Glossary · every term explained
Plain-English definitions of every label, abbreviation, and Greek used on this dashboard. Hover any underlined term anywhere on the site to see a short version of these definitions.
The trade
Convertible bond
"CB"A bond that can be exchanged for the issuer's stock at a fixed price. Has two values bundled together: the bond's debt value + the embedded equity option's value.
Conversion price
in JPY per shareThe price at which the bond converts to stock. Example: conversion price ¥9,346 means each ¥100 of bond face becomes 100/9,346 = 0.0107 shares.
Parity
What the bond is worth if you converted it to stock right now. Parity = stock price × conversion ratio. If parity > bond price, you can convert and profit immediately (rare).
Premium
How much extra you're paying for the bond above its conversion value. Example: bond price 131, parity 119 → premium 10%. High premium = bond is more like debt; low premium = bond is more like equity.
Cheap% (cheap to model)
Our model's view: (model fair value − market price) / market price. Positive = market is too low vs the model's fair value (potential buy). Negative = market is too high (potential short or just avoid).
Δ-hedge (delta hedge)
To isolate the bond's pricing edge from general stock direction, you short Δ × shares of the underlying stock. This way you profit if the bond moves to fair value regardless of where the stock goes.
The pricing inputs
Mkt
market price, % of parWhat you'd pay (or receive) for the bond right now. Quoted as % of par. 100 = at par. 130 = trading 30% above issue price.
Model
model fair value, % of parWhat our pricer thinks the bond should be worth, given the stock price, vol, credit spread, and yield curve. Same units as Mkt.
HV60
historical volatility, 60-dayHow wildly the underlying stock has moved over the last 60 trading days, expressed as annualized standard deviation of log returns. 40% means the stock typically moves ±40% over a year.
IV
implied volatilityWhat volatility number would make the pricing model output equal to the bond's market price. If IV is much lower than HV60, the market is underpricing future volatility → bond looks cheap to model.
IV-HV
Implied vol minus realized vol. Negative means the bond market is paying for less vol than the stock has actually been showing — the option is on sale.
Spread (Spd)
credit spread, basis pointsHow much extra yield this issuer pays over the risk-free JGB yield. AAA = ~25bp. BBB = ~190bp. Unrated junk = 400-700bp. Used in the model to discount the bond's cash flows.
Rtg
issuer credit ratingS&P/Moody's-equivalent letter grade for the issuer. AAA is best, D is default. Investment Grade = BBB- and above. Below = high yield / junk.
The Greeks (sensitivities)
Δ (delta)
How much the bond's price moves for a ¥1 move in the underlying stock. Δ=0.05 means if stock rises ¥100, bond rises ¥5. Used as the hedge ratio: short Δ × shares per ¥1 of bond face.
Γ (gamma)
How fast delta itself changes. High gamma = the hedge ratio drifts quickly, so you need to rebalance the equity short more often.
Vega
How much the bond moves for a 1 percentage-point change in volatility. Vega of 0.5 means if vol rises from 30% to 31%, the bond gains 0.5 (% of par).
Theta (θ)
time decayHow much the bond loses per day as time passes (the embedded option decays). Negative theta = you're paying rent to hold the position.
Rho (ρ)
How much the bond moves for a 1bp change in the risk-free rate. Usually small for short-tenor CBs.
The agentic workflow
Tsiveriotis-Fernandes
The 1998 model the pricer uses. Splits the bond's value into an equity-driven component (discounted at the risk-free rate, since converting eliminates default risk) and a debt-driven component (discounted at risk-free + credit spread, since default risk applies). The standard CB pricing approach.
Binomial tree
Discrete grid simulating the stock price over time. At each node, the model decides "convert / put / hold" optimally. With 250+ time steps it converges to the continuous Black-Scholes-style solution.
Monte Carlo (MC)
Used here when a bond has a path-dependent reset feature. Simulates thousands of stock paths, applying the reset rule along the way, then averages the payoffs. Slower than the tree but handles features the tree can't.
Reset / refixing clause
Common in Japanese small-cap CBs. If the stock falls below a trigger (e.g. 80% of conversion price) on a specified date, the conversion price drops (e.g. to max(stock, 70% of original)). Bondholders get more shares per ¥100 face — protects against equity decay.
Soft call (130% trigger)
The issuer can call (redeem) the bond if the stock trades above 130% of the conversion price for 20+ days. Forces the bondholder to convert or take cash — caps upside for the bondholder.
JGB
Japanese Government BondThe risk-free rate used in pricing. We pull 2y, 5y, 7y, 10y, 20y, 30y yields directly from Refinitiv and interpolate to match each CB's maturity.
BUY / WATCH / AVOID
BUY = cheap ≥ 5%, high confidence, no data flags.
WATCH = cheap ≥ 5% but has reset clause (model uses MC pricer; needs human review).
AVOID = rich vs model (model price < market price).
Hit rate
Out of all the times the model flagged a bond as cheap, the % where the bond actually went up over the next N days. 71% hit rate at 60d in the 5-10% bucket means: of bonds the model called cheap by 5-10%, 71% had a positive return 60 days later.
Anomaly / stale conv price
When a stock split happens but the dealer's quoted conversion price hasn't been updated, the bond looks artificially "cheap" by 30-50%. Backtest showed these have 0% hit rate — the system auto-suppresses them.