Config iEvery editable input for the model — globals, markets, KAT price, agents, and Monte Carlo distributions. Edits update all charts and the verdict live.
id iPerp market name. Free-form text — used as a key in the simulator. Markets cannot be added/removed mid-horizon in this version; whitelist curation is treated as static.bps iTaker fee rate in basis points (1 bp = 0.01%). Charged on volume traded in this market.baseline ADV iDaily USD volume on this market BEFORE the elasticity response. Realized volume per epoch grows above this depending on rebate ratio (see honest taker config).
i`linear` = interpolate between a start and end price (set both equal for a flat path). `custom` = edit each epoch individually.
KAT price process
Start price ($)real
Drift (annual log)assumed
Volatility (annual)real
Jump intensity (/yr)real
Jump mean (log)real
Jump volreal
Price floor ($)assumed
Taker leverage
Median leverage (×)proxyiMedian per-taker effective leverage (lognormal). Posted margin = notional / leverage, so higher leverage shrinks margin and the vault yield it earns — the audit proved this is the protocol's most load-bearing assumption. Calibrated from Katana Perps venue data (median max leverage = 1/initialMarginFraction × typical utilization).
Lognormal σreal
Minassumed
Maxreal
Takers / agentassumediEach agent represents a POPULATION of this many independent takers; its posted margin pools their leverage draws (harmonic mean). Larger ⇒ leverage uncertainty averages out within a trial rather than acting as one coin-flip per agent. 1 = legacy single-draw behavior.
Hold-time noise σproxyiσ of a per-agent, per-epoch multiplicative shock on posted margin, standing in for variation in position hold-time (longer holds = more margin-days earning yield). Mean-1, so it widens the margin spread without shifting its mean. 0 = off.
Stress deleverage βassumediStress-linked deleveraging β_L. In a bad epoch (the same systemic factor that crashes price) takers cut positions or get liquidated, shrinking posted margin — the vault-yield revenue base — exactly when the protocol is most exposed. Applied as a mean-1 multiplier on margin, so it adds the negative margin↔stress correlation without shifting the baseline mean. 0 = no deleveraging response. A stress knob, not a fitted constant.
vbUSDC vault yield
Base APYreal
APY noise (rel.)proxy
Vote-escrow dynamics (dynamic vKAT)
Lock-growth βassumed
Unlock hurdle APRassumed
Capitulation βassumed
Max change / epochassumed
Volume noise
Common factor σproxy
Idiosyncratic σproxy
Market βassumed
Systemic stress (tail risk)
Couples the KAT price and volume crashes so a bad epoch hits both at once. Shapes the worst-case percentiles — not the median break-even.
Systemic stress ρassumediSystemic-stress correlation ρ (Gaussian copula). One latent factor per epoch couples the KAT price diffusion and volume so a bad epoch co-moves them — the joint tail (low price + thin volume at once) that independent draws miss. Each factor's marginal is preserved, so calibration stays valid. ρ=0 = fully independent (legacy); price jumps stay idiosyncratic in v1. A stress knob, not a fitted constant.
Liquidation
The only perp friction that touches protocol revenue: stress-driven wipeouts shrink the posted-margin base and thus margin yield. Funding & slippage were removed — trader-side only, no effect on the go/no-go. β is the load-bearing knob; the two below are refinements.
Liquidation βassumediLiquidation sensitivity to systemic stress: fraction of posted margin liquidated = clamp(β × max(0, systemic) × leverage^exp, 0, 1). Only bites in stress epochs, rising with leverage — a fraction-of-pool wipeout coupled to the same factor that crashes price. The liquidated slice is a taker loss and is removed from the yield base. Default 0.01 (median leverage 5 under a 2σ stress ≈10% liquidated). 0 = liquidation off. A stress knob.
Liq leverage expassumediLeverage exponent in the liquidation fraction (β × max(0,systemic) × leverage^exp). 1 = liquidation scales linearly with leverage.
Liq penalty fracassumediFraction of the liquidated margin slice actually lost by the taker. 1 = the whole slice is a loss; <1 = partial recovery (a softer stress case).
Provenance: real = fit/sourced · proxy = comparable venue/asset · assumed = prior / stress knob. Hover a chip for the source.
Worth building? iA plain-language go/no-go synthesized from the deterministic (single-path) base case. It combines three build-or-not checks — Katana solvency (does revenue cover emissions?), attack resistance (does any critical break-condition fire?), and holder incentive (do external KAT holders earn enough to lock?). BUILD = all three pass; WITH CHANGES = at least one is a warning but none fail; NOT YET = at least one hard fail. This is one path; run the Monte Carlo risk profile and Break-even surface below to stress-test it across randomized conditions and the full price × volume region.
Not worth building yet
At least one check fails in the base case — the design needs a structural fix first.
Katana net / yr
-$68.0K
Coverage
77%
Holder APR
10.5%
✗
Katana stays solventiDoes protocol revenue (fees kept + margin yield + treasury rebate) cover KAT emissions on the median path? Pass ≥ 125% coverage (comfortable margin of safety); warn 100–125% (covered but thin); fail < 100% (Katana subsidizes the system).
Katana subsidizes the system
77% coverage
✗
Survives the attack harnessiDoes the deterministic verdict fire any critical break condition (insolvency, P&L spiral, persistent pump, or value capture with damage)? Pass = no triggers; warn = only soft warnings (vote concentration, low/underwater holder APR); fail = a critical economic break fires.
Critical break: Katana net negative, Attacker extraction with system damage, Persistent market pump
attacker net $18.3K
✓
Holders want to lock KATiDo external KAT holders earn enough return on their locked vKAT to make locking rational? Pass ≥ 5% APR; warn 0–5% (weak but positive); fail < 0% (holders lose money locking — the vote-escrow flywheel cannot bootstrap).
Locking earns a competitive return
10.5% holder APR
This is the deterministic base case— one path, recomputed live. It answers the go/no-go at today's assumptions. For confidence, run the Monte Carlo risk profile below (probability each check fails under randomized conditions) and the Break-even surface (the full KAT price × projected volume region where it stays solvent).
verdict — brokeniHeuristic checks for "broken" conditions: Katana net negative (incl. treasury), attacker extraction with system damage, persistent market pump, bribe-dominated voter rewards, vote concentration, external APR collapse, and Katana P&L sign flip. Wash trading alone is permitted by design per spec — only flagged when paired with a system harm.
- Katana net negative— Cumulative -$67.8K (incl. treasury rebate) over horizon
- Fees kept$103.0K
- Margin yield$95.5K
- Treasury rebate$29.7K
- Emissions paid$296.1K
- Attacker extraction with system damage— 4 attackers extract value while system is in deficit
- MarginParker(ZEC-USD)$18.4K
- VoteWhale (market in deficit)$7.5K
- Cartel(cartel_1) (market in deficit)$2.9K
- Cartel(cartel_2) (market in deficit)$2.9K
- Persistent market pump— 7 markets pumped ≥3 epochs (fees + margin yield < emissions)
- BTC-USD23 epochs · -$10.3K
- ETH-USD26 epochs · -$5.5K
- SOL-USD26 epochs · -$34.5K
- KAT-USD26 epochs · -$14.7K
- XRP-USD26 epochs · -$28.0K
- DOGE-USD26 epochs · -$19.8K
- TAO-USD26 epochs · -$15.1K
Breakeven analysis (annualized) iAnnualized check of whether Katana revenue (fees kept + margin yield + treasury rebate) covers KAT emissions, plus the single-lever moves that would close any gap.
fee-only: 35%iFees kept / Emissions. Shows what coverage looks like if vbUSDC yield + treasury contributions are stripped out. Useful for stress-testing dependency on the vault-bridge yield side.
77% coverediKatana revenue / Emissions. 100% = exact breakeven. Below = deficit, above = surplus.
Fees kept / yriAnnualized USDC fees retained by the protocol — total fees × (1 − voter split).
$103.3K
Margin yield / yriAnnualized vbUSDC vault-bridge yield generated by posted taker margin and retained by the protocol.
$95.8K
Treasury / yriAnnualized treasury net (voter fees + bribes received). Included in Katana-effective revenue per spec.
$29.8K
Emissions / yriAnnualized USD value of all KAT emissions distributed to takers.
$296.9K
Gap / yri(Fees kept + margin yield + treasury rebate) − emissions, annualized. Negative = Katana is in deficit, positive = surplus.
-$68.0K
To close the gap — pick one (others held constant)
- Grow volumeiVolume multiplier needed (other levers held constant) to scale Katana revenue up enough to cover emissions. Volume drives fees, margin yield, and treasury rebate together.from $1.54M/day→ $2.00M/day× 1.30
- Cut KAT emissionsiKAT emissions per epoch needed (other levers held constant) to bring emission cost below Katana revenue.from 2.00M/ep→ 1.54M/ep× 0.77
- Tilt voter splitiVoter fee split needed so the protocol retains enough fees to cover emissions. Marked impossible when (fees + margin yield) alone < emissions even at 0% to voters.from 50%→ 3%47 pp lower
- Raise fee bps (uniform)iMultiplier on every market's fee bps that would cover emissions given current margin yield + treasury rebate. Only scales fees, not margin or treasury revenue.from current→ × 1.6666% higher
Each row is isolated. Combine in practice. Volume growth scales fees, margin yield, and treasury rebate together. Fee-bps growth only scales fees — so it's usually less powerful when margin yield already carries part of the load.
Monte Carlo risk profile iRuns the model over thousands of randomized trials — leverage, volume, and price shocks all drawn from the Distributions panel — to show the probability and severity of each failure mode. The verdict above is a single path; this is the full distribution.
distributions over 2,000 randomized trialsTrials iNumber of randomized Monte Carlo runs. More trials shrink the Monte Carlo error on every probability and band, at the cost of runtime.
Seed iSeed for the random number generator. The same seed reproduces the exact same trials; change it to draw an independent sample.
Break-even surface — is it worth building? iThe go/no-go surface. Katana Perps (perps.katana.network) is live but young, so per-market volume over the modeling horizon is still a projection and the emission cost is dominated by KAT price. Each cell runs the full ensemble over a (KAT price × projected ADV) grid: green = solvent and attack-resistant, amber = solvent but fragile, red = Katana subsidizes. The green frontier is the volume the venue must attract at a given KAT price.
Katana Perps (perps.katana.network) is live but young, so per-market volume over the modeling horizon is still a projection and the emission cost is dominated by KAT price. Each cell runs the full ensemble over a (KAT price × projected ADV) grid. Green= comfortably solvent (median net > 0 and P(net<0) ≤ 20%); amber = solvent on the median but a meaningful subsidy tail; red = Katana subsidizes (median net ≤ 0). The green frontier is the volume the venue must attract at a given KAT price. (P(critical break) is in the tooltip / rule table.)
Frontier surface — gauge × trader-acquisition efficiency iThe go/no-go surface. Katana Perps (perps.katana.network) is live but young, so per-market volume over the modeling horizon is still a projection and the emission cost is dominated by KAT price. Each cell runs the full ensemble over a (KAT price × projected ADV) grid: green = solvent and attack-resistant, amber = solvent but fragile, red = Katana subsidizes. The green frontier is the volume the venue must attract at a given KAT price.
No emission cap — the two dials that move break-even. Rows are the gauge (KAT/epoch, the subsidy size); columns are incentive efficiency (trading $ each $1 of emission attracts, calibrated at 2,125). Each cell is a full ensemble. Efficiency (→) buys the median; the gauge (↓) sets the tail, because emissions are KAT-priced. Green / amber / redfollow the selected metric; the cyan outline is the P(net<0) ≤ 20% frontier.
Set your beliefs in the Distributions panel, then Run to see the probability of each failure mode — with confidence bands and data provenance. The deterministic verdict above is a single path; this is the full distribution.
Details iSupporting breakdowns behind the headline verdict: annualized summary cards, cumulative P&L, per-market volume, and the subsidy heatmap.
Katana net (annualized)iAnnualized Katana-effective net USD: USDC fees retained + vbUSDC margin yield + treasury voter rebate − USD value of KAT emissions paid out. Treasury rewards are included because the treasury is owned by Katana per spec. Negative = Katana subsidizes the system.
-$68.0K
fees kept $103.3K + margin yield $95.8K + treasury rebate $29.8K − emissions $296.9K
External holder APRiAnnualized yield for honest vKAT holders. (Voter fee share + bribes received) / (KAT locked × starting KAT price) / years. Margin yield is NOT shared with voters.
10.5%
on 80.0M vKAT
Attacker net totaliCumulative net USD across all attacker agents over the horizon. Positive = they extracted value. Per spec, attacker profit alone isn't a break condition — it's only flagged when paired with Katana net loss or persistent market deficit.
$18.3K
8 agents
Volume / epoch (avg)iAverage traded volume per epoch across all markets — the top-of-funnel flow that drives fees (volume × fee bps) and, via posted margin, the vault yield. Headline is per-epoch; the sub shows the implied daily ADV (÷ days per epoch) and the cumulative horizon total.
$20.86M
$1.49M/day · $542.27M total
Fees to vKAT / epochiUSDC trading fees routed to vKAT voters, i.e. the voterFeeSplit share of fees on markets that received votes (bribes excluded). Shown per epoch (total ÷ epochs) with the horizon total. This is the fee stream vKAT lockers actually earn — margin yield is NOT shared with voters.
$4.0K
$103.0K over 26 epochs
New vKAT stakeiEstimated NEW vKAT stake the integration attracts. The voter-fee yield makes locking attractive, so KAT flows in until the locker APR compresses to the opportunity-cost rate (Target lock APR): supported vKAT = annual voter rewards ÷ (Target lock APR × KAT price), and new stake = that minus what is already locked. Inversely proportional to Target lock APR — a display estimate, not part of the sim economics. Bounded at 0 when the current APR is already at/below target.
0 KAT
APR 10.5% already ≤ 10.0% target
Trader incentivesiTotal USD value of KAT emissions paid out to takers as trading rebates over the horizon, with the per-epoch average. Emissions are distributed only to agents who generated revenue (fees + margin yield) pro-rata — this is the cost of incentivizing trading. Priced at each epoch's KAT price.
$296.1K
$11.4K/epoch · 52.00M KAT total
Gross revenueiGross protocol revenue generated over the horizon BEFORE emissions: all USDC fees collected + vbUSDC margin yield retained. This is top-line revenue; the Katana-net card nets emissions out of it.
$301.6K
fees $206.1K + margin yield $95.5K
Margin held (avg)iAverage taker margin (USDC collateral) held on Katana per epoch — the capital base that produced the vbUSDC margin yield. This is the surviving (post-liquidation) open margin summed across markets and averaged over epochs, i.e. a stock (level held), not a flow. Yield ≈ margin held × vbUSDC APY × horizon-years.
$2.98M
3.2% eff. APY → $95.5K margin yield
Cumulative P&L (USD)iRunning USD P&L over time for each entity. Lines are cumulative (epoch 0 + epoch 1 + ...). Protocol line includes fees kept and margin yield.
Realized daily volume per market (stacked)iStacked daily volume per market over time. Flat is expected: with emissions, KAT price and organic baselines all held constant across epochs, the system hits a stationary equilibrium within ~1 epoch. To see how volume RESPONDS to incentives, see the emissions→volume chart (a comparative-static sweep), not this time series.
Volume vs KAT emissions (with efficiency cap)iComparative-static sweep: re-runs the sim across a range of KAT emissions/epoch and plots steady-state daily volume against it. The dashed line is the uncapped (linear) response — each $1 of emission buys volumePerEmissionUSD $ of volume forever. The solid line is the modeled response with the efficiency cap (Michaelis–Menten): incremental volume saturates toward volumeCapMultiple × organic ADV, so $/$ efficiency (right axis) decays as emissions rise. The gap between the two lines is what the diminishing-returns cap costs. Marker = current emissions.
cap = ×31 organic ADV
Each point re-runs the whole sim at that emission level. Dashed = uncapped (linear, each $1 of emission buys the same volume forever). Solid = modeled with diminishing returns. The gap is what the efficiency cap costs; the dotted line (right axis) is $-volume bought per $1 emitted.
Per-market summary (cumulative revenue, emissions, net subsidy)iCumulative revenue, emissions, and net subsidy for each market over the horizon. Negative net subsidy = that market drains the protocol.
| market | avg vote iAverage share of total vote weight directed to this market across all epochs. | realized ADV iAverage daily volume over the last 3 epochs — the elasticity-driven realized volume. | cumulative fees iCumulative USDC fees paid by takers in this market. | cumulative margin yield iCumulative vbUSDC vault-bridge yield generated by posted margin on this market and retained by the protocol. | cumulative emissions $ iCumulative USD value of KAT emissions distributed to takers of this market. | cumulative net subsidy i(Fees kept + margin yield) − emissions $ on this market. Negative = market drains the protocol. Per spec, single-epoch deficits are expected; ≥3 consecutive deficit epochs trigger a verdict. | |
|---|---|---|---|---|---|---|---|
| BTC-USD | 17.7% | $443.1K/d | $57.8K | $13.4K | $52.5K | -$10.3K | |
| ETH-USD | 8.9% | $210.0K/d | $28.4K | $6.6K | $26.3K | -$5.5K | |
| HYPE-USD | 0.3% | $0/d | $0 | $0 | $822 | -$822 | |
| SOL-USD | 18.5% | $209.0K/d | $27.6K | $6.4K | $54.7K | -$34.5K | |
| KAT-USD | 12.2% | $218.1K/d | $29.4K | $6.8K | $36.2K | -$14.7K | |
| XRP-USD | 18.6% | $272.8K/d | $37.0K | $8.6K | $55.0K | -$28.0K | |
| DOGE-USD | 8.0% | $40.0K/d | $5.5K | $1.3K | $23.8K | -$19.8K | |
| TAO-USD | 8.6% | $104.1K/d | $14.2K | $3.3K | $25.5K | -$15.1K | |
| ZEC-USD | 7.2% | $46.9K/d | $6.3K | $49.3K | $21.2K | $31.3K |