2
$\begingroup$

I realize that this question may be verging on asking for the proprietary/"secret", so if suggestion of a general approach that doesn't divulge details isn't really possible, I understand.

From what I've seen of the literature, the usual approach to profiting off of modeling options seems to be of a sell-side/market-making nature.

To my knowledge, the approach usually goes like this: Some sort of potentially-modified model (e.g. BSM, Heston, etc.) is chosen, the model is then fit to quoted vanilla options prices on said underlying, then this resulting model is either used to price more exotic options (in the sell-side context) or to calibrate greeks so as to market-make, hedge, and, as a result, profit off of the spread (in the market-making context).

Let us assume that, henceforth, "options" will generally refer to European or American options, specifically, calls and/or puts.

I realize that there are ways to leverage these pricing models to make some sort of "speculative" profit by, for example: using the model fitted on something other than market prices to identify mispriced (in the context of the model) options, buying and/or selling the options accordingly, and then delta-hedging our position until (hopeful) convergence to capture the disparity.

I'm wanting to know if there is a general approach that is used in practice to predict the future value of an option or set of options, compare it to it's current price, and determine if a profit can be made through either going long or short the option or set of options without the need to dynamically (or even statically?) hedge (since this may not be feasible/possible) or invoke discounting expectations under the real-world measure (due to the numerous ways that this may complicate things).

Does this just come down to doing something like this post suggests and just choosing/creating a pricing model, forecasting the parameters over some future time horizon (say until expiration), and then plugging these forecasts into the pricing model to get the prediction of the future value of the option over said time horizon?

Alternatively, would there be anything inherently "wrong" with treating the option prices acoss strikes for the same expiration (although, this could be extended to multiple expirations as well) as a multivariate time series/sequence and then applying the various methods designed to model those to the option prices? Or do the complications inherent in non-European options (early exercise, etc.) undermine this approach (for prediction of American Options)?

Any input/references would be greatly appreciated, thanks!

$\endgroup$
1
  • 2
    $\begingroup$ In addition to the bid-offer spread that you mention in your post ("profit off of the spread (in the market-making context)"), options are usually traded to trade the volatility. At inception, the option seller prices the option using the "forward-looking" implied volatility. Then, if the realized volatility proves smaller than the IV priced at inception, the option seller would make money whilst delta-hedging the option from inception through to maturity (quant.stackexchange.com/questions/60762/…) $\endgroup$ May 2, 2023 at 11:30

1 Answer 1

2
$\begingroup$

In addition to my comment, these are some option trading strategies that involve "examining the future possible prices of options" (and I look forward to what other contributors might add):

1) Gamma Trading: the option seller sets the IV at inception to a value that he / she believes is higher than what will be the realized volatility. After selling the option, the option seller would delta-hedge this strategy from inception through to maturity. If the realized volatility turns lower than what the IV was set to at inception, the strategy will be profitable (see Jim Gatheral, Nassim Nicholas Taleb - The volatility surface: A practitioner's guide (2006), page 154. PS: would be interesting to compute how the frequency of the hedging and bid-offer spreads affect the profitability though).

2) Vega trading: a market participant will go "long vega" (i.e. long call or long put or both) in the hope that options will become more expensive due to a higher volatility in the future (i.e. you can have a view on the SPX500 taking a dive and buy some put options: then you will make money not only on the delta, but also on the vega: i.e. the put options becoming substantially more expensive if the realized volatility increases, driving also the future IV higher; then you sell the options, to make money on the vega + delta).

(the Vega-trading strategy works obv. in reverse too: say the SPX500 had already taken a dive and option prices have increased due to higher IV: if your view is that the volatility will dissipate, you sell vega (put or an ATM call that you can delta-hedge via futures) to pocket the high premium. Then you close the vega in the future at lower IV).

In general, if you look at today's volatility surface priced by the market and you believe that something about the way that the surface "implies" future option prices is wrong, you can take a corresponding view.

An example would be a vol surface which prices short-dated options (say 1-week expiry) at a relatively high IV (say because the next week involves a FED decision, so some uncertainty), but if you look further into the future, the implied prices of short-dated forward starting options covering the same event (i.e. FED in six months) are relatively low (I am talking about 1-week options starting in six months).

It is quite probable that if the market prices the current one-week options at high IV, as we approach the 1-week options starting in six months, the IV of those will also increase: so the right strategy would be to go long the 1-week options starting in six months and hold until we get closer to the date (in fact, non-stochastic volatility models for pricing options do have this problem of miss-pricing forward-starting options, so if one bank uses a non-stochastic vol model to price forward-starting options, they can easily get exploited by a more sophisticated bank that uses - say - the Heston model).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.