Investment Memo · CyberAgent

The bond trades at 107.25 (% of par). Our binomial-tree pricer, using the underlying stock at ¥1,316, 60-day realized vol of 36%, and a credit spread of 429bp (based on the NR issuer rating), arrives at fair value of 103.84.

Implied vol (42%) is roughly in line with realized vol (36%). The mispricing is primarily in the bond's level relative to model, not a vol gap.

Three convergence paths can close the gap: (1) the dealer updates the indicative quote toward fair value (the most common path on illiquid JP CBs), (2) realized volatility forces a re-mark of the embedded option, or (3) a corporate event (coupon, conversion period start, call window opening) prompts a fresh price.

For each ¥100M face, buy the bond at 107.25 (= ¥107,250,000 outlay). Short delta × shares of 4751.T (37,465 shares at ¥1,316 = ¥49,322,724 short proceeds). Rebalance the hedge every 5 trading days or when delta drifts more than ±10%.

Greeks: Δ 0.037 · Γ -0.00004 · Vega 0.57 · θ +0.003 per day per ¥100 face. Daily theta on a ¥100M position is approximately ¥2,513 of carry you collect.

Across the 13-month backtest, the strategy generated 38 trades total. Top 5 issuers contributed 113% of P&L — meaning the rest of the book was net-flat. This signal is not broadly distributed; it lives in a handful of issuers.

• Earnings release for CyberAgent — usually triggers a fresh dealer mark on the bond
• Conversion window milestones — when conversion becomes possible, or close to call dates, dealers re-quote
• Stock breakout above conversion price × 130% — triggers the soft call clause; can force convergence
• Realized vol spike — option's value re-prices upward; bond catches up
• Index inclusion / removal of the equity — forced flow that moves the stock and the option

• The gap widens before it closes. Dealer marks the bond down further before catching up — mark-to-market pain even if eventually right.
• Issuer-specific event: credit downgrade, profit warning, or M&A invalidates the spread assumption.
• Realized vol crashes: the option loses value faster than the dealer catches up. Theta bleed exceeds gap-closure.
• Liquidity: bid-ask spreads on this bond may be 50-100bp; entry/exit slippage erodes the edge.
• Stale conversion price: if a corporate action (split, special dividend) has hit the underlying without the conversion price being adjusted in our data, the model is wrong. Always cross-check parity.

Recommendation: HOLD at ¥100M face, Δ-hedged with 37,465 shares of 4751.T. Target convergence within 60 days. Hard stop if cheap% widens past +2% (gap blowing out) or narrows below +1% (take profit).