Banker Behavior Lab

Background net ETH flow after day 5 −3 ETH/day Staggered-exit sell delay 7 days Tokens sold after retirement 90% Starting Branches per Banker 3 Simulation horizon 26 days

Download this run (.json)

Observed in this run

Hypothesis generator—not a contract replica, exploit claim, or profitability forecast.

Expired-origin outflow days —

Final illustrative multiplier —

Withdrawn STANDARD still unsold —

Two exit clocks

Withdrawal pressure vs. market outflow

7-day withdrawal pressure Modeled ETH outflow

The shaded columns mark days when the two signals describe different stages of exit. The lines use separate vertical scales; compare timing, not height.

Decision log

What each Banker actually did

Show the illustrative decision rules

Controlled counterfactual

Same Branch retirements. Only sale timing changes.

Against a small positive background flow, every focal Branch retires at the same modeled step in every run. Only the later liquidation policy changes: same step, next step, six daily steps later, or hold.

— higher cumulative issuance is recorded for the focal Banker after waiting six daily steps than after selling in the retirement step.

The delayed run emits less in total than the same-step run, yet the focal Banker's cumulative total is higher. Against the next-step baseline, aggregate issuance reconverges but cumulative Banker totals do not.

Same step Next step Six daily steps Hold

Illustrative multiplier paths. They reconverge, but issuance accrued while the active Branch set differed is not recalculated later.

Decision boundary

The cumulative issuance difference is conditional, not universal.

SALE IMPACT

BACKGROUND NET ETH FLOW (ETH/STEP) →

Cell = focal issuance change0% to ≈8%

1,200-run stress test

Higher cumulative issuance is not automatically an economic advantage.

The seven-day Resolution window stays fixed at 168 hours while policy epochs, market direction, pool depth, cohort size and sale fraction vary.

Try the boundary

When does waiting actually change anything?

Same Branch-retirement hours in both runs. Only liquidation changes: retirement epoch versus exactly 144 hours later.

Policy epoch Background pool flow Exiting cohort's Branch share Initial pool depth Withdrawn STANDARD sold90%

Issuance received—

Final ETH exit value—

Exit value means ETH already realized plus a hypothetical liquidation of the focal Banker's remaining STANDARD at the shared horizon. Pool-depth options preserve the same normalized starting price.

Does the unknown epoch length kill the result?

Epoch	Issuance Δ	Exit value Δ	Issuance Δ across 1h boundary

The baseline direction survives at 1h, 6h, 12h and 24h, but its size changes. This is robustness within the tested policy family—not validation against deployed parameters.

Is the Banker actually better off?

Market flow	Issuance Δ	Exit value Δ	Market timing	Policy increment

“Market timing” freezes monetary policy; “policy increment” is the remaining ETH difference after subtracting that control. Most value change comes from waiting in a rising or falling market.

Baseline + sensitivity

Does the result depend on an alternate accounting interpretation?

W records	Policy ETH flow reads	144h issuance Δ	144h exit value Δ	169h sales outside W

Important: gross ETH entering on buys and leaving on sells is the whitepaper-stated policy input. After-fee flow is shown only as a conservative adverse control. Gross/net W remains an implementation question. Separately, 144 hours is inside the 168-hour window; the post-window control uses 169 hours.

Protocol-defense sensitivity

Can deeper POL and rate-limited buybacks absorb the timing effect?

Starting pool depth	Starting vault / pool ETH	144h issuance Δ	Exit value Δ	Delayed-path buybacks

Whitepaper rule used: each hourly contraction tick spends the smaller of 10% of vault balance and 0.2% of pool ETH reserve. The 70% contraction-fee share replenishes the modeled vault using an explicit current-sign settlement order. Pool-depth multiples proxy already-accumulated POL; they do not reconstruct the unpublished fee-to-POL execution path. The 20% vault case is a normalized stress point, not an expected launch balance.

Still illustrative: the pool is constant-product, initial reserves and defense balances are normalized, and redacted policy parameters, arbitrage, MEV and Charter transfers are not reconstructed.

Research boundary

Confirmed mechanics and illustrative assumptions are kept separate.

From the whitepaper

Branch retirement realizes accrued STANDARD.
Resolution pressure uses system withdrawals over a trailing seven-day window.
Half of the Resolution Fee burns; half rewards Bankers who stayed.
Pool trades create the net ETH-flow policy signal.
The hook counts gross ETH entering on buys and gross ETH leaving on sells per epoch; policy reads the last two completed epochs.
Protocol-owned liquidity cannot be withdrawn and only grows; Contraction Vault buybacks are rate-limited each hourly tick.

Illustrative in this model

The main behavior chart uses one day per step; the stress study advances hourly.
Fee floor, ceiling, saturation and multiplier steps are placeholders.
Retirement commits are processed sequentially.
A Banker retiring one Branch receives none of that commit's redistribution; the whitepaper does not specify partial-retirement eligibility.
The main chart uses a normalized index; the stress test uses a simplified constant-product pool, not a full Uniswap v4 implementation.
The design overview does not resolve whether W records gross retired STANDARD or net wallet mint. Both interpretations are tested.
Gross ETH flow is the whitepaper baseline; after-fee flow is only an adverse sensitivity control for execution-order uncertainty.
The hourly pool charges an illustrative ETH-denominated hook fee; 0–100 bps is used only as a sensitivity range.
Defense tests vary unknown starting vault balance and use pool-depth multiples as a proxy for accumulated POL.

Contribution boundary

What this experiment establishes—and what could change the result.

It does establish

In this model, retirement timing can be held fixed while sale timing alone changes the two-epoch policy path.
With the pro-rata Branch rule held fixed, different temporary multiplier paths can leave different cumulative Banker totals after aggregate policy reconverges.
More issuance is not the same as a better ETH-denominated outcome.

It does not establish

A deployed-protocol threshold, profit strategy, vulnerability or forecast.
The exact deployed effect after real vault balances, accumulated POL, arbitrage, MEV and redacted safeguards.
Who qualifies as a “Banker who stayed” after a partial retirement.

This result may change or disappear if

Branch retirement and market liquidation are atomically coupled, so withdrawn STANDARD cannot be sold on an independently chosen schedule.
Sale timing cannot move modeled ETH outflow between the completed epochs subsequently consumed by monetary policy.
Deployed settlement ordering or safeguards prevent temporary policy divergence from changing actual cumulative Banker issuance.