The Standard Deviation: a thesis

Standard Reserve is an onchain monetary system that pays the banks running on it in its own currency, $STANDARD. Its rules set four games: a fixed daily issue shared by every branch, a daily auction for the right to open one, a payout that can only be taken by closing branches, and an exit fee that punishes crowds and pays those who stay. Played alone, the standard strategy is weak. This paper argues that pooling is the profitable deviation from it, sets out the policies by which The Standard Deviation acquires, retires and repays, and shows what the pool is modelled to earn in $STANDARD and what that becomes in ETH at any $STANDARD price.

Standard Reserve describes itself as a closed monetary economy: one currency, $STANDARD; one market, where ETH trades against it; one signal, the net flow of ETH through that market; and one authority, a central bank written as immutable code. No company can vote on it, pause it or upgrade it.

The people who run it are bankers. A banker holds a charter, and a charter operates between one and ten branches. One thousand founding charters were sold at launch. They cannot yet be transferred, and the daily auction for new ones has not started.

Three rules matter to anyone who owns a bank.

1.1Issuance is one pie, split by branch

Every day the central bank issues 700,000 $STANDARD multiplied by a policy rate m, and divides it equally among all N branches in existence, second by second. One branch earns, per day:

1.2Money exists only when a branch is closed

Earnings accrue as a balance inside the bank, not as tokens in a wallet. $STANDARD is minted at one moment only: when a banker retires branches, which the Reserve’s own app calls dissolving them. Retiring k of a charter’s n branches releases

where B is the charter’s balance and φ the exit fee of section 2.4. The retired branches are destroyed. Retiring the last one destroys the charter.

1.3Everything else burns or taxes

A new branch needs a licence, bought at a daily auction in $STANDARD that is burned. Trades are taxed 2% on the way in and 3% on the way out. Exits pay the fee in (2), and thirty days of inactivity costs a bank 70% of its balance. Half of each of those penalties is burned and half is paid to the banks that remain.

THE MARKET

ETH ⇄ $STANDARD

One pool. 2% tax in, 3% out. It measures the net flow of ETH.

→

THE CENTRAL BANK

700,000 × m a day

Immutable code. It sets m from the net flow, between 0.2 and 1.25.

→

THE BANKS

1,000 charters, N branches

Each branch is credited an equal share, second by second, as a balance inside its bank.

OUT BY LICENCE

A new branch

Bought in $STANDARD at the daily auction, 100 a day. All of it is burned.

OUT BY RETIREMENT

Minted $STANDARD

Close k of n branches for k/n of the balance, less the exit fee. Half the fee is burned, half goes to the banks that stay.

OUT BY SILENCE

70% forfeited

After thirty days without acting, anyone may close the bank. Half is burned, half goes to the banks that stay.

At block 65,629,152 the Reserve had 1,000 charters, 1,299 branches and 95,131,712 $STANDARD in circulation.

Those rules set four games. Each has a formula, and each formula has a best response.

2.1One pie

By (1), every new branch anywhere makes every existing branch earn less. With 1,299 branches a branch earns about 539 $STANDARD a day. At 5,500 it earns 127, and at the 10,000 that the founding charters can hold, 70.

Entry is rationed three ways: a fixed number of seats, ten slots to a seat, and 100 licences a day for the whole system, three per charter. The pie cannot be cut faster than 100 slices a day, so a branch opened early collects the large early slices.

$STANDARD A DAY, PER BRANCHBRANCHES IN THE RESERVE N020040060002,0004,0006,0008,00010,000TODAY1,299 branches: 539 a day each.12770

2.2The licence

Licences sell by Dutch auction. The price opens at twice the previous day’s close and decays towards a floor of two days of one branch’s issuance:

Every auction so far has sold out within hours, far above the floor. The useful measure of the price is the number of days a branch takes to earn its own licence:

The first three auctions cleared at about 19, 33 and 24 days. At 24 days a branch returns its cost about fifteen times a year before dilution and costs, which is why they sell out.

2.3The payout problem

By (2), earnings can be taken only by destroying the branches that earn them, and a destroyed branch must be bought again at the day’s price. Measure a charter’s balance in days of its current earnings, t = B / (n × y). The cost of re-opening, as a share of what a retirement releases, is then

The number of branches retired cancels out. The cost of a payout depends only on how long the balance has been building. Paying out early is expensive: while t is below τ, a payout costs more than it releases.

A charter with a single branch cannot take anything without destroying itself. After three auctions, 861 of the 1,000 founding charters had never bought a second.

COST OF RE-OPENING, AS A SHARE OF THE RELEASE cBALANCE, IN DAYS OF CURRENT EARNINGS t050%100%150%200%050100150200TOO EARLYIn the hatched band a payout costs more than it releases.THE POOL’S RULE25% or less: from about 97 days.

2.4The exit

A banker leaves by retiring branches, and the Reserve charges for it. The charge depends on how many others are leaving at the same time. Exit pressure is the share of all bank balances withdrawn over the trailing seven days:

where W is the $STANDARD withdrawn in the last seven days and D is the $STANDARD still held in bank balances.

The fee rises with the square of that pressure, from 2% in calm to 60% in a run:

Half of every fee is burned. The other half is paid to the banks that stayed.

FEE φEXIT PRESSURE P020%40%60%02%4%6%8%10%12%7%16%CALMThe pool retires branches only in this band, where the fee is at most 5%.A RUNFrom 10% pressure the fee is 60%.

In an ordinary bank run, leaving first is the winning move, so everyone runs. Here the order is reversed.

	While others stay	While others run
You leave	You keep 98% of your balance and give up the branch’s future issue.	You keep as little as 40% and give up the same future issue.
You stay	You keep earning your share of the issue.	Your share of the issue grows as branches close, and you collect half of every fee the leavers pay.

The more bankers leave, the more leaving costs and the more staying pays. For a banker with no need to leave, staying is the dominant strategy. The only losing position is being forced to exit in a crowd.

2.5The central bank’s bias

The policy rate m starts at 1 and lives between 0.2 and 1.25. It is reset every epoch, currently three days, from the net flow of ETH over the last two:

Net flow over the last two epochs	The rate becomes
Negative	max(0.2, m − 0.15), at once
Positive, two epochs running	min(1.25, m + 0.10)
Anything else	m, unchanged

Cuts are immediate and raises must be earned. Capital flowing in lifts what every branch earns by up to a quarter. Capital flowing out cuts it quickly, by up to four fifths, which protects the currency at the bankers’ expense.

In game theory a deviation is a player’s move away from the strategy everyone else is playing, and a position is stable only when no deviation pays. The standard strategy in the Reserve is the solo banker: one charter, one branch, run by hand. Section 02 shows where it is weak.

A branch is productive capacity of an unusual kind: scarce, destructive to realise, demanding to operate, and worth more the longer it is held. Standard Reserve rewards scale, patience and liveness, and a banker alone is short of all three.

	Alone	Pooled
Access	One of 1,000 seats, none for sale.	A share of a seat, for anyone admitted.
Capacity	One branch. A second takes about 24 days of earnings.	Ten branches filled within days, at the early yields of (1).
Taking profit	Impossible with one branch without destroying the charter.	A tenth at a time, only when (5) says it is cheap.
Exit fee	Paid whenever the owner needs the money.	Retirements only in calm, never in a run.
Silence	One missed month costs 70%.	Never silent, and on the receiving side of that penalty.
Leaving	Closing every branch burns the seat.	Members leave through the pool. The seat survives.

Each row is a best response to one formula. Access and capacity answer (1): the early slices are the large ones, and only capital that arrives together can fill ten slots at once. Patience answers (5): a payout is cheap only for a bank that can wait. Calm answers (7): the fee is a transfer from those who must leave to those who need not. A pool is built to be the bank that need not.

The last row matters most. A solo banker who wants out must close every branch, pay the fee of the day and burn the seat. A member of the pool leaves by redeeming units in a weekly round. Every request in a round is met at one price and in the same proportion, from $STANDARD the pool has already realised, so inside the pool there is nothing to gain by running first. The Reserve itself still lets the first banker out at the lowest fee. The pool’s own rounds give no one that advantage. The bank itself never exits.

Pooling is therefore not a wrapper around the standard strategy. It turns scarce banking capacity into a strategy that scales and is operated every day, and under the Reserve’s own rules it is the better one.

The pool begins with founder-owned charter #729 and funds its other nine branches. It grows by contribution rounds, opened as the Reserve makes charters available. Its policies follow from section 02. They are published, and they change only with seven days’ public notice.

4.1Acquisition

Licences. Up to three a day on each charter until it holds ten branches, and only at or below 25 days of one branch’s current issuance, τ ≤ 25.

Charters. None can be bought today. When the Reserve sells or frees them, the pool buys only with new ETH raised in a contribution round, and each round first publishes its mandate: what it will buy, the most it will pay, and the modelled payback. Every charter acquired is then filled to ten branches.

Rounds. Every round joins the same pool. The principle is that a newcomer pays what capacity costs on the day they enter: units are issued against the pool’s $STANDARD, its releasable balances, and its branches and charters marked from recent auction prices, which is what the newcomer’s own ETH is about to pay for them. A branch is not liquid inventory. Its licence is burned, it cannot be sold, and its balance is released only by closing it, so what a member leaving is paid is a different and lower figure than what a member entering pays. The valuation policy that sets both, including how recent auction prices are averaged and how a charter bought in ETH is carried, will be published before public contributions open.

The pool never sells $STANDARD to buy a charter, and never opens a round to fund withdrawals.

4.2Dissolution

Retiring branches is the only way to be paid, so the pool does it by rule, charter by charter.

1. Each charter pays for itself first. Nothing is retired on a charter until a single retirement, of every branch but the ones that must stay, would return what the pool paid for that charter and its licences, swap costs included. That retirement is made on the first day it can be. On charter #729 it is eight of the pool’s nine branches.
2. Then regular payouts. One branch is retired whenever re-opening it costs at most 25% of what the retirement releases, c ≤ 0.25 by (5), and it is re-opened under 4.1.
3. Only in calm. No retirement while the exit fee is above 5%.
4. Members waiting to leave. If a withdrawal request has waited 90 days, the pool retires the fewest branches that pay it, provided the exit fee is at most 10%.
5. Never. The pool never retires the founder’s branch or a charter’s last branch, and retires nothing above a 10% exit fee.

Policy v1	17 September 2026
Licence price ceiling	25 days of one branch’s issuance
Licences a day	3 per charter, to ten branches
A charter’s first retirement	when it returns the charter’s cost
Regular retirement	1 branch, when c ≤ 25%
Exit fee ceiling	5% (10% for rule 4)
Waiting-member trigger	90 days
Notice of any change	7 days

4.3Repayments

Everything a retirement brings in, from any charter, is paid to every member in proportion to their units, in $STANDARD, whichever round they joined in. A payout does not redeem units: ownership stays. What a charter returns up to its own cost carries no fee. The pool charges one fee, 10% of realised profit: what retirements bring in above the cost of the branches retired, after costs and earlier losses are recovered. It charges nothing to enter and nothing to leave.

Leaving is separate. A member queues units and is paid in weekly withdrawal rounds at that week’s unit value, from realised $STANDARD, with a price floor the member sets. Unpaid units stay invested and keep receiving payouts. There is no promised date and no guaranteed amount.

PAID OUT BY ONE CHARTER, × WHAT ITS BRANCHES COSTDAYS SINCE ITS FIRST LICENCE01×2×3×090180270365IT PAYS FOR ITSELFDay 79: one retirement of eight branches returns what the branches cost.THEN REGULAR PAYOUTSOne branch at a time from day 186, each about 11% of the cost.

4.4Growth

The Reserve is not selling charters yet, so the pool’s first year is one charter’s year: nothing for about eleven weeks, one large payout, a pause while the balance rebuilds, then a payout every fortnight.

Growth changes that pattern. Rounds open as charters come up for sale, each with its mandate, and every charter bought runs the same cycle from a different day. Their payouts overlap. A bank of charters at different stages pays its members something most weeks, which is how the pool’s two aims, capital back first and frequent payouts, are met together.

DAYS SINCE THE FIRST ROUND090180270360450540CHARTER #729A SECOND CHARTERA THIRDA FOURTHPAID TO EVERY MEMBERIT PAYS FOR ITSELFThe tall tick: a charter’s first retirement returns its cost.

4.5The founder

Charter #729 was bought by the founder in the founding sale, for 1.2397 ETH, and is held in the founder’s wallet, which is also the pool’s custodian. The review of the pool’s code was internal. The arrangement between the founder and the pool is set out here in full, as a related-party statement.

Item	Terms
Founder entitlement	The balance an untouched one-branch founding charter would hold, measured from the mint. Every branch earns the same, so any untouched charter shows it.
Pool entitlement	Everything the charter holds above that benchmark: the nine branches the pool funds, and what they earn.
Fee	10% of realised profit: what retirements bring in above the cost of the branches retired, after recorded costs and earlier losses. It vests for 30 days and is clawed back by later losses while unclaimed.
Operating costs	Transaction costs are borne by the operator, not charged to members.
Purchase of #729	Deferred: no payment and no liability in version 1. The price will reference the median of arm’s-length sales of founding charters over a stated window, once the Reserve makes them transferable, normalised to a one-branch charter and excluding the branches and balances the pool funded. If too few sales occur, the terms are revisited rather than estimated.
Where repayment ranks	Paid only from a disclosed share of realised surplus, after funded claims and essential reserves; never from contributed principal or reserves. No date is promised.
After repayment	The charter is committed to the pool and transferred into its custody when the Reserve allows transfers. Until then the founder’s wallet remains its holder and operator.

There are two figures, and they answer different questions.

5.1Pool APY: what the pool earns, in $STANDARD

Pool APY is the modelled return on capital deployed in one round at that round’s prices. The figures here are for the first round: nine branches on charter #729 at today’s prices. Every later round publishes its own in its mandate, because the licence price, the branch count and the charter’s price will differ. A later round also buys a share of balances the pool has already earned, which come straight back and earn nothing in between, so its figure on the whole contribution sits a little below the figure for new branches alone.

The model takes the latest auction’s licence price and the Reserve’s emission rate, runs the policy of 4.2 day by day for a year, and counts the 3% cost of swapping ETH for $STANDARD as money put in.

Two things the pool does not control decide the figure: how many branches the Reserve ends with, and what a licence costs. Only a charter’s holder can buy its licences, ten to a charter, so 10,000 branches is the ceiling for as long as the Reserve sells no new charters. After three auctions, 139 of the 1,000 charters had expanded. The table runs the model across both, with the Reserve adding 100 branches a day until the ceiling and the policy rate stepping up to 1.1 on 21 September, as measured inflows imply.

Licences clear at	3,000 branches	5,500	7,500	10,000
17 days	+660%day 38	+360%day 47	+270%day 47	+210%day 47
24 days, the latest	+440%day 55	+230%day 79	+170%day 89	+130%day 95
33 days	+285%day 80	+140%day 123	+100%day 149	+70%day 175

Each cell is Pool APY over twelve months after fees, then the day the charter pays for itself.

The reference case used in the rest of this paper is 5,500 branches at the latest auction’s 24 days: +230%. It is a scenario, not an equilibrium. At 100 licences a day the Reserve reaches 5,500 branches in about six weeks and 10,000 in about three months, so the reference case holds only if most charters never fill.

The policy rate moves every cell. At 5,500 branches and 24 days:

Net flow of ETH	The rate goes to	Pool APY
Sustained inflow	1.25	+270%
As measured so far	1.1	+230%
None	1.0	+205%
One month in, a cut	0.7	+135%
Sustained outflow	0.2	−15%

Pool APY does not depend on the price of $STANDARD. Licences are paid in $STANDARD, branches earn $STANDARD, and the pace of new branches is capped by rule, not by price. What moves it is in the two tables above, and in the levers the Reserve’s owner holds, which section 08 sets out.

5.2Contribution APY: what your ETH becomes

Members contribute ETH and are paid in $STANDARD. What that is worth in ETH depends on where $STANDARD trades when it is sold:

where x is the $STANDARD price as a multiple of the price at entry, and 0.96 is what remains after the Reserve’s 3% sell tax and the 1% pool fee on the way back to ETH.

$STANDARD price	Contribution APY	1 ETH becomes
0.5×	+60%	1.6 ETH
1×, today	+215%	3.2 ETH
2×	+535%	6.3 ETH
3×	+850%	9.5 ETH
5×	+1,485%	15.8 ETH
10×	+3,070%	31.7 ETH

Break-even is x = 0.32. In this model a contributor gets their ETH back even if $STANDARD loses two thirds of its value over the year. That cushion is thinner than it looks, because a falling price usually means capital leaving, and by section 2.5 capital leaving cuts what every branch earns.

5.3Against simply holding $STANDARD

A holder of $STANDARD earns nothing from the Reserve. A contributor with the same ETH ends the modelled year with

times as many $STANDARD as the holder, whatever the price does, since both sell into the same market. Pool APY is the edge over holding the currency. The price decides what both are worth.

ETH AFTER TWELVE MONTHS, FROM 1 ETH$STANDARD PRICE, × TODAY’S x05101501×2×3×4×5×THROUGH THE POOL3.2 ETH at today’s price.HOLDING $STANDARD0.93 ETH at today’s price.

None of these figures is a forecast or a promise. They are a model at one block, and a contributor can lose money.

The pool is paid in $STANDARD, measures itself in $STANDARD and never converts back to ETH. Its view of the currency rests on how supply behaves.

Supply can be minted only by closing branches, and the cap can only fall:

where M is minted on withdrawal, B burned, C deposited back into banks and R removed from bank balances, mostly by licences.

So far the burn is winning. By block 65,629,152 about 1.8 million $STANDARD had been issued to bank balances, while 4.44 million had come off the cap and circulating supply had fallen by 4.87 million, to 95,131,712. Almost all of it was bankers buying $STANDARD in the market and burning it for branches.

From licences alone, the Reserve burns more than it issues whenever

one hundred licences a day, each costing τ days of a branch’s 1/N share of the issue.

On 17 September that was 2,377 against 1,299. If τ stays where it is, issuance overtakes the licence burn once the Reserve passes about 2,400 branches. What pushes the other way is demand. There are only 100 licences a day however many bankers want them, so more charters means more bidders and a higher τ. Charter sales, when they start, add bidders, and the ETH they raise goes to the Reserve’s reserves, its permanent liquidity and its buybacks. Behind the licence burn sit the other sinks: half of every exit fee and every penalty for silence, the buybacks, and the tax on every trade.

LICENCE PRICE, IN DAYS OF ONE BRANCH’S YIELD τBRANCHES IN THE RESERVE N025507510002,0004,0006,0008,00010,000BURN EXCEEDS ISSUANCEAbove the line, 100 licences a day destroy more $STANDARD than the Reserve issues.THE FIRST THREE AUCTIONS19, 33, 24 days of yield.ISSUANCE EXCEEDS BURNBelow it, new supply builds in bank balances.

Issuance is also finite. The budget is 900 million $STANDARD. At m = 1 it lasts about three and a half years, after which banks are paid from recycled fees alone. The banks worth owning then are the ones that were built early and stayed.

That is the pool’s position. Every ETH contributed is a market purchase of $STANDARD that is burned into permanent capacity. The pool retires branches only in calm, so it adds nothing to a run and is paid by those who join one. It expects a currency with a falling cap, an asymmetric central bank and a growing reserve behind it to be worth more as the system grows. It does not need that to be true to out-earn holding the currency, by (10), and it states the downside in (9).

A monetary system whose yield is reserved for a thousand seat-holders needs intermediaries, as every banking system has. The Standard Deviation is built to be four things.

An access layer. Capital that lacks a charter, enough scale, enough patience or the means to operate a bank every day has one alternative today, which is holding the currency. The pool gives it a strategy that is otherwise hard or impossible to run well.

Patient capital. The Reserve pays those who stay and taxes those who run. A pool with published rules, no need to exit and no silent months is the counterparty that design rewards, and a stabilising one: its purchases burn supply and its retirements happen only in calm.

A buyer of seats. When charters become transferable, selling a seat is an exit with no sell pressure on $STANDARD. The pool intends to be a standing bidder, by mandate, and to fill every seat it buys.

A protocol. $STANDEV is The Standard Deviation’s own proposed token. Its rights, its issuance and any link to the 10% fee are not yet defined. Pool units are not $STANDEV, and neither is a claim on the other.

• Custody. The founder’s wallet holds the charter and can override the pool’s rules. Contributing means trusting the founder. Charters bought in later rounds are held the same way until the Reserve allows transfers, so that trust grows with the pool. The review was internal.
• The Reserve’s owner. The Reserve’s contracts cannot be upgraded, but its owner keeps bounded control of several economic settings: licences a day (100 today, up to 2,000), charters a day (none today, up to 100), trading taxes (up to 10%), the length of an epoch (one to seven days) and the split of fees. The pool treats those controls as an explicit strategy risk. Faster licence sales bring dilution forward, and charter sales lift the ceiling on branches altogether: in the model, a Reserve that reaches 20,000 branches cuts the reference Pool APY from +230% to about +75%.
• Contraction. Net outflows cut the policy rate to as little as 0.2, which lowers Pool APY at the same time as the price falls.
• Exit-fee spikes. Above 5% the pool does not retire branches, so payouts and withdrawals pause.
• Liquidity. Units are not transferable. Withdrawals are paid from realised $STANDARD over time, with no promised date or amount.
• The model. The Reserve has existed for days and the pool has not yet made a payout. Pool APY is a calculation at one block, not a track record, and it moves with every auction.
• Price. Contributions are ETH and payouts are $STANDARD. Below x = 0.32 a contributor loses ETH even if the model holds.

Standard Reserve whitepaper, version 1.0, at standardreserve.xyz/whitepaper. Chain figures were read from Robinhood Chain (4663) at block 65,629,152 on 17 September 2026. Returns come from the pool’s published payout model at the same block. Current figures: App coming soon.

The Standard Deviation is independent of Standard Reserve and is not affiliated with or endorsed by it. Public contributions are not yet open. Thesis version 1.0.