ManualReference

Glossary

Every label, value, and parameter in the desk, defined and grouped by area. This is the same glossary that powers the inline tooltips, rendered in full with its math.

IV Surface

18 terms

Arbitrage-free: The vol shape leaves no risk-free money on the table — green only when every no-arbitrage check passes.
Calendar: total variance non-decreasing in $T$ .
Butterfly: Durrleman $g (k) \geq 0$ across the $k$ -grid.
Green when both hold; otherwise flagged.
ATM term structure: At-the-money implied vol at the nearest expiry vs the farthest — the term-structure of ATM vol.
$frontIV = rows [0] . atmIV$ , $backIV = rows [last] . atmIV$ .
Arrow reads near→far, rendered as integer %.
Butterfly arbitrage: A smile shaped so a butterfly spread would cost less than nothing — an impossible negative implied density.
$g (k) = (1 - \frac{k w ^{'}}{2 w})^{2} - \frac{w ^{'2}}{4} (\frac{1}{w} + \frac{1}{4}) + \frac{w ^{''}}{2}$ $g (k) = (1 - \frac{k w ^{'}}{2 w})^{2} - \frac{w ^{'2}}{4} (\frac{1}{w} + \frac{1}{4}) + \frac{w ^{''}}{2}$
Arb-free when Durrleman $g (k) \geq 0$ across the strike grid.
Flagged red the moment $g (k) < 0$ at any sampled $k$ .
Calendar arbitrage: Total variance must never fall as expiry lengthens. ✕ means a longer-dated tenor is priced cheaper (in variance) than a nearer one at the same strike — free money on the table.
$w_{T_{2}} (k) \geq w_{T_{1}} (k) (T_{2} > T_{1})$ $w_{T_{2}} (k) \geq w_{T_{1}} (k) (T_{2} > T_{1})$
The pill flips red the moment this is violated for any sampled $k$ .
Checked across the live term structure by checkCalendar.
IV color key: The surface's color encodes implied vol — dark blue is the lowest vol on screen, bright cyan/white the highest.
Ramp maps normalized IV in $[0, 1]$ across the current surface.
Relative scale, not an absolute vol scale.
IV surface: A 3-D map of how BTC option-implied volatility varies by strike and time-to-expiry — per-expiry SVI smiles stacked into a $(k, T) \to IV$ surface.
$IV (k) = w (k) / T, w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$ $IV (k) = w (k) / T, w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$
Moneyness readout: How far the cursor's strike sits above ( $+$ ) or below ( $-$ ) the forward, in percent.
$(e^{k} - 1) \times 100$ $(e^{k} - 1) \times 100$
$k$ is log-moneyness; $k = 0$ is at-the-money.
Ticks labeled ±10/20% are log-moneyness, so $\pm 0.2 \approx - 18%.. + 22%$ actual.
Surface IV readout: Implied volatility at the point under your cursor — the headline value, an annualized vol fraction.
$iv = min (w / T, 1.0)$ $iv = min (w / T, 1.0)$
Evaluated at the snapped real tenor; shown ×100 to 1 dp.
SVI model: The Stochastic-Volatility-Inspired model (Gatheral & Jacquier 2014) that fits the whole vol smile with just five numbers.
$w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$ $w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$
Parameterizes total variance vs log-moneyness $k$ .
Real vs interpolated ribs: Bright grid lines are real on-chain expiries; dimmer lines are smooth fill between them.
Interp ribs multiply color by $0.42$ .
Built in total-variance space (linear in $T$ ) to stay calendar-arb-free.
Strike axis: Moving right raises the option strike across the surface.
Columns span log-moneyness $k = - 0.2..0.2$ ( $\approx - 18%$ to $+ 22%$ of forward).
No numeric strike ticks are drawn.
Surface height: How tall a point rises encodes relatively higher IV on this surface.
$y = norm \cdot Y_{SCALE}, norm = \frac{iv - ivMin}{span}$ $y = norm \cdot Y_{SCALE}, norm = \frac{iv - ivMin}{span}$
Normalized across the whole surface to fill the frame.
Not an absolute IV height.
Tenor countdown: A live countdown to an expiry along the surface depth.
$fmtDuration (expiry - now)$ $fmtDuration (expiry - now)$
Formats $D d H h$ / $H h M M m$ / $M m S S s$ / $S s$ , or 'expired' if $\leq 0$ .
Nearest tenor highlighted cerulean; hovered tenor underlined.
Tick pulse: The wireframe briefly brightens when a fresh on-chain SVI update arrives, signaling the desk is live.
On version change (checksum of SVI data), wire opacity ramps $0.8 \to 0.5$ over 700 ms.
Drag to orbit: Drag to rotate the camera around the surface.
Camera-only orbit, damped; no pan or zoom.
Polar angle clamped; auto-rotates while not hovering.
Tenor count: How many distinct real on-chain expiries are stacked into the surface right now.
Count of active oracle tenors with valid SVI and $T > 0$ .
Interpolated fill-ribs between them are not counted.
Surface resolving: The surface is still loading and not yet drawable.
Shown until at least two expiries have data.
buildSurface needs $\geq 2$ tenors to render a surface.
Tenor axis: The time-to-expiry axis running along the surface's depth.
Each entry is the live countdown to one real oracle expiry.
Nearest tenor sits closest to the camera; structure recedes as $T$ grows.

Smile / Skew

17 terms

ATM implied volatility: The market's expected annualized BTC swing — the single most-watched vol number. Also sizes the vault's stress shock.
$σ_{IV} (0) = w (0) / T$ $σ_{IV} (0) = w (0) / T$
Evaluated at-the-money ( $k = 0$ ); live BTC ≈ 42–45%.
Call wing: Implied vol 10% above the forward (the upside/call wing) — how expensive upside bets are.
$IV (+ 0.1) = w (0.1) / T$ $IV (+ 0.1) = w (0.1) / T$
$impliedVol (+ 0.1, p, T) \times 100$ ; strike ≈ 10% over forward.
Shown to 1 dp.
Butterfly — fly: How much the wings curve up versus the at-the-money point — a measure of smile convexity.
$fly = \frac{put + call}{2} - atm$ $fly = \frac{put + call}{2} - atm$
In IV points; positive = wings richer than ATM.
fmtSigned, 1 dp.
Put wing: Implied vol 10% below the forward (the downside/put wing) — how expensive crash protection is.
$IV (- 0.1) = w (- 0.1) / T$ $IV (- 0.1) = w (- 0.1) / T$
$impliedVol (- 0.1, p, T) \times 100$ ; strike ≈ 10% under forward.
Shown to 1 dp.
Risk reversal — skew: How much pricier downside is than upside — put minus call IV at ±10% moneyness, the headline crash-fear gauge.
$RR = IV (- 0.1) - IV (+ 0.1)$ $RR = IV (- 0.1) - IV (+ 0.1)$
Negative = market paying up for puts (put skew, normal for BTC).
fmtSigned, 1 dp; warn tone when $< 0$ .
OI chart x-axis: Strike distance below spot on the left, ATM at center, strike distance above on the right — so OI lines up with the smile.
$reach = max (0.005, max ∣ mny ∣) \times 1.25$ $reach = max (0.005, max ∣ mny ∣) \times 1.25$
Span is symmetric and adaptive so near-ATM strikes stay legible.
OI bar hover: Hovering an open-interest bar reveals that strike's price, its distance from spot, and the Up versus Dn position counts.
$fmtPrice (strike)$ , then signed $fmtMny (strike / spot - 1)$ .
Raw up / dn position counts shown after.
Live / Replay: Live is the newest on-chain smile; Replay is a scrubbed past snapshot stamped with its UTC capture time.
Replay pill (warn tone) shows $utcClock (checkpoint_timestamp_ms)$ .
Appears only when you scrub back in history.
Open interest by strike: How many live binary positions sit at each strike, grouped and aligned under the smile.
PositionMinted grouped by strike, filtered to $∣ strike / spot - 1∣ \leq 0.5$ .
Bar height = up + dn count per strike.
Smile hover: Hovering the smile reads off the implied vol and moneyness right under the cursor.
$IV \times 100$ to 1 dp in accent; signed $k \times 100$ in faint.
Snaps to the nearest of 80 sampled points.
Vol smile: One expiry's vol smile — how IV changes across strikes and whether it tilts toward puts or calls.
$IV (k) = w (k) / T, k = ln (K / F)$ $IV (k) = w (k) / T, k = ln (K / F)$
Plotted from the live SVI fit, with skew/fly readouts.
Smile x-axis: Deep-put edge at −30% moneyness on the left, ATM at center, deep-call edge at +30% on the right.
Endpoints are SMILE_OPTS $k_{m i n} / k_{m a x} = - 0.3/ + 0.3$ .
Center strike = forward.
Smile y-axis top: The top of the IV scale on the smile chart marks the highest vol shown.
$y_{top} = (max (IV) + 12% pad) \times 100$ $y_{top} = (max (IV) + 12% pad) \times 100$
Top-left label; only the high end is ticked, the low end has none.
Time travel: A scrubber that replays how this expiry's smile evolved over past SVI updates.
Slider $0 \dots liveIdx$ across the SVI history array.
Shown only when $> 1$ snapshot; pushing to liveIdx returns to live.
Up / Dn legend: Color key for binary direction: green Up bets the price ends above strike, red Dn bets it ends below.
$upFrac = \frac{up}{up + dn}$ $upFrac = \frac{up}{up + dn}$
In the OI histogram each bar is a stacked green-over-red split.
Jump to live: Resets the replay scrubber back to the newest live smile.
Faint 'Live' when already live; accent '→ Jump to live' when scrubbed back.
Sets scrub = null.
Smile status states: Status messages when there's no live curve — no oracle selected, the feed failed, or this expiry already settled.
Settled when $status = settled$ or $T \leq 0$ .
Settled still shows the arb verdict on the last fitted params.

Oracle State · SVI Params

18 terms

ATM total variance: The raw un-annualized variance at-the-money that ATM IV is derived from – near-zero for short expiries.
$σ_{ATM}^{2} T = w (0) = a + b (ρ (- m) + m^{2} + σ^{2})$ $σ_{ATM}^{2} T = w (0) = a + b (ρ (- m) + m^{2} + σ^{2})$
Source $atmVar = totalVariance (0, p)$ ; shown in $toExponential (2)$ .
IV recovered as $w (0) / T$ .
Checkpoint: The Sui checkpoint – a block-height-like sequence number – this snapshot came from, its proof of freshness.
Roughly 9–11 digit monotonic ledger height.
Shown with thousands separators.
Expiry: Time left until this contract settles, or 'Settled' once it is done.
$σ_{IV} (k) = w (k) / T$ $σ_{IV} (k) = w (k) / T$
From $fmtDuration (expiry - now)$ ; drives the time-to-expiry $T$ .
Renders 'expired' when $T \leq 0$ .
Forward — $F$: The price the market expects BTC to have at this option's expiry – the reference the whole smile is centered on.
$k = ln (K / F)$ $k = ln (K / F)$
Source $latest_price.forward$ , $1 e 9$ -scaled $.
$F$ is the denominator of log-moneyness $k$ .
Oracle panel: The live on-chain oracle: current price plus the five SVI numbers that define BTC's whole implied-vol smile.
Data from GET $/oracles/:id/state$ .
Carries meta, $latest_price$ and $latest_svi$ .
Strike matrix ticks: Strike positions as percent moneyness with ATM in the middle, so you can read each cell's distance from the forward.
$k = ln (K / F)$ $k = ln (K / F)$
Labels read like $\pm 10/20%$ but $k$ is log-moneyness.
So $- 0.2$ is really $\approx - 18%$ actual strike offset.
SVI level — $a$: The baseline floor of total variance that shifts the whole smile up or down.
$w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$ $w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$
Decoded as on-chain $raw. a /1 e 9$ .
Shown $toExponential (3)$ , e.g. $2.767 e - 5$ .
SVI slope — $b$: How steep the smile's wings are – the angle between its two asymptotes.
$w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$ $w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$
Raw-SVI $b \geq 0$ ; decoded $raw. b /1 e 9$ .
Shown $toExponential (3)$ , e.g. $4.037 e - 4$ .
SVI shift — $m$: Horizontal position of the smile's low point relative to at-the-money.
$w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$ $w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$
Log-moneyness of the vertex shift; $w (k)$ centers on $(k - m)$ .
Source $(raw. m /1 e 9) \cdot sign$ ; shown $toFixed (5)$ .
Correlation — $ρ$: The smile's skew or tilt: how much richer downside (put) vol is than upside — the usual fear premium.
$w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$ $w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$
Range $ρ \in [- 1, 1]$ ; more negative = steeper put skew.
Live BTC $\approx - 0.94$ ; tinted warn when $ρ < 0$ .
SVI curvature — $σ$: How rounded vs sharp the bottom of the smile is – small is a sharp V, large a gentle bowl.
$w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$ $w (k) = a + b (ρ (k - m) + (k - m)^{2} + σ^{2})$
Source $σ > 0 = raw. σ /1 e 9$ ; shown $toFixed (5)$ .
Unrelated to the stress $σ$ .
Term structure: How ATM vol changes across expiries, showing an upward or inverted curve.
$σ_{ATM} (T) = w (0) / T$ $σ_{ATM} (T) = w (0) / T$
Readout = nearest $\to$ farthest $atmIV %$ over active oracles.
Each $impliedVol (0, params, T) \times 100$ ; tenor labels are near/far countdowns.
Feed age: How long ago the oracle last updated – the panel's freshness verdict.
$age = now - checkpoint_ts$ $age = now - checkpoint_ts$
Rendered compact: 's / m / h / d h' (a dead feed reads '4d 19h', not raw seconds).
Green if $< 120 s$ , warn otherwise.
Oracle status: Whether this expiry is still trading, has settled, or was just created.
Derived server-side from timestamps.
$settled_at \to$ settled (neutral), $activated_at \to$ active (green), else created (warn).
Vol by strike: A row of IV readings across strikes showing how vol rises out-of-the-money – the skew.
$IV (k) = w (k) / T$ $IV (k) = w (k) / T$
Each cell = $impliedVol (k, p, T) \times 100$ , integer %.
Sampled at $k \in [- 0.2, - 0.1, 0, 0.1, 0.2]$ .
Oracle ID: The on-chain object ID of this oracle, shortened for explorer lookup.
$truncateAddr (oracle_id, 6, 4) \to$ '0x8263…d6cf'.
One oracle = one rolling expiry.
Spot: Current BTC price the oracle reports – the headline number this panel verdicts on.
Source $latest_price.spot$ , $1 e 9$ -scaled.
$fmtPrice$ to '$69,315' (0 dp); per-digit pop on tick.
Spot sparkline: Mini trend of recent spot prices with the % move over the window, green up and red down.
$Δ_{spot} = \frac{last - first}{first} \times 100$ $Δ_{spot} = \frac{last - first}{first} \times 100$
Over the spot-history samples.
Shown $fmtSigned$ to 2 dp.

Desk · Vault

25 terms

APY: The vault's annualized LP return over its recorded history, projected to a full year — the headline yield for liquidity providers.
$APY = \frac{last / first - 1}{years} \times 100$ $APY = \frac{last / first - 1}{years} \times 100$
Share-price ratio over the perf window; simple, non-compounded.
Green when positive; null with $< 2$ data points.
Available liquidity: Free capital the vault can pay out or let LPs withdraw right now — funds not reserved against max payout, and the numerator of the stress buffer.
Source data.available_liquidity, scaled by $1 e 6$ .
LP net flow: Whether LPs are net depositing into or withdrawing from the vault, read off a running cumulative balance.
$net = \sum (supplies) - \sum (withdrawals)$ $net = \sum (supplies) - \sum (withdrawals)$
Source /lp/*; sparkline plots the cumulative series.
Reads $N$ deposits, $M$ withdrawals, net dUSDC in.
Max payout: The largest total the vault could owe if every open position pays out fully — a hard liability ceiling, since Predict payouts are capped.
Source data.total_max_payout, scaled by $1 e 6$ ; warn tone.
Denominator of max-payout util and the stress model.
Max-payout utilization: Tail-risk gauge: how close the worst-case total payout sits to exhausting the vault's capacity — the core LP-safety tail metric.
$util = \frac{total_max_payout}{capacity}$ $util = \frac{total_max_payout}{capacity}$
Range $0..1$ , shown $\times 100$ to 2 dp; warn-tone meter.
Modeled payout: The dUSDC the vault is modeled to owe traders in the simulated crash — a heuristic stress estimate, not position-level pricing.
$payout = maxPayout \cdot (mtmFrac + (1 - mtmFrac) (1 - e^{- σ /2.5}))$ $payout = maxPayout \cdot (mtmFrac + (1 - mtmFrac) (1 - e^{- σ /2.5}))$
Payout fraction ramps toward the cap as $σ$ grows.
PLP share price: What one PLP vault share is worth now — above $1.0000 means LPs are in profit.
$share = \frac{vault_value}{plp_total_supply}$ $share = \frac{vault_value}{plp_total_supply}$
Starts at $1.0$ at inception; header shows 4 dp.
Perf chart plots its time series.
PLP supply: Total PLP shares outstanding across all LPs — the denominator that turns vault value into a per-share price.
Source data.plp_total_supply, scaled by $1 e 6$ .
A share count, though rendered like a $ figure.
$σ$ move: How far BTC would drop in the simulated 1-day N-sigma crash — downside only.
$move = σ_{IV} \cdot 1/365.25 \cdot σ$ $move = σ_{IV} \cdot 1/365.25 \cdot σ$
Shown as $- (moveFrac \times 100)$ ; warn tone.
Stress ladder: Table previewing crash size and coverage at fixed 1 / 2 / 3 / 5 sigma so you can read the whole tail at a glance.
$Buffer = \frac{availLiq}{maxPayout \cdot payoutFrac}$ $Buffer = \frac{availLiq}{maxPayout \cdot payoutFrac}$
Move = modeled $- %$ drop per $σ$ row.
Color safe $\geq 2$ , warn $\geq 1$ , breach $< 1$ .
Stress $σ$: Number of standard deviations in the modeled 1-day BTC move — bigger means rarer and more extreme.
$moveFrac = σ_{IV} \cdot 1/365.25 \cdot σ$ $moveFrac = σ_{IV} \cdot 1/365.25 \cdot σ$
Unrelated to the SVI $σ$ curvature param.
Stress verdict: The flagship LP-safety headline: would the vault survive an N-sigma 1-day crash?
$buffer = \frac{availLiq}{stressedPayout}$ $buffer = \frac{availLiq}{stressedPayout}$
Reads 'Survives N $σ$ · $X \times$ buffer' (safe $\geq 2$ , warn $\geq 1$ ).
'Breach' when buffer $< 1$ .
Stressed spot: Where BTC's price would land after the simulated crash.
$stressedSpot = spot \cdot (1 - moveFrac)$ $stressedSpot = spot \cdot (1 - moveFrac)$
Scaled by $1 e 9$ , 0 dp; shown only when spot is known.
Total mark-to-market: The current net value of all open positions the vault holds against traders — what it would book if it closed everything now.
$mtmFrac = clamp (mtm / maxPayout, 0, 1)$ $mtmFrac = clamp (mtm / maxPayout, 0, 1)$
Source data.total_mtm, scaled by $1 e 6$ ; feeds the stress model.
Utilization: How much of the vault's capital is currently tied up backing open positions.
Source data.utilization, a $0..1$ fraction shown $\times 100$ to 2 dp.
Meter fills value $\times 100%$ .
Vault value: Total mark-to-market worth of the vault's assets right now — distinct from raw cash vault_balance.
Source data.vault_value: quote balance adjusted for open-position MtM.
Scaled by $1 e 6$ , fmtUsdCompact e.g. $1.01M.
dUSDC: Predict's Testnet-only quote stablecoin — not official Testnet USDC — in which every vault USD figure is denominated.
All vault amounts are dUSDC scaled by $1 e 6$ .
Net deposits: Lifetime LP money put in minus taken out — the cost basis of LP capital, read against vault value for aggregate LP PnL.
$net_deposits = total_supplied - total_withdrawn$ $net_deposits = total_supplied - total_withdrawn$
Source data.net_deposits, scaled by $1 e 6$ .
Sparkline: A tiny axis-less trend line — left end is the start of the window, right end is now — showing pure shape, not absolute level.
$y$ normalized to the series min/max.
Renders only with $\geq 2$ points; stroke = accent-brand.
Stress available: Free vault capital available to absorb the modeled payout — the numerator of the stress buffer.
$buffer = \frac{availLiq}{stressedPayout}$ $buffer = \frac{availLiq}{stressedPayout}$
Source data.available_liquidity, scaled by $1 e 6$ ; safe tone.
Stress slider: Drag to choose crash severity in standard deviations, driving the modeled move, stressed spot, payout and buffer.
Range $0..6 σ$ , step $0.5$ , default $5$ .
Readout shows $σ$ to 1 dp.
Supplied: All dUSDC ever deposited into the vault by LPs — the gross inflow side of net deposits.
Source data.total_supplied, scaled by $1 e 6$ ; green tone.
Vault panel: Answers 'is it safe for liquidity providers to underwrite this market?' — the PLP is the pooled dUSDC taking the other side of every binary trade.
PLP = Predict Liquidity Pool.
Withdrawn: All dUSDC ever pulled out of the vault by LPs — the gross outflow side of net deposits.
Source data.total_withdrawn, scaled by $1 e 6$ ; dim tone.
Performance chart: Area chart of PLP share price over time, the vault's running track record for LPs.
lightweight-charts; right price-axis at 4 dp (minMove $0.0001$ ).
Hover shows exact share price (4 dp) plus that point's date.

Desk · Live Flow & Settle

14 terms

Flow amount: Each feed row's dollar figure, with one column overloaded by sign and color to read open-cost, win, or worthless-loss at a glance.
$-$\text{cost}$ = premium paid to open (mint, dim).
$+$\text{payout}$ = winning settle (redeem, green); $$0.00$ = expired worthless (faint).
2 dp, U+2212 minus glyph.
Feed columns: What each column in the live trade feed means, ordered newest-first.
Time = UTC HH:MM:SS the event hit chain; Side = UP/DN direction.
Strike = settle price level (1e9-scaled $).
Prem/payout = premium out (mint, $-$ ) or payout in (redeem, $+$ ).
Taker win rate: The share of settled bets where the trader made money – under 50% means the PLP vault has the edge.
$winRate = \frac{wins}{total} \times 100$ $winRate = \frac{wins}{total} \times 100$
Wins = settled positions with $payout > 0$ .
Green when $< 50%$ , amber when $\geq 50%$ .
Up / Dn bar: A green-for-UP / red-for-DN bar showing the share of recent bets that are bullish vs bearish.
$upPct = \frac{upCount}{mints} \times 100$ $upPct = \frac{upCount}{mints} \times 100$
Count-weighted, not premium-weighted.
Defaults to 50% when there are no mints.
Up / Dn win split: How often bullish vs bearish bets won among settled positions – a big gap hints at directional mispricing of UP vs DN binaries.
$upWinRate, dnWinRate = \frac{wins}{count} \times 100$ $upWinRate, dnWinRate = \frac{wins}{count} \times 100$
Computed within each UP/DN subset separately.
Side glyph: The per-row direction marker: green UP for a binary call, red DN for a binary put.
UP wins if $settle \geq strike$ ; DN wins if $settle < strike$ .
Maps FlowItem.isUp.
Payouts: Total dollars paid to winning takers across all settled positions – the vault's gross loss to traders.
Sum of PositionRedeemed.payout over wins only ( $payout > 0$ ).
1e6-scaled.
Positioning: Crowd sentiment over the last N opened positions, i.e. which way recent traders are betting.
N = count of mint FlowItems in the window; redeems excluded.
Order-flow direction, not settled outcomes.
Your row: A faint blue tint marks trades from your own connected wallet.
Applied when your account address equals the mint trader or redeem owner.
Requires a connected Sui wallet.
Settlement count: How many positions have actually settled in the redeemed feed.
N = count of redeemed positions with $is_settled = true$ .
The block renders only when $N > 0$ .
Volume premium: Total premium paid to open all recent positions in the window – premium spent, not notional or payout exposure.
Sum of mint cost (raw 1e6 dUSDC).
fmtUsdCompact, e.g. $12.3K.
Desk panel: The 'can I act on it?' room – a live merged feed of position mints (order flow) and redeems (settlements) with aggregate positioning and settlement stats.
Polls /positions/minted and /redeemed every 6s.
Flow event count: How many recent trade events are loaded in the feed, capped at 80.
Merged mint+redeem FlowItems, sorted newest-first.
Shows 'Live' until the first events arrive.
Flow unreachable: Error state: the trade-flow source could not be reached – either /positions/minted or /redeemed errored.
Both endpoints are polled every 6s.

Command Strip, Status Bar & Selectors

12 terms

Indexer status: Health of the data indexer feeding every panel – OK means it has caught up to the chain.
From GET /status: green OK / amber Sync / red Offline.
The footer's authoritative health readout, same source as the live pill.
Feed lag: How many seconds behind the chain the feed is – small is fresh, large means stale prices.
$data.max_time_lag_seconds$ from GET /status, raw integer seconds.
Shown in both the live pill and the footer.
Live pill: Whether fresh on-chain data is flowing right now.
Dot from GET /status: red = Offline (error), green = Live (OK), amber = Sync.
The ping halo gates on the Tier-2 checkpoint push channel, distinct from the indexer poll.
Connect wallet: Connect a Sui wallet – once connected the pill is your account and gates the on-chain write actions.
Connected pill shows SuiNS name or truncateAddr (6/4) with a green dot; Testnet.
The write path for the vault (supply / withdraw) and trading (mint / redeem).
Market selector: Picker of BTC markets still open for trading.
Lists oracles with derived status = active (activated, not settled), sorted by expiry.
/oracles (~3.3k) is filtered client-side; Loading / 'No market' states disable selection.
UTC clock: Current time in UTC – the universal trading clock so expiries and lag read unambiguously.
ISO slice HH:MM:SS; renders '--:--:--' until hydrated.
Markets settle on UTC timestamps.
Command palette — $Cmd+K$: Keyboard shortcut that opens the quick command palette.
Dispatches a 'deepskew:cmdk' window event the CommandPalette listens for.
Trigger hidden below sm.
Footer links: Footer chrome – external links plus a network note, not metrics.
Links to DeepBook Predict docs and the DeepBook X account.
'Mainnet when Predict ships' notes that deepskew follows Predict to Mainnet.
Market symbol: The underlying asset of the market you're viewing, currently always Bitcoin.
Predict's Testnet lists only BTC rolling expiries.
Symbol is fixed while the expiry varies.
Network: Which Sui network the terminal reads – Testnet during the hackathon.
SUI_NETWORK constant.
Predict is Testnet-only today; deepskew follows it to Mainnet at launch.
Pill tones: Reusable colored capsule where color carries meaning – never decoration.
green = up, red = down/breach, cerulean = key/accent, amber = warn.
Maps to semantic tokens $-safe$ / $-breach$ / $-warn$ / $-accent-brand$ .
Wordmark: Product name and home link – clicking returns to the desk root.
DeepSkewMark glyph in cerulean.
The authority signal of the desk.

Vol Analytics

12 terms

Risk-neutral density: The full distribution of where the market prices BTC ending at expiry, recovered from one SVI smile — not a single number but the whole shape.
$q (k) = \frac{g ( k )}{2 π w ( k )} e^{- d_{-} (k)^{2} /2}, d_{-} = - \frac{k}{w} - \frac{w}{2}$ $q (k) = \frac{g ( k )}{2 π w ( k )} e^{- d_{-} (k)^{2} /2}, d_{-} = - \frac{k}{w} - \frac{w}{2}$
Breeden–Litzenberger via Gatheral; $g (k)$ is Durrleman's function.
Normalised to integrate to 1 over the log-moneyness grid.
Tail shape: Which side of the distribution carries more weight — a fat left tail means crash risk is priced richer than upside.
Compares the 10% and 90% move quantiles: $q_{90} + q_{10} < 0$ ⇒ fat left.
Warn tone for left-skew (downside), safe for right-skew.
Modal move: The single most-likely outcome the density implies, as a percentage move versus the forward.
$mode = e^{k^{*}} - 1, k^{*} = ar g k max q (k)$ $mode = e^{k^{*}} - 1, k^{*} = ar g k max q (k)$
Signed; near 0% when the smile is centred on the forward.
10–90% range: The market-priced range of moves: from the 10th to the 90th percentile of the implied distribution.
Both legs are $e^{k} - 1$ at the 10%/90% cumulative density.
Risk-neutral P(up): The probability the underlying finishes above the forward, read off the implied density.
$P (up) = \int_{k > 0} q (k) d k$ $P (up) = \int_{k > 0} q (k) d k$
P(down) = 1 − P(up); both integrate the same density.
Density non-negativity: A valid probability density can never go negative — which is exactly the butterfly-arbitrage condition on the smile.
$q (k) \geq 0 ⟺ g (k) \geq 0$ (Durrleman).
Green when the minimum sampled $g (k) \geq 0$ ; red flags an arb.
25Δ risk reversal: How much richer 25-delta calls are than 25-delta puts, in vol points — the headline directional-skew gauge.
$RR_{25} = IV (25Δ call) - IV (25Δ put)$ $RR_{25} = IV (25Δ call) - IV (25Δ put)$
Negative = puts bid (crash fear), the normal BTC regime; warn tone.
25Δ strikes solved off the smile by inverting $N (d_{1})$ .
25Δ butterfly: How much the 25-delta wings sit above the at-the-money vol — a measure of how convex (fat-tailed) the smile is.
$BF_{25} = \frac{1}{2} (IV_{c} + IV_{p}) - IV_{atm}$ $BF_{25} = \frac{1}{2} (IV_{c} + IV_{p}) - IV_{atm}$
In vol points; positive = wings richer than ATM.
Implied move: The 1-standard-deviation move the market prices into this expiry — the straddle-width breakeven.
$\pm σ_{ATM} T$ $\pm σ_{ATM} T$
Shown as a percentage of the forward.
Forward vol: The volatility priced for the window between two expiries — exposes event risk a flat ATM curve hides.
$σ_{fwd} = \frac{w _{2} - w _{1}}{T _{2} - T _{1}}$ $σ_{fwd} = \frac{w _{2} - w _{1}}{T _{2} - T _{1}}$
From ATM total variance $w = totalVariance (0)$ at each tenor.
Blank when variance isn't increasing (calendar-arb in the window).
ATM term structure: At-the-money vol across every live expiry — upward (contango) means longer-dated vol is richer; inverted (backwardation) means near-term stress.
$σ_{ATM} (T) = w (0) / T$ $σ_{ATM} (T) = w (0) / T$
Plotted nearest→farthest; verdict from back minus front IV.
Skew rotation: How fast the risk reversal is moving — whether the desk is actively re-skewing the smile (a signal) or holding it steady.
$slope = Δ RR_{25} /Δ t$ $slope = Δ RR_{25} /Δ t$
Vol points per hour over the cached SVI history.
Falling = steepening put skew; rising = calls richening.

Flow & Edge

14 terms

Vault edge per fill: How much more the vault charged than the binary is worth, per fill, in basis points — the premium-weighted average over the window.
$edge = paid - Φ (d_{2}), d_{2} = - \frac{k}{w} - \frac{w}{2}$ $edge = paid - Φ (d_{2}), d_{2} = - \frac{k}{w} - \frac{w}{2}$
Model-fair $Φ (d_{2})$ uses each fill's own oracle SVI + forward.
Green = vault overcharged (edge); red = takers got it cheap.
Vol-risk-premium captured: Cumulative signed dollar edge across the window — the realized premium the vault harvests for taking the other side.
$\sum_{i} (paid_{i} - fair_{i}) \cdot q_{i}$ $\sum_{i} (paid_{i} - fair_{i}) \cdot q_{i}$
Positive when the book of fills was sold above model fair.
Overpay rate: The share of fills where the taker paid more than the model-fair probability — i.e. the vault had the edge on that trade.
Count of fills with paid > fair, over all marked fills.
Marked fills: How many recent binary mints could be priced against a live SVI smile and scored for edge.
Only fills on active, quotable oracles are marked.
Settled / un-fitted oracles are skipped.
Paid vs fair: Every fill plotted by the price paid against its model-fair value, with the break-even diagonal.
Below the diagonal (paid > fair) = vault edge (green).
Dot size scales with contract quantity.
Settlement outcomes: Every settled position as a dot over time — UP lane above, DN below, green when the taker won, sized by payout.
From the redeemed feed where is_settled = true.
Calibration: Does the price takers paid match how often that side actually won? A well-priced market sits on the diagonal.
Bar = average paid probability per side; tick = realized win rate.
A gap means UP or DN binaries were systematically mis-priced.
Pending settlement: Oracles whose expiry has passed but which haven't settled yet — a 4th state the indexer's 3-state status can't express.
Active oracles with expiry < now.
Positions there can't be redeemed until settlement lands.
Net directional flow: Premium spent opening UP binaries minus premium spent on DN over the window — the crowd's net directional bet, by money not count.
$net = \sum cost_{↑} - \sum cost_{↓}$ $net = \sum cost_{↑} - \sum cost_{↓}$
Green when net-long upside; red when net-long downside.
Positioning by notional: The UP/DN split of opening premium weighted by dollars, so one whale counts more than ten dust trades — the real order-flow skew.
Contrast with the count-based OI histogram on the Desk.
Mint / redeem momentum: Premium paid to open new positions vs payouts flowing out on redemptions — whether risk is being opened or unwound.
Mint-heavy = risk-on; redeem-heavy = unwind.
Whale flow: The largest mints by premium in the window, with side, strike, moneyness and wallet — the size that actually moves the book.
Your own fills are tinted; sorted by premium.
Range flow: Vertical-spread (range) instruments that pay inside a strike band — a separate product line from the up/down binaries.
Premium and count of range mints over the window.
Band ladder: Where range positions cluster: each top band drawn from its lower to upper strike, brighter for more notional.
Bands aggregated by (lower, higher) strike pair.

Vault

17 terms

MtM concentration: How much of the vault's open risk rides on a single expiry — concentrated books blow up faster on one bad settlement.
$HHI = \sum_{i} s_{i}^{2}, s_{i} = \frac{liability _{i}}{\sum liability}$ $HHI = \sum_{i} s_{i}^{2}, s_{i} = \frac{liability _{i}}{\sum liability}$
HHI ≥ 0.5 = concentrated; ≤ 0.25 = diversified.
Top-expiry and top-3 shares of marked liability.
Marked liability: What the vault owes per expiry, valuing every open binary at its model-fair Φ(d₂) — the reconstructed contribution to total MtM.
Open quantity per (oracle, strike, side) marked at fair value.
Bar colour = net directional inventory (green up, red down).
MtM tie-out: The protocol's own total MtM, shown alongside the reconstructed liability as a sanity check — they should track, but the reconstruction is an estimate.
Reconstruction marks at client Φ(d₂); the contract uses cached matrix MtM.
Bounded by the indexer flow window (see 'partial book').
Partial book: The flow window hit the indexer's row limit, so the reconstructed exposure is a lower bound — older open positions may be missing.
Fetch capped at 1000 rows; treat totals as a floor.
Breach σ: The worst-direction σ-shock the vault survives before LP equity is wiped — repricing the whole book through the SVI smile.
$σ_{breach} = min {∣ n ∣ : vaultValue (n) \leq 0}$ $σ_{breach} = min {∣ n ∣ : vaultValue (n) \leq 0}$
Forward shocked $F e^{n σ_{T}}$ per oracle, smile held fixed.
Stressed value applies Δliability to the live vault value.
Scenario curve: Vault value across the full ±5σ shock range — where it crosses zero is the breach point.
x = σ multiple; y = stressed vault value in dUSDC.
Scenario ladder: The stressed vault value and buffer at fixed ±1/3/5σ, so you can read the whole tail at a glance.
Buffer = stressed value ÷ current value; red below zero.
Withdrawal limiter: The cap on how much LPs can pull out right now — the binding minimum of free liquidity and the limiter's budget.
$exit = min (available_liquidity, available_withdrawal)$ $exit = min (available_liquidity, available_withdrawal)$
Open = limiter not binding; Throttled = limiter caps exit; Drained ≈ 0.
Exit capacity: The limiter's current withdrawal budget against the vault's free liquidity — how much can leave before the throttle bites.
available_withdrawal from the vault summary (1e6).
Solvency floor: Cash left after the worst-case payout is reserved — the hard floor under withdrawals independent of the limiter.
$max (0, vault_balance - total_max_payout)$ $max (0, vault_balance - total_max_payout)$
Whichever of this and the limiter budget is smaller binds.
Max drawdown: The deepest peak-to-trough fall in PLP share price over the range — the loss an LP who bought the top would have worn.
$t min (\frac{share _{t}}{m a x _{s \leq t} share _{s}} - 1)$ $t min (\frac{share _{t}}{m a x _{s \leq t} share _{s}} - 1)$
0% means share price only ever made new highs.
Underwater curve: How far below its running peak the PLP share price sits at each point — flat at 0 on the highs, dipping when underwater.
Range selector replays 1D / 1W / 1M / 3M / ALL.
LP volatility: Annualised volatility of PLP share-price returns — how bumpy the LP ride has been.
$stdev (ln r) \cdot n / years$ $stdev (ln r) \cdot n / years$
From log share-price returns over the range.
Risk grade: A composite LP-safety read — GREEN/AMBER/RED from utilization, MtM concentration and exit capacity.
Worst of: utilization, max-payout utilization, top-expiry share, exit ratio.
Export the metrics as CSV or print the one-pager.
House rules: The live on-chain pricing and risk config the desk operates under — spread parameters and the trading-paused kill-switch.
From GET /config; null fields mean not yet published on-chain.
Spreads shown in basis points (1e9-scaled probability units).
Exposure ceiling: The maximum share of the vault that may back open positions before new mints are blocked — the protocol's hard risk gate.
RiskConfig.max_total_exposure_pct; null until published.
LP economics: The vault's running LP track record — share price, net deposits, and gross supplied/withdrawn — shown when the on-chain pricing/risk config isn't published yet.
All from the vault summary (share price + 1e6 amounts).

Accounts

7 terms

Manager leaderboard: Per-manager realized PnL attribution, ranked — the indexer's authoritative account roll-ups for the most active desks.
Cohort = highest-volume managers in the flow window.
Rank by realized PnL, account value or open exposure.
Equity curve: A desk's cumulative realized PnL over time — the shape of how it made (or lost) money.
From the manager's realized-PnL series.
Account blotter: The connected wallet's positions marked to market server-side, plus the account roll-up — including claimable winnings awaiting settlement.
Aggregated across all of the wallet's manager accounts.
Market managers: Desk-level activity across the whole venue — how many PredictManager accounts exist and how many traded in the recent window.
Total from the /managers directory; active = distinct managers in flow.
Vault net = premium in minus settled payouts out.
Account value: Total worth of the account: trading balance plus the mark value of open positions and redeemable winnings.
From the manager summary (1e6).
Open exposure: Capital tied up in open positions right now — the account's at-risk notional.
From the manager summary (1e6).
Redeemable: Winnings on settled positions that can be claimed now — value sitting idle until redeemed.
From the manager summary (1e6).

Ops / Health

4 terms

Feed staleness: How long since each oracle last published a price/SVI update. Past the 30s window the contract treats the feed as stale and mints revert.
Active = fresh; Stale = > 30s; Pending = expired, not yet settled.
Verdict is HALTED if trading is paused, else DEGRADED if any are bad.
Expired, not settled: This expiry has passed but the oracle isn't settled on-chain yet, so its price/SVI feed is frozen and the binary stays un-mintable until settlement clears it.
The label shows how long ago it expired.
Drops out of the feed list once the oracle settles.
Pipeline lag: Per-pipeline indexer lag — how far behind chain each event pipeline is, beyond the single aggregate lag the footer shows.
Time lag in seconds and checkpoint lag per pipeline.
Backfill pipelines are excluded.
Kill-switch & assets: The trading-paused flag and the set of enabled quote assets the market currently accepts.
From GET /config; paused halts all mints.

Cross-Venue

5 terms

Vol-arb spread: Predict's ATM vol minus Deribit's BTC DVOL index — when Predict trades rich the vault is selling vol above the reference.
≥ 2 vol points flags an actionable cross-venue spread.
Rich = sell Predict vol; cheap = buy it.
Deribit DVOL: Deribit's BTC volatility index — the market-standard reference for 30-day implied vol, pulled hourly.
External reference; CORS-open public API.
Vol-risk-premium: Implied vol minus trailing realized vol — what option sellers earn for bearing risk. Positive means options are rich, the LP's edge.
$VRP = σ_{implied} - σ_{realized}$ $VRP = σ_{implied} - σ_{realized}$
Sell-vol when positive; buy-vol when negative.
Realized vol: Annualised volatility of actual BTC returns over the trailing window, from Binance hourly closes.
$stdev (ln r) \cdot 24 \cdot 365.25$ $stdev (ln r) \cdot 24 \cdot 365.25$
24h rolling window of hourly log returns.
Implied vs realized: The two vol lines whose gap is the vol-risk-premium — implied (DVOL) over realized (Binance).
Shaded gap is the premium option sellers capture.