Frustrated with derivatives pricing. Where else can I use my pricing skills?

Joined
6/12/17
Messages
17
Points
13
Hi folks, I work in GS on pricing models for exotic equities and rates. I am frustrated with just working on pricing models. Although I have deep expertise in this area (understanding very complex math, numerical methods like Monte Carlo, programming skills, etc) , I wanted some help in understanding where can I go now, with following specific questions:

1. Is there any overlap with the skills required in quantitative trading/investing? I have a very vague idea of what they do. In HFT, they mostly do smart linear regression and nothing fancy and they dont care about pricing logic a lot, once they have a codebase for it. For mid-freq maybe they do more complicated predictive modeling. Happy to be corrected if my understanding is wrong. Would love to know if and how there's an overlap between pricing quants and the work done by signal/alpha generation folks.

2. Are pricing quants compartmentalized by buyside quant firms, as quants that aren't relevant for them, or they are ready to consider them in case they show some work which has overlap with required skills? I'm talking specifically for signal/alpha generation roles, and not for derivatives modeling roles within buyside.

3. Are there any areas where you need to know BOTH - pricing models and signal generation? From what I've been reading, volatility trading seems to be that area, but I am not sure and I'm probably not looking at the right sources. Would highly appreciate if you could share a bit about areas where both these skills are needed.

Thanks a ton! Also if anyone has questions about the pricing world, I'd be happy to help in any way I can. Feel free!
 
Hi - I am a quant at an options trading firm. The below is based on my experience at my current firm and might not apply to others. My team works roughly on three topics: pricing, market making algorithms and signal-based volatility trading. A solid understanding of pricing theory is useful in all three areas.

As opposed to banks, we don’t trade exotics and thus typically don’t deal with very complex underlying dynamics (with a few exceptions). Our models are highly specialized for just vanillas. They have to be extremely fast and we require numerically very stable Greeks. We try to incorporate as many market nuances as possible directly into the models - unfortunately I cannot be more explicit here. Coming from a front-office / desk quant / strats job would make you a good candidate for this. If you’re working in a back office role (risk, model validation, …) then this would be a much tougher sell.

The market making algos involve things like fitters and retreaters. They are often a bit less model-driven / ad-hoc but try to stay parsimonious and intuitive in their parametrization since traders interact with them. Having a good grasp of pricing (e.g. no arbitrage conditions) is very useful here but not strictly necessary.

The signal based trading part is indeed the furthest away from pricing and requires a strong statistics foundation. It’s more than just “smart linear regression” as you phrase it. The better strategies typically involve a good understanding of market dynamics and drivers to build relevant features. Getting “clean” input signals heavily relies on the underlying pricers normalizing away as many effects as possible (dividends, jump events, early exercise, default risk, …) that would otherwise manifest in the volatility surface.
 
Last edited:
@CrossGamma whats wrong with model validation if they have the same skills? (and risk equally)
Strats are typically the ones building the models and writing the production code used for trading while model validation reviews the existing implementation and maybe builds their own version of it to check against. The desirable skillset/experience on the buy side is the former. Strat roles are also much closer to the trading desks / the revenue generation, which means they typically have a better understanding of markets and strategies. Their work is more fast-paced and involves more ad-hoc analysis and problems beyond just pricing models which require a certain degree of pragmatism. All of this is much closer to what you find at proprietary trading firms. As you see from my first post, only a small part of my role is pure model work. The rest has some overlap with strats roles but almost none with mode validation and risk.

See also: Can't choose between QIS vs MO quant
 
Last edited:
@CrossGamma your answer was extremely helpful! Yes, thankfully when I had offers I consciously choose this role since it was a front office role working on actually building pricing models. I was pleasantly surprised with the amount of math I had to learn, which I enjoy.

Follow-up questions: about signal based vol trading, is the pricing logic handled by a different individual and the stats guy focuses purely on alpha signals? Or its not compartmentalized at all and the stats guy knows enough pricing to handle it, and then processing signals is the more important part, which is what the focus is on?

@CrossGamma, mind if I DM you on quantnet?
 
Last edited:
@CrossGamma your answer was extremely helpful! Yes, thankfully when I had offers I consciously choose this role since it was a front office role working on actually building pricing models. I was pleasantly surprised with the amount of math I had to learn, which I enjoy.

Follow-up questions: about signal based vol trading, is the pricing logic handled by a different individual and the stats guy focuses purely on alpha signals? Or its not compartmentalized at all and the stats guy knows enough pricing to handle it, and then processing signals is the more important part, which is what the focus is on?

@CrossGamma, mind if I DM you on quantnet?
This will really depend on the company / team how it is structured. My team is fairly open and people get to work on all three aspects depending on needs and preferences. While some are specialized on pricing and naturally tend to pick up the tasks that require a deeper understanding of e.g. stochastic processes, numerics, ... the majority of effort is currently spent on the strategy layer and signal-based trading part. This again very much depends on the company and the maturity of their respective stacks though.
 
I think @CrossGamma said it all.

Since you are in exotic, I assume most of your work is related to Local vol and numerical solutions. It is a highly specialized area which might not be directly applied in vanilla.

But your pricing theory skills are actually quite useful if you join a quant trading firm specialized in listed options (exotics. not so much). The skills are very similar as the core is still no arb pricing. But unlike local vol where you try to brute force everything. In vanilla, a robust (with fewer params ideally) is probably preferred. The theory is still interesting which involves jump diffusion, stochastic vol and even complex analysis (which you don't see much in exotics).
 
Last edited:
A possibility is to upgrade or learn new skills. e.g. become a professional programmer or data science.
Do you use Slang etc.?
 
Last edited:
I think @CrossGamma said it all.

Since you are in exotic, I assume most of your work is related to Local vol and numerical solutions. It is a highly specialized area which might not be directly applied in vanilla.

But your pricing theory skills are actually quite useful if you join a quant trading firm specialized in listed options (exotics. not so much). The skills are very similar as the core is still no arb pricing. But unlike local vol where you try to brute force everything. In vanilla, a robust (with fewer params ideally) is probably preferred. The theory is still interesting which involves jump diffusion, stochastic vol and even complex analysis (which you don't see much in exotics).

Yep that makes sense. Yes, half of my work revolves around stoch local vol and numerics. BTW, I acknowledge that pricing is a little outdated in that it was an attractive area before the GFC, but I enjoy pricing theory and I think its not so straightforward, which IMO can make it a great filtering tool for candidates. Like if someone understands pricing theory really well you can place trust in their abilities etc. I was also curious if industry has the same view in that if a candidate did well in a pricing role, they consider that to be a favorable signal, at least in terms of giving them a shot at the interview?
 
A possibility is to upgrade or learn new skills. e.g. become a professional programmer or data science.
Do you use Slang etc.?
I did give a thought to being an SWE or DS in FAANG etc., but its not that I hate pricing and my current role, in fact I do enjoy pricing theory and sometimes I get projects that are interesting. Just that its a little outdated and more exciting stuff is happening in other areas. So I'm willing to develop skills in other areas and stay within finance, and in that context I was thinking if my current skills were relevant in other areas. Like what CrossGamma mentioned is quite interesting to me and something I'd love to explore.
WRT languages, I'm also familiar with C++ and Python.
 
Last edited:
Yep that makes sense. Yes, half of my work revolves around stoch local vol and numerics. BTW, I acknowledge that pricing is a little outdated in that it was an attractive area before the GFC, but I enjoy pricing theory and I think its not so straightforward, which IMO can make it a great filtering tool for candidates. Like if someone understands pricing theory really well you can place trust in their abilities etc. I was also curious if industry has the same view in that if a candidate did well in a pricing role, they consider that to be a favorable signal, at least in terms of giving them a shot at the interview?
It depends on the firm. If you go for Citadel probably no because they trade mostly cash. But if you go for SIG/Optiver, then yes your pricing theory exp is highly desirable.
 
What about energy markets? a lot of pricing, hedging and optimisation of high-factor American spread options and much more. (real optons as they are called).
I imagine a lot of opportunity here, based on anecdotal evidence.
 
Last edited:
Back
Top Bottom