Welcome to Sparks, Friktion’s DAO contributor driven quantitative research insights on market structure, trading dynamics, and Volt strategy written by our research collaborators. Please check out our previous articles here and here for background context.
Intro
We’re excited to share the results of some of the research problems we’ve been investigating at Friktion, the most nimble and sophisticated options vault in the crypto space.
The popularity of defi option vaults (DOVs) has increased quickly in recent months. Friktion alone has amassed over $100m in TVL in just four weeks of being online on Mainnet. There has been a lot of publicity around the effects of DOVs on the volatility landscape. Implied volatility has been compressed to the lowest we’ve seen in over a year due to DOVs selling the same parts of the curve (5–15 delta region).
Term structure (difference in implied volatility between longer and shorter dated expiries) has steepened due to DOVs preferring to sell shorter term expiries to collect more yield.
What is happening is that short dated out of the money options are becoming cheaper due to an imbalance of weekly sellers from vaults and from external market participants anticipating this flow and preemptively selling before vault auctions.
This is what typically what happens as markets get larger and more mature. Though we at Friktion agree that DOVs can provide a sustainable source of alpha for everyone, we realize that the amount of yield available in the market is finite. As more people start chasing after the yield, the individual share begins to diminish. At the end of the day, the participants that have the tools to understand the market and the agility to react to market dynamics will capture more yield.
The Problem
How do we minimize the market impact and predictability of our Volt strategies?
Given that everyone knows that we have $100M+ of options flow to sell at the same time every week, how do we minimize our price impact and avoid being front-run?For some context, here’s what the current meta looks like.
Deribit (the most liquid centralized options exchange) weekly options expire every Friday at 8am UTC. Options vaults typically hold their auction to sell their options this time as well, with expiries lining up with the Deribit expiry. This makes it easy for market makers to deal with the flow they get from DOVs. Since it’s operationally easier to handle, they can offer better pricing and more yield to the vaults. Right now, every single vault has its auction within 2–5 hours after Deribit expiry. Many other market participants like crypto hedge funds looking for yield will typically sell weekly options during this time. What happens when $500M+ of options flow gets sold at the same delta region at the same expiry at the same time every week? The market reacts and front-runs the flow, eating away at the yield for the vaults.
To highlight this effect, we examined the volatility dynamics in the market around the weekly expiry using a dataset beginning in May 2021. On average, 1-week expiry 10 delta BTC call options fell 2 vol points on average leading up the expiry before recovering 12 hours afterwards.
The difference for ETH calls is even larger. On average, 1-week expiry 10 delta ETH call options fell 3–4 vol points on average leading up the expiry before recovering 12 hours afterwards.
The expiry recently on January 21, 2022 was especially egregious for DOVs with market front-running, with implied volatilities dipping by over 15 vol points before vault auctions with it recovering by 15 points afterwards. This resulted in vaults selling at a much lower implied volatility (collecting less yield) in a volatile market environment.
With implied volatility at around 58 at the time of expiry, a 15 vol point difference for a 10 delta call would mean going from 21% APY -> 52% APY. A huge and unacceptable difference!
Volt Strategy Changes
How should we adjust our Volt strategies to avoid getting front-run by the market? The data tells us that we should hold our auctions 6 hours before or 12 hours after Deribit expiry to take advantage of higher implied volatility environments. There are two complexities with moving around the auction time:
- If the expiry doesn’t line up with Deribit expiry, market makers have to pay more to hedge the position. This eats into the pricing they can give and lowers yield. This makes it difficult to start the auction earlier than Deribit expiry.
- If the expiry lines up with Deribit expiry, but starts 12 hours after, then we lose out on 12 hours of time value on our option.
Altcoin Strategy
More than 70% of the TVL in Friktion’s protocol is concentrated in non-BTC/ETH tokens, which don’t currently have CEX listed options. Because we don’t have to worry about lining up expiries, we can choose whenever we want to hold alt-coin auctions. Market makers typically price alt coins as a multiple of ETH volatility, so the effects of holding alt-coin auctions earlier when ETH volatility is systemically higher will result in even more yield for Friktion Volts!
Below is a table of historical implied volatility and yield is for Friktion vault listed alt-coins alongside a hypothetical increase in yield if the implied volatility were to increase by 5 points on average by moving the auction 6 hours forward.
That’s pretty juicy!
BTC/ETH Strategy
Since we have to worry about lining up expiries with BTC/ETH, our adjustment to the Volt strategy requires a little more nuance.
Selling BTC/ETH along with the altcoins 6 hours earlier is an option, but there are two complexities to doing that:
- Having market makers to lining up the expiries eats into our yield. They typically hedge BTC/ETH options with Deribit options/futures that expire at the same time and perps. If the options expired 6 hours earlier, they would have to unwind their hedges instead of holding it into expiry, adding additional cost. It’s unclear if 2–4 extra vol points of yield is worth the decrease in yield from extra hedging.
- Moving the BTC/ETH options times will cause the market to react eventually. Friktion’s BTC Volt is the largest BTC vault in defi. If we move our BTC/ETH auction 6 hours earlier, eventually the market will start anticipating our move and reduce the efficacy of having our auction earlier. Furthermore, this might reduce the benefit of moving our altcoin auctions earlier to get more yield, since altcoin pricing is largely a function of BTC/ETH volatilities.
The other alternative is to sell the BTC/ETH options 12 hours later, at 20:00 UTC. The downside to this is that we lose 12 hours of time value as mentioned earlier. As much as we want to avoid situations like 1/21/22 where we sell at the pico-bottom of implied volatility,
We don’t want to over-adjust and have situations like 1/14/22 play out where if we were to hypothetically sell 12 hours later, we’d lose 12 hours of time value and 3 vol points. The worst of both worlds.
How much is 12 hours of time value worth? It depends on what the implied volatility is, but it’s roughly 3.5% APY.
Historically, has it been worth it to move all auctions back? Does the increase in implied volatility make up for the decrease in time value we collect?
If we do a backtest where we assume we always sell the 10 delta strike at the implied vol for each hour after the auction and adjust for a decrease in time value, what does that look like?
The below plots show historical yields by week of moving the auction back 0 to 14 hours (which each line representing the cumulative yield of moving the auction back x hours) subtracted by the baseline (doing the auction exactly at Deribit expiry). For example, the blue line corresponding to 0 is always 0 because we are subtracting it by itself. The orange line corresponding to 1 is the cumulative difference of yield by moving the auction 1 hour back and the baseline. Notice how no lines end up positive. This means that naively moving auctions back doesn’t make up for the loss in time value for all time periods.
So moving all auctions back doesn’t make up for the loss in time value. Notice however, that during certain weeks, it is very much worth it to move the auction. Can we predict those weeks and adjust accordingly?
One such signal we can use is market implied volatility vs. realized volatility. When realized volatility >> implied volatility, we probably want to avoid selling since we are less likely to be fairly compensated for the volatility we are shorting.
Though the data is dominated by a few outliers, the signal can help us prevent catastrophic weeks like in January 21 where we undersell ourselves.
After investigating the data, the optimal strategy is to hold an auction at normal times immediately after Deribit expiry. If we notice that the market makers are offering prices with implied volatility << 24 hour realized volatility, then delay the auction until implied volatility is closer to realized volatility.
Conclusion
Based on the data we’ve seen, we will move all altcoin auctions 6 hours earlier at 18:00 UTC on Thursdays to take advantage of systemically higher volatility during that time. We’re keeping BTC/ETH at the normal expiry time with the option of moving it 12 hours later during times where implied volatility is under realized volatility.
We think that these Volt strategy adjustments will capitalize on current market dynamics to capture more yield for our users. Other protocols will no doubt copy our analysis and we will have to adapt accordingly, but our work shows we are leading the efforts to produce the best risk-reward DOV returns in the space. At Friktion, we believe our deep understanding of the markets and trading coupled with our agility in reacting to whatever the market throws at us gives us an edge over other DOV protocols in seeking yield for our users.
Interested in contributing to the Friktion quantitative research and strategy effort? Join one of our Taskforces and let’s build the future of finance together.
Website: app.friktion.fi
Twitter: twitter.com/friktion_labs
Discord: discord.gg/beCbA5VpJu
Medium: friktionlabs.medium.com
Documentation: docs.friktion.fi
Shoutouts to our friends at Genesis Volatility and Deribit Insights for their contributions to the research. Please check them out below
