Flirting with Models: Wayne Himelsein

Link: https://blog.thinknewfound.com/podcast/s2e7-wayne-himelsein/

About Wayne: CIO of Logica Capital

Transcription: Otter.ai


Overview

Every trade is implicitly long or short volatility or optionality

  • There is variability in every asset and its distribution dictates whether you are long or short.
  • Every trade is either a bet on convergence or divergence. Convergence trades are short volatility

Quant vs Discretionary

“There’s good and bad in all of it. So the best you can do for yourself by going with what you know because you’ll be able to ask better questions and be more comfortable with what’s happening day-to-day.”

  • Myth of quants building a black box then “going to the beach”

“The market is always changing. In fact, it’s funny even the idea of factors and categories, if you think of something like value and growth. These two big facets of the market, even those are evolving. [Consider] that you buy a value stock, and it turns around and starts moving in your favor. Well, now it’s a growth stock. So literally, the categories are changing on us. So if you bought a value book, and you leave it for six months, you’re now a growth book, if you were right on your picks.”

  • Using quant to “mechanize” what works vs mining for patterns

“Finance algorithms that developed from logic and experience that simply seek to mechanize what is already well understood, have a chance at success. Those that begin in data analysis, categorization, quantification, or statistical or numerical gymnastics do not.”

Opportunities in volatility trading

Traders have different “assumptions across the volatility surface, the strikes up and down and across the calendar upwards and outwards, There are different prices for every option. Because of all this modeling and people having demand for different options at different calendars in different strikes, there’s going to be cheaper and more expensive….Take advantage of the weirdness and pricing and model variants across the option surface.”

An inverse relationship between signal strength and opportunity size

  • As your signal strength declines you need to diversify more. “To have more probabilities repeated more often, [so] more positions”
  • Hoffstein: “Information ratio is equal to your information coefficient times square to breadth. If you have to lower your information coefficient, but your breadth goes way up, you can actually end up with higher information ratio”

Re-phrasing a bit: expectancy scales with number of trials but volatility scales with square root of number of trials. If your bankroll is large and your business diversified, it follows that your focus should be on hunting for high expectancy games, not minimizing risk.

Evaluating a strategy

  1. Use daily returns to get more data points. Monthly returns mask too much.
  2. Are you achieving your premise?

    “So you’ve said yourself, I know where I want to neutralize, and I know where I want to get my alpha. And if that’s where you get your alpha, you have to know that number one, you have alpha there. So if you look at your growth tilt and measure that against Fama growth factor, do you beat it? If not, you’ve got no edge.”

    • Map the strategy.
      • Compare the exposures to time series of different exposures to see how it behaves. This requires using mathematical tools that do not rely on linearity (ie regressions).
        • “I don’t ever listen to what [the manager] tells me. I just run it versus we have in here about 180 different exposures that we have time series for factors or exposures [to find out] “what is inside this thing?”
      • How intentional are the exposures?
        • Managers will tell you that they’re doing something but don’t even know what they’re exposed to. “Did you know you have a 30% exposure to momentum? Oh, no, I didn’t. I’m actually a value investor.” (Me: sounds similar to performance attribution frameworks behind “hedge fund replication” strategies)

Risk

Beta is a poor quantity to use to balance your portfolio

  • Beta equals correlation times vol ratio
    • It’s easy to compute which makes it popular
    • …but since its inputs are non-stationary, non-linear and themselves volatile it’s garbage in/garbage out.
    • Important to understand if a beta-hedge portfolio will bleed longer or shorter as correlation increases. (Me: This is why gross exposures are important to constrain)
  • How to balance a portfolio without relying on beta?
    • Geometric approaches that account for non-linearity
      • Clustering distance approaches
      • Stochastic dominance

Market neutrality is a “funny” concept

  • What does it mean to even be neutral?
    • “What do you want to be neutral to? Are you directionally neutral? Are you factor neutral? You can [initiate] a directionally neutral portfolio that has equal long shorts, with a complete growth, tilt, or a value tilt or some other factor tilt like a volatility tilt.

Overcrowding

“If we find a good pair trade, rest assured, many others have found it. And there’s just gobs of computing power, and PhDs and all the rest doing the same thing. And so we’re all going after the same edge. When things start to go wrong, the differences between the different groups is that they manage the risk differently. And one of the best means of managing risk in these markets [is to manage leverage]. The overcrowding risk is that everybody’s in this trade, and it’s a good trade. That’s why everybody’s in it. So you’ve done the right thing. But as some of these bigger shops start to unwind, it becomes everything going the wrong way. Others are needing to exit because they have LPs to answer to or they have risk that they’re managing to, so as long as you’re in it, you’re exposed to that. And it’s difficult to manage because at the get-go, you made the right bet.”

Walking away or sticking with a “broken” strategy?

Difficult question since the pricing may be more favorable as anomaly gets stretched but unclear whether the relationship will revert and on what timeline. There’s career risk is sticking with it vs the weight of the historical evidence for the opportunity.

“The more your measure won’t determine whether something’s out of favor, the more time you might give it to try to fix it”

“Comes down to a personal decision. How much time am I willing to spend tweaking and contorting to try to figure out whether I can fix it. And we all have our limits. It comes down to a business question as well. It’s not just tweaking and contorting and trying to fix it. But how much time can you spend defending it? How sticky is your capital? Even if it does come back still be in business?”

An easy example was the trade that shorted both the triple long and triple short ETFs on the same reference asset. The trade was over once the cost to borrow the shares exceeded the edge in the trade. This was easy to measure and therefore abandon when it became too crowded.

Hedging non-linearity or skew

  • “The only way to get rid of the left tail is to balance it with the right tail. And to have that obviously, you have to have the right offset temporarily. You need the time association to match that when this thing goes down, the other thing goes up. So you need to understand the time relationship between the two.”
    • Stop-losses are “synthetic left tail mitigator”. They are not fully reliable because of:
      1. Gaps
      2. Discipline
    • Tradeoffs between hit rate and cost of the hedge. Need to define what type of exposure you are ok with to target the right option hedge. Just like insurance has cost levers like premiums, coverage amounts, durations, and deductibles options portfolios can be custom tailored.
    • Flight to quality assets like gold, USD, treasuries in a permanent portfolio
    • Managers who engineer defensive market-neutral portfolios

Final words on hedging

  • Depending on the nature of the crisis hedges behave differently. Since we cannot predict the nature nor timing of a crisis it’s best to be diversified across hedges.
    • “Back to the larger insurance analogy, you have your medical and you have your dental and you have your vision. And so I don’t know where I’m going to get hurt. But either way it’s covered.”
  • Tolerating the cost
    • “Optionality being potentially the heaviest cost again, to me, it’s not expensive when you get what you want. But since it is more often a bleed than a payoff, perhaps people should have more treasures and gold and a little bit less optionality. But definitely all concurrently.”

Thought experiment

You can only own 1 asset and never trade it again, what do you pick?

SP500. The only reason people underperform the market is they want to control volatility and liquidity needs. But if we remove these concerns the best thing is to just own the market in perpetuity.

Flirting with Models: Benn Eifert

Link: https://blog.thinknewfound.com/podcast/s2e2-benn-eifert/

About Benn: Founder of QVR Advisors specializing in option-based strategies


Selected notes from his conversation with Corey Hoffstein, co-founder of quant management firm Newfound Research

Can you maybe explain the difference between what you would consider to be more of an option strategy versus what actual volatility investing is?

  • A common option strategy is call overwriting or put writing. They are both long equity exposures. That exposure is going to be the biggest risk factor. I would contrast that with volatility strategy which tries to isolate features of the distribution of returns, but not the direction of returns.

You mentioned that you guys focus somewhat heavily on relative value strategies in the volatility space. Can you explain what constitutes a relative value strategy? What’s a trade that you might put on?

  • Create value buying cheap exposures and selling expensive exposures at the same time, trying to hedge out the main directional market risks that would dominate a traditional asset allocation.
  • Identifying trades is really an important part of the process as they move around all the time over different cycle frequencies. Imagine, for example, long volatility in large-cap energy companies versus short volatility of smaller energy companies. That might be an opportunity at a point in time driven by a series of large transactions in the equity market. A large fund may have done a bunch of overwriting in their large-cap energy names, which suddenly made them very cheap. You really have to monitor, see the prices move and understand why there are dislocations and other relative value opportunities that might cycle over time.
  • Pension fund overwriting or cash-secured put selling are types of opportunities that might last for several years while those strategies are popular. You can imagine them becoming too popular over some number of years, then the pendulum swinging the other way. I wouldn’t say that there’s risk premia that you would expect in the space to just exist in perpetuity in a relative value sense.
  • There’s a large need to provide liquidity for end-users of options and distribute risk from where options are being heavily supplying to where they’re heavily demanded. These markets really developed on the back of end-user demand and their need to transfer risk. The key thing in relative value investing and in volatility, is that the marginal price setter for the probabilities and the market prices that prevail is not a volatility investor who is thinking about nuances in implied volatility. For a specialist volatility investor, many of the best opportunities really arise from either explicitly or implicitly providing liquidity to meet the needs of end-users, and to warehouse basis risk between what they’re buying and what they’re selling.

Analogs and Differences from traditional investing?

  • Selling vol and overwriting are expressions of carry styles
  • Rather than using traditional factor language to describe volatility trading he prefers a “Star Wars” analogy: Derivative users do things in big herds. And they typically have very large size relative to the absolute return community. Those flows are very sticky and implemented in similar ways with similar benchmarks, for example selling one-month index options. This creates congestion in one segment of the overall options market creating a ‘disturbance in the force’ — this creates really steep term structures, market makers get stuffed with short term options, they don’t have the risk limits to hold. And the relative value community’s job is really to distribute that risk much more broadly, throughout the ecosystem. A nice risk-reward profile is the payment to provide the liquidity to that market. Concentrated flows need someone on the other side to warehouse and distribute that basis risk.

How do you think about identifying trades in this space? How do you think about managing trades? How do you think about exiting trades? How does the book come together? It seems like a very overwhelming landscape to try to get your arms around.

  • Our investment process as a collection of bottoms-up strategy sleeves. So an individual strategy sleeve would really be a theme that’s driven by some particular type of dislocation or some particular type of underlying flow that end-users are generating. In a cross-sectional portfolio, opportunities are more fleeting, as opposed to being structural flows that are very consistent over long periods of time. It involves building out quite a lot of infrastructure, to identify those opportunities quickly.
  • In the example of a fund running a big overwrite sale on their long, large-cap, equity names portfolio, that would feed through quickly into the prices of options within that universe, and you’d see a significant reduction in those prices, relative to the prices of the small-cap energy names. You’d see it probably wasn’t driven by underlying realized volatility dynamics, it wasn’t that the spread compressed(because the names in the short baskets started becoming very volatile, and their prices started rising). You’d have various other ways of quantitatively triangulating that which would trigger an investigation into a type of the trade to add to the portfolio.
  • Where the dislocations are potentially more persistent, it might be more a question of measuring those dislocations. How do you track the ebb and flow over time? Is it a particularly attractive opportunity set? Do you want to have maximum risk on? Is it a less attractive opportunity set? Do you only want to have 30, or 40% of risk on? The identification of those type of opportunities is a starting point in the design of the strategies.
  • Again much is driven by what end-users of derivatives are doing in really big size and affecting markets. It’s not hard to see if you are an active market participant. You spend a lot of time talking to market makers and talking to the end-users of derivatives so you see it very quickly.

How do you think about the trade off between systematic versus discretionary and volatility investing?

  • I think in terms of a spectrum between, on the one hand, fully discretionary, and gut feel based investing all the way to the other end of the spectrum of fully automated back to front, systematic trading. Most volatility managers lie somewhere in between on that spectrum. It’s really hard to get that last mile to full automation. Since options are non-linear, you need to manage the very small risk of automation failures which also makes full automation elusive.

Are there any examples that come to mind where either an opportunity was systematically identified and you had a discretionary override? How about the opposite, where you thought there was an opportunity and the systems were not flagging it?

  • Back in the early days of Abenomics, in Japan, when the Nikkei was incredibly depressed, there was an interesting dynamic showing up in skew on Japanese equity indices. So skew is the relative price and an implied volatility sense of upside, call options versus downside put options. And in Japan, it actually started to go positive, which is very unusual. In other words, upside call options, were trading at a higher implied volatility than downside put options. A lot of folks in the volatility community got really excited about how silly it was, that an upside call option would trade at a higher implied probability than a downside put option, and really aggressively sold upside call options. But the key thing to remember back then was the Japanese equity market had just been incredibly depressed for a long time. There was a tremendous macro narrative building around big structural reforms and a great unconventional monetary policy. What followed was a very volatile rally! It was really a sucker’s trap to look at skew based on the historical data set because you were selling an upside crash scenario.
  • Another example was the model not appreciating how cheap the options on VIX were when the sizing in XIV became extreme and creating a very negatively convex profile in VIX due to the size of the rebalance. If you have a fund that requires a mechanical response that has to buy volatility when vol is up, it creates a problem if the size in the market became too large. It was just a market microstructure time bomb waiting to happen. The timing of that type of event happening was uncertain. But the sizes of those positions made it almost inevitable.

When you see a very steep VIX futures curve, in your opinion, is that an expression of the markets viewpoint? Or do you think that’s just an expression of a market imbalance?

  • Typically, it’s more related to risk premium than it is some kind of unbiased forecast of future volatility. If you look academic research or practitioner research there are some fundamentals to that term structure and some expectation element but often quite a lot of element of risk premium.

If you were doing due diligence, on a volatility strategy, describe red flags (besides leverage and are they selling tail insurance) and other concerns.

  • I would want to drill into sophisticated, top-down risk systems that stress all of the main risk factors in the portfolio to very extreme levels, and see that the risks were acceptable. There is no portfolio that makes money under all circumstances which is fine. But if there’s a major risk factor in the portfolio, you should be able to take it to a very extreme unprecedented level and see that the portfolio is not going to be getting liquidated at that level.
  • It should be contained in a level that’s acceptable to the end investor.
  • I’d want to understand the assumptions they’re making in those stress tests.
  • I would really want to see the actual positions and hear them explain what other parts of the market and what other market participants are doing to understand what the squeeze risk looks like.
  • I would want to see that they had at least contemplated thoughtfully and analytically how the strategy should be expected to perform going forward. And really a thought about how the market changed in the past 20 or 30 years versus right now. Markets in general change over time but volatility and options markets have changed dramatically.

Notes from EconTalk: Anja Shortland

Link: http://www.econtalk.org/anja-shortland-on-kidnap/

About Anja: Researcher and author of  Kidnap: Inside the Ransom Business


Economist Russ Roberts interviews Anja Shortland

Kidnapping for ransom as a business

The hint that kidnapping was in fact a business: 97% are resolved peacefully

How can the chance of a peaceful resolution be so high if all these things must go right:

  • Both sides must negotiate a price from a wide range
  • How to payment, typically unmarked cash, to the kidnapper?
  • Trust that the kidnapper will acknowledge payment
  • The kidnapper to trust they will not be arrested during the hand-off
  • The kidnapper must expect that the hostage will not be a witness

“The only reason for this kind of trade to go smoothly is what economists call the shadow of the future. So, people behave well this time ’round because it will help them in their business in future interactions.”

“This will only work if the kidnapper understands that he’s better off keeping the promises than breaking the promises. And that works because there must be a mechanism for information about good and bad behavior to be transmitted to future victims. So, if you have a kidnapping gang working in a city, then local gossip will probably ensure that people know whether or not they can trust the kidnappers. However, how does that work for transnational hostages? How does it work for the tourist that gets picked up in a bar late at night? How does that work for the aid-worker? How does that work for the expatriate?

Enter kidnap insurance

“There’s a very limited number of insurers, syndicates, underwrite kidnap-for-ransom, and they exchange information about trustworthy kidnappers and rogue kidnappers.”

  • Insurance actually ‘orders the market’, creating moral hazard in the process.
  • Corps buy ‘kidnap for ransom’ insurance with conditions:
    • Insured cannot know about it
    • Corporation provides security
  • In some areas, kidnapping occurs because corp didn’t know who to pay protection money to
  • Lloyd’s of London brokers a market of insurance companies willing to ensure special risks (like a basketball player’s knee)
    • The market settles into a civil equilibrium
    • Small supply. Crisis responders (often ex-special forces) retained by the insurer will have specific experience with a class of kidnapper
    • Insurers share info and more coordinated than the heterogenous kidnappers which keep prices down. However, when gov’t come in splashing the pot it changes the dynamics of the game as it raises the expectations of kidnappers b/c of public pressures and gov’t large resources and because unlike insurers they are in a one-off game (France hopes the next victim is Swiss)
  • Each kidnap market has local conventions
    • Example: Pirates want money dropped in canisters next to the ship so that kidnappers can stay high enough to avoid capture himself
    • Businesses that provide secure common ground for handoffs(almost like escrow!)
    • Trustworthy middlemen — again ‘shadow of the future’; reputations and long-running exchanges (reminds me of my open-outcry trading past. In the pits, your “word was your bond”)
    • While any one transaction can go wrong on average the market hovers around a going price.
    • If kidnappers make mistakes, then they are out of business.
      • “Sometimes you have very emotional kidnappers. Sometimes you have stupid kidnappers. But stupid kidnappers will reveal information. And ultimately it is in the insurer’s interest to eliminate stupid kidnappers–well, eliminate kidnappers where possible. But if you have stupid kidnappers who make mistakes, you can remove them from the market by dropping some hints to the police.”

On the game theory of negotiation

  • Manage kidnapper’s expectation of ransom size (hide the fact that the captive is insured)
  • “Squeezing the towel” process as the concessions offered to the kidnappers turn in to a slow drip
    • Eventually, the concessions are below the kidnappers’ cost to hold the victim. For example, the longer a hostage in custody the more expensive (via bribes) to keep it secret
  • Can’t reward kidnapper’s bad behavior or threats (“parenting lesson”)
  • Negotiators help the kidnappers see things through a more rational perspective. And, they educate them. And say, ‘Yes, we don’t want you to hurt Uncle Ted.’ And, ‘You’re not going to get anything out of hurting Uncle Ted.’ And they just help the kidnappers see how that strategy is not going to be helpful.

Notes from Alpha Exchange: Harley Bassman

Link: https://www.youtube.com/watch?v=X8wioRF0434&t=26s

About Harley: There is but one “Convexity Maven” in the world, a moniker that belongs uniquely to Harley Bassman. A 35-year career in financial markets has left Harley steeped in all things relating to the price of and characteristics of optionality.


Dean Curnutt of Macro Risk Advisors interviews “Convexity Maven” Harley Bassman

  • Is there too much short convexity out there?
    • Not in listed option markets where there’s a clearinghouse and vol is explicit traded and monitored
    • Risk is in the implicit convexity similar to portfolio insurance
  • Bassman on volatility surfaces
    • Term structure reflect flows; SPX has option sellers near term and insurance company buying in the longer term
    • Skew in bond markets has flipped since GFC. Pre-GFC puts were richer than calls as large asset managers hedged their bond exposures buying puts. Since GFC, the market recognizes that low interest rates are more coincident with financial stress which has re-priced the upside higher.
    • Forwards will typically price in line with long term options
    • Structured note issuance has vol-suppressing influence on surfaces
      • Europe has more structured note issuance b/c older more income-demanding demographic (looks more like covered calls)
      • Auto-callables in Asia suppresses downside vol (until roughly 10-15% knockout levels)
  • Bassman on a low interest rate worldWith central banks setting policy rates negative, the market is setting pricing across the curve very low.
    • Germany is -.20% out to 10 years yet have nominal positive growth and breakeven inflation is priced at 90 bps, so an extremely negative real interest rate out 10 years.

    Demographic motivated argument for secular stagnation

    • Negative short term rates are not unprecedented and typically accompany short-term market stress. Insurance premium to secure assets
    • Longer term negative rates are a symptom of market expectations for slower growth due to demographic headwinds.
      • In US boomers are getting older. Japan is further ahead and Europe behind Japan.
      • Declining labor force participation is biggest concern since growth = total hours worked x productivity
      • Labor force participation and yields are correlated over long periods
      • The trend of each decade is bluntly explained by demographics but it’s slow moving and difficult to trade
      • Immigration necessary to balance the ratio of workers to retirees. Immigration very important.
    • Trump is a symptom of low wage growth
      • Bassman believes QE1 was necessary to save economic system but later rounds of stimulus should have been fiscal not monetary. Monetary has caused asset inflation without wage growth. Inflation therefore was uneven and regressive leading to Trump and dissatisfied public
    • MMT
      • It’s coming. 2029 boomers will be fully retired and Republicans will not want to cut spending so there will be no check on Democrats
      • Japan a good example that MMT can work in the short term if you borrow in your own currency. The issue is that MMT will not be restrained even if inflation starts to emerge so is likely bad idea in grand scheme
      • The fallout can take decades but it’s not sustainable to print money at a faster rate than the economy grows

    Trade idea

    • Since bond vol term structure is flat, buy long dated (10 year) vol to hedge against longer term seismic shift while levering coupons on CEFs, MLPs, REITs and/or sell puts in 1 to 3 years bond options since demographics will limit rate upside to 3-4%. Can lever the near dated trades while owning the vol protection. This is a version of long time spread since near-dated levering or outright option selling is all short vol.
    • Outright tail protection too expensive and path dependent to be relied upon