Backtesting option-writing strategies

ETH Covered Call

Options, as we’ve mentioned before, are a lot like chess (or warfare). For every market situation, there is an optimal combination. That’s why we’ve decided to explore the possibilities opened by derivative products, combining those with web3 capabilities to create something easy to use yet customizable.

Neuron Pools

So, we’ve settled on Neuron pools — a product that disrupts the DeFi landscape with summarized multi-layered yield from depositing otherwise idle collaterals into liquidity pools, along with smart option writing. More on the latter below (composable yield concept deserves an article on its own!).

Why test at all? Why all this option fuss?

Providing liquidity is virtually sans market risks (not counting hacking and impermanent loss in, of course). When we up the capital efficiency game by adding another layer of yield — option-writing strategies — both APY and risk start to depend on mainly one variable: the right strike price to mint options at.

This “capital efficiency”, as you call it — is it on the same network, this same protocol, with us?

Too far out-of-the-money — and you’re not covering the fees. Too close, being greedy selling expensive options — and you’re seriously risking going bust: option writing is notorious for being relentless risks concerned.

When your own $30.000 are at risk, no problem; yet with potentially millions at stake, millions you’ve been trusted with — it all demands for serious preliminary research, especially for an automated protocol like Neuron. So, we’ve started developing a methodology for choosing the optimal strike price with available information each week.

One thing to consider: think of the importance of strike selection in Neuron as a world-class athlete’s shoes. If they are here, he’s able to show incredible feats of power, adding up to 15% APY on top of one’s already owned assets. No shoes or wrong shoes — a lot less efficient and potentially riskier (one only needs one broken leg to drop out of competitions for several months).

So choosing the optimal strike price is crucial for bringing Neuron’s true powers — summarized yield via combining option-writing strategies with increased collateral efficiency.

Whatcha gonna do? Listen to Bob Dylan, of course, and do a little DIY backtesting.

Backtesting process

First in line was ironing out the following preliminary process for our small research, which went as follows:

• Getting the ETH price and volatility data, or option tick prices (prices trades were executed at, e.g. “real” prices)— preferably, from several sources for comparison
• Pricing the option chain using the Black–Scholes–Merton model if using ETH prices
• Simulating a simple covered call strategy over several deltas
• Choosing the optimal delta for selecting the strike price reward/risk-wise
• Presenting the results and then open-sourcing the code — we’re [hand for hands in team] for open research! And yeah, we’d done it all in Python…

Assumptions

• We’re assuming expiration price > strike price = exercising when backtesting
• Also, a crucial fieldnote: this process is dynamic; we will be doing this every week(more — we’ll be improving on the process), with a rolling time window. The world’s constantly changing, fitting the Buddhist concept of Aniccha — it is only practical that variables change, too.

Data

We’ve used two data sources (diversify thy sources of truth, as the old maxim goes):

• Messari, one of crypto research and analytics mastodons, is providing up to 5 years’ worth of price data — and volatility, too
• Deribit, de-facto the biggest crypto derivatives exchange (off-chain, though), provides an easy-to-use API for downloading option chain prices

P.S. For those aching for an example of working with Deribit — here it is.

Results

Mr. Taleb preparing to be as antifragile as possible.

Disclaimer: the first week’21 revealed an event worthy of being called a black swan for such an asset— both ETH price and volatility soaring up 25%+ in a couple of days. This is the worst possible scenario for covered call strategy:

• We are betting on price staying roughly the same or increasing — a little bit. When price moons, we’re having our FOMO literally materialized, missing on asset price appreciation.
• We’re banking on option’s extrinsic value, basically selling its time and implied volatility value. With implied volatility getting higher, the option price rises, and we’re experiencing unrealized loss during the week.
• Last but not the least — if the expiration price exceeds strike, we’re at a very real possibility of being exercised. We’d want to avoid; think of exercising when selling options as a big bad raid boss in any RPG game you’ve played. Only the boss’s 20 levels higher…

Still, we’ve got no choice but to advance (aye). First, let’s find the optimal delta:

Apr’19 — May’21

Jan’20 — May’21

Suffice to say ~0.1 delta strikes are the best regardless of time periods and data sources used — finally, something robust and stable in our VUCA world!

Then, off to backtesting with optimal delta (actually, everything’s already on the previous charts — each point on the delta selection chart is the last point on some simulation chart; in our case, for 0.1 delta). Spoiler: the “eat like a butterfly, poop like an elephant” adage is frequently used with options writing for a reason, but we’re always over buy&hodl.

USD-denominated yields since Apr’19. As can be seen, we’re exceeding B&H:

Why not tread on more realistic, calmer waters? 2021, although rough, had been lighter on black swans. Let’s try backtesting after the 2–8th Jan:

Still over B&H, wgmi!

Some data…

A research piece can’t be legitimately called so without some crude and honest-to-earth data, and we’ve got it (for the 2021–2022 range, in particular):

Basically, it’s 203.6% APY for our strategy vs. 174.8% for buy and hold — a hefty difference one can pocket. Mind you, it’s an almost 29% yield per year for one’s assets otherwise sitting idle.

Conclusion

Here, we’ve presented an approach for choosing the strike price by closest to the optimal delta. It transfers and is scaled to any future and/or existing markets easily; we’ll continue to advance on-chain derivatives further and publish more research.

Neuron makes collateral management for options easy and flexible.

Stay updated on:

Twitter: https://twitter.com/neuronfund

Discord: https://discord.com/invite/SFasvmAwSr

Telegram: https://t.me/neuronfund