2 spreads per bet and you can't easily contract or expand the position? Do you just take off individual trades, or roll down your shorts to effectively shrink the bet size? And if so, where do you draw the line? Intuitively, given the volatility inherent to Kelly trading strategies, it seems really easy to overmanage positions - it's down 5% today, but it's up 7% tomorrow, so how do you "remove the noise" from your management strategy? Do you need to calculate the expected move per the equity's IV and only manage if it nears that? Or perhaps just manage trades when they hit the 1SD mark to either side? I'm so lost here, lol.

I have so many more questions too, but I've already dropped way too long of a post as is. I'm really just hoping to find someone I can bounce some ideas off and discuss the things I know, the things I don't know, the things I have wrong, where I'm right, and where to go to further develop my understanding of these types of trading systems.

Thanks in advance. Sorry for being so long-winded! Peace.
 
Last edited:
On the risk management front, strategies like Kelly criterion are really designed to ensure that you are taking the optimal level of risk while still ensuring the ability to survive a series of losses. A good rule of thumb is to max your cost limit by the expected volatility of your positions x your drawdown threshold.

E.g. max you can afford to lose on an annualized basis is 20%, if each trade you make has an expected volatility of 50%, then it means that the max position size should be below 35%.

You can rebalance this on a daily basis. For example, say you hold a position at 35% on day 1. On day 2, the trades goes down by 10%, and now only represents a 32% weight in your portfolio. You can rebalance to increase the weight back up to 35%. You do this by adding to the position.
 
longandshort said:
On the risk management front, strategies like Kelly criterion are really designed to ensure that you are taking the optimal level of risk while still ensuring the ability to survive a series of losses. A good rule of thumb is to max your cost limit by the expected volatility of your positions x your drawdown threshold.

E.g. max you can afford to lose on an annualized basis is 20%, if each trade you make has an expected volatility of 50%, then it means that the max position size should be below 35%.

You can rebalance this on a daily basis. For example, say you hold a position at 35% on day 1. On day 2, the trades goes down by 10%, and now only represents a 32% weight in your portfolio. You can rebalance to increase the weight back up to 35%. You do this by adding to the position.
More...
thanks for the tip about vol weighting.. that's been one of the things i kind of intuitively have been leaning towards, the idea that position sizing needs to be vol weighted, and that trades should be managed once they hit whatever threshold i've set.. thank you for that!

to the second part, that's kind of where i've been at, but i've been trying to figure out how to do it the right way.. using a daily rebalance with a set P/L threshold is right in line with my thoughts..

but say my overall portfolio threshold for management is +/- 7%, then if i'm up 7% on the day i'm going to a) close out enough positions to actually realize that 7% gain, and b) expand my overall cash deployment by 7%.. am i on the right track there?

i considered just expanding the position sizes, but if you aren't realizing any of those profits, i'm not sure it pans out like i'm hoping?

and +1 at the purpose of kelly, seeing how the mere act of creating a positively skewed volatility curve in your portfolio increases your exposure to tail risk was a HUGE eye opener - explained why my account has been so volatile, and also what the big risk was in the whole positive-expected-value trading system!

am i conceptualizing that right tho in regard to how kelly mgmt affects your portfolio vol curve? the continued position expansion on profits creates an exponential growth cruve, while continued position contraction on losses creates an ever-flattening curve? and since probabilities tend to hit in waves, it actually plays out in practice that way?
 
You want to be consistent on both the long and short side. There’s a difference between adding to a position based upon conviction (e.g. starting at 1% and moving up to, say, 5%) versus drifting (a position goes from 1% to 10% from gains).

In poker terms letting positions drift in weight is like upping the ante on a hand that was good preflop, without incorporation new information revealed post flop.
 
Also— important distinction to make between position pnl, position weight, and portfolio pnl.

If you have a single position and your weight is 100%, if it moves up 10%, the position is still at a 100% weighting.

So if you have a position of 2%, with a max weight of 5%, and it goes up 100% overnight, you will trim the position down to your threshold the next day.

Risk management is all about sizing your bet — making sure that your size stays within your risk limits and that it is aligned to your conviction.
 
that definitely makes sense.. it's a little tricky figuring out how to manage spreads that are doing the opposite of my overall account - if the portfolio is green, does management start with winning or losing trades? or if it's red, same thing.. still wrapping my brain around all that!

what's getting me most is the best way to go about resizing the positions if i'm not running 100 lot trades.. given i trade a lot of 90/40 call debit spreads on equities with call skew, my best guess for trimming position sizing on a winner is to add ATM call credit spreads if the stock breaches my short strike.. selling ATM will reduce the effects of the call skew working against the bear spreads vs OTM spreads, while ensuring i only lose on the bear spreads if i win on the bull spread.. and if the stock drops back down, now the bear spreads are hedging the down turn..

theoretically, if the initial spread has a 1:2 risk-reward ratio, and the stock breaches the short strike, you could sell enough ATM credit spreads to completely cover the initial cost of the trade and essentially convert the trade into a riskless broken wing butterfly... since you entered on a 1:2 risk-reward, and the bear spreads ATM will yield about $0.50 on the $1.00, you can sell enough credit spreads to cover the initial trade cost while still ensuring you'll see profit to the upside despite removing any downside risk...

>pay 4.00 for a 10.00 debit spread (not 1:2, i know lol)
>short strike gets breached
>sell 8 1.00 call credit spreads for $0.50 each at your short strike
>trade is now a riskless butterfly with a potential 2.00 profit

i go with narrow bear credit spreads there because if you widen the bear spreads you start going OTM and the skew starts to really fight against you.. if you're removing the risk, you don't have to worry about having to wait till expiry for volatility to run out on the spreads, so having to hold narrow spreads longer doesn't matter.. and the binary effect, again, is irrelevant, cuz you've got no money on the table..

managing losers seems simpler - just take off trades that are losing, OR (if you're running super wide verticals like i am), roll the short call down to your breakeven strike to downsize the risk exposure...
 
Last edited:
Good points. At the end of the day, lining up conviction with size is what’s key. If you’re playing with butterflies, you are either expressing a view on the direction of the underlying or its volatility. That’s the driver of your conviction. If a trade moves against you, is your thesis still solid or not? If your thesis changes then exit, but there’s a difference between a trade that’s down versus being a loser. Bet sizing allows you to expect a total loss on the trade (which does happen when trading options).
 
You must log in or register to reply here.
Back
Top