I’ve started building a 1 DTE ETH options trading bot for Bybit, with the flexibility to extend to Deribit and potentially any exchange that offers an options chain and API access.
Bybit is the primary focus, as it’s where most options trading for the Terramatris crypto hedge fund currently takes place.
In the early days of Terramatris, the options strategy was entirely manual. I focused on trading 1 DTE Ethereum options, primarily selling far out-of-the-money contracts with very low delta.
Over roughly a one-year period, the performance was surprisingly consistent. Only about 5–6 trades—with delta below -0.1—were meaningfully challenged.
That said, those weren’t truly systematic trades. They were entered manually, driven by human judgment—and at times, emotion. Some trades were more aggressive, position sizes were occasionally increased, and decisions weren’t always consistent. It was, in many ways, discretionary trading shaped by experience rather than a fully disciplined, rule-based system.
Over the years, the strategy evolved from daily 1 DTE trades to a more structured weekly approach, and along the way we developed our own Ethereum strategy.
With the rise of AI and the accessibility of rapid “vibe coding,” I decided to build a simple app on my MacBook using ChatGPT. There’s nothing sophisticated in terms of AI decision-making yet—the trading logic is purely rule-based, implemented in Python.
Looking ahead, the plan is to incorporate machine learning elements that can better replicate aspects of human decision-making—such as adjusting aggressiveness based on volatility, scaling position sizes, and adapting to changing market conditions.
At its current stage, the bot scans Bybit for 1 DTE ETH options, ranks available contracts, and selects the trade with the highest probability of expiring worthless. The parameters can be adjusted to be more aggressive or conservative, but the core logic reflects my own approach—shaped by roughly three years of experience running Terramatris—favoring 1 DTE puts with delta below -0.1.
Once candidates are ranked, the bot automatically executes the trade.
Low-delta, short-dated options—when selected carefully—tend to expire worthless most of the time. That’s the core edge: not prediction, but probability.
At its current stage, a basic hedging mechanism is in place. If a position is in the money at expiry, the bot automatically buys the underlying on Bybit, effectively mimicking a more traditional spot assignment. From there, it does not yet transition into a covered call strategy, although that’s a logical next step for future development.
For now, the post-assignment handling remains partly manual. The bot continues selling new puts, while any acquired spot is managed separately with a longer-term perspective—potentially to be used later for weekly, monthly, or quarterly options strategies.
So far, after the first five days of trading, there have been no assignments, resulting in a 100% win rate. That said, this is highly dependent on conservative settings—more aggressive parameters would naturally increase assignment risk. The core objective remains minimizing assignments as much as possible.
Early results look promising, with projected yields around ~18% annually from very deep out-of-the-money puts, though this is based on a very limited sample and should be treated cautiously.
This is still an early-stage project, and there are no plans for public access for at least a year. It’s being developed strictly as an in-house tool for the Terramatris crypto hedge fund.
Right now, the priority is simple: ensure the bot can execute trades end-to-end reliably, without any manual intervention.
Expanding beyond ETH—to BTC or other supported assets—is relatively straightforward in theory, but the focus is on getting ETH to work flawlessly first.
What’s next
The core idea is to expand it into a more complete options engine. That means adding advanced hedging logic (e.g., automatically going long on assignment and switching into call writing — essentially a wheel strategy), and structuring it around predefined strategies like 1DTE, 7DTE, and 30DTE. In theory, it already supports longer-dated options, but it needs clearer strategy layers rather than ad hoc execution.
On the infrastructure side, the plan is to add Deribit support, expand to multiple cryptocurrencies, and introduce user accounts. Right now, it’s just running locally on my MacBook—far from production-ready.
The next step is to move it to a VPS once most core features are in place. From there, development would shift toward more structured updates, likely through periodic version upgrades (e.g., monthly iterations).
There are no plans to make the code public.
Before anything goes public, a few non-negotiables:
- stable execution across multiple cycles
- basic risk controls
- logging and monitoring
- edge-case handling (API failures, liquidity issues, etc.)
If you're curious
The bot is not available yet. But if you’re interested in seeing how it works in practice—or want to follow its development—feel free to reach out.
One important caveat: the ~18% premium starts to look meaningful primarily on larger account sizes, roughly $100K and above. At that scale, liquidity depth on Bybit becomes a relevant constraint, which is why expanding to Deribit is under consideration. Realistically, around $10K would be a minimum to make the strategy worthwhile from a practical standpoint.
At Terramatris, however, we’re approaching this differently—starting small, from around $100 and gradually adding $25 weekly (generated from the core Ethereum strategy). This allows the bot to evolve in a controlled, low-risk environment while validating execution and logic over time.
