The core problem

A 10x leveraged long on BTC with margin $M = \frac{S_0}{10}$ faces liquidation risk on relatively small moves. A 5% drop against you wipes out half your margin. A 10% drop puts you at liquidation. The question: can you structure the trade so that the downside at a predefined stop-loss is approximately zero, while preserving the leveraged upside?

Structure of the hedged position

The strategy pairs a 10x leveraged perpetual long with a protective put. The combined P&L at any spot price $S$ is:

$$\text{P\&L}(S) = \underbrace{L \cdot M \cdot \left(\frac{S - S_0}{S_0}\right)}_{\text{leveraged long}} + \underbrace{\max(K - S, 0) - C_{\text{put}}}_{\text{protective put}}$$

where $L = 10$ is the leverage, $M$ is the margin, $S_0$ is the entry price, $K$ is the put strike, and $C_{\text{put}}$ is the premium paid.

The put strike $K$ is chosen near the stop-loss level. At the stop-loss price $S_{\text{SL}}$, the loss on the leveraged position equals:

$$\text{Loss}_{\text{lev}} = L \cdot M \cdot \frac{S_0 - S_{\text{SL}}}{S_0} = S_0 \cdot \frac{S_0 - S_{\text{SL}}}{S_0}$$

For the hedge to work, the put's intrinsic value at the stop-loss must offset this loss:

$$K - S_{\text{SL}} \geq \text{Loss}_{\text{lev}} + C_{\text{put}}$$

This determines the minimum number of put contracts needed. In practice, I solve for the contract count that brings the net P&L at stop-loss to approximately zero.

The IV dependency

Everything hinges on what you pay for the put. And what you pay depends almost entirely on implied volatility at entry. This is where the strategy either works or does not.

I developed an entry checklist based on backtesting across different IV regimes:

  • IV below 55%: Ideal conditions. The hedge cost is low enough that effective leverage after premium remains attractive. The break-even move on the upside is small.
  • IV between 55% and 62%: Acceptable, but the upside is compressed. Effective leverage drops from roughly 10x to approximately 8.5x after accounting for the premium.
  • IV above 62%: Skip. The premium consumed by the put compresses the payoff profile to the point where a simple unhedged position with tighter sizing produces comparable risk-adjusted returns with less complexity.

The volatility skew problem

One insight from backtesting with real Deribit options chain data: the OTM puts you need for the hedge sit on the steep part of the volatility skew. At 0-DTE, a put with a strike 5-7% below spot can have an IV of 65-128%, even when ATM IV is at 52%. This is the classic skew effect, and it gets more extreme as time to expiry shrinks.

In a concrete backtest with BTC at ~$69,000 and options expiring the next day (BTC-3MAR26 chain), every put in the relevant strike range (64k-67.5k) had an IV ask above 62%. The strategy produced no valid entry signal. Near-ATM puts at 68k-69k had acceptable IVs but were too close to entry to serve as meaningful stop-loss protection.

This confirmed an important structural finding: the strategy works better with longer-dated options (2-4 week DTE), where the skew is flatter and OTM puts are not yet inflated by short-dated gamma effects.

Backtesting methodology

The backtesting engine parses Deribit options chain exports (CSV format with instrument names, deltas, IV bid/ask, and prices). For each chain, it:

  1. Estimates spot from the ATM call delta (the call with delta closest to 0.5)
  2. Defines stop-loss scenarios at -5%, -7%, and -10% below entry
  3. For each scenario, finds the 3 closest puts to the SL level
  4. Computes hedge cost, put payoff at SL, and net P&L
  5. Applies the IV checklist to flag valid entries

The output is a full matrix of scenarios with a binary verdict: entry valid or not. Over time, logging these results across different IV regimes and DTE windows builds the empirical foundation for refining the entry criteria.

Automation

The monitoring component connects to the Deribit API in real time, pulls the current options chain, runs the entry checklist, and calculates optimal position sizing when conditions are met. It runs on a VPS and sends alerts. Trade execution remains manual by design. The goal is informed decision-making, not blind automation of a strategy whose edge depends on context the bot cannot fully evaluate.