Skip to content
← The wizards
DeclassifiedThe Position Sizing Theorist

Ralph Vince

Operationalized by

Throttle Control Panel — the 7-tier deployment ladder

Integration rate

67% — computed, with the gaps below

The record

What is publicly documented.

Verified against multiple public sources. Where sources disagree, the disagreement is stated rather than resolved silently.

Profession
Computer programmer — analytical software for funds, large traders and professional gamblers
Field
Portfolio management, position sizing, trade optimisation
Portfolio Management Formulas
1990 — optimal f introduced
The Mathematics of Money Management
1992 — the standard reference
The New Money Management
1995
The Handbook of Portfolio Mathematics
2007 — drawdown as the risk metric
The Leverage Space Trading Model
2009 — growth-optimal within a drawdown constraint
Risk-Opportunity Analysis
2012
Core constructs
Optimal f · geometric mean maximisation · leverage space · risk of ruin

Vince is not a fund manager and has never claimed to be. He is a programmer who writes analytical software for funds, large traders and professional gamblers, and the gambling half of that client list is the tell — the mathematics of how much to bet, given an edge, is the same problem whether the edge comes from a market or a table. He arrived at trading through that door rather than through a trading floor.

The 1990 and 1992 books put optimal f into the retail vocabulary. The result is uncomfortable and correct: for any distribution of outcomes there is a single fraction of capital that maximises geometric growth, it is computable, and both under-betting and over-betting it cost you. Over-betting costs you catastrophically. The corollary — that two traders taking identical signals can produce a compounding machine and a smoking crater purely through sizing — is the most important idea in the retail money-management canon and remains the least acted on.

The part most readers stop before reaching is that Vince himself moved on. By the 2007 and 2009 books the risk metric had shifted from variance to drawdown, and the framing had become explicitly constrained: find the growth-optimal allocation subject to a drawdown limit you can actually tolerate. He argued that unconstrained geometric maximisation, while mathematically correct, describes an investor nobody actually is. That later Vince — drawdown-constrained rather than growth-maximising — is the one MARS is built on, and the distinction matters more than any other line in this dossier.

The framework

The idea underneath the method.

Not the trades — the reasoning that decides which trades are permitted, at what size, and when they stop.

Sizing is a strategy variable, not an afterthought

The founding argument, and it survives every later revision. Analysis produces a stream of outcomes; sizing converts that stream into an equity curve. Two operators with the same signals, the same entries and the same exits will not have the same result, the same drawdown, or the same probability of surviving the year, and the only thing separating them is the fraction of capital committed per trade. Treating that fraction as a matter of comfort rather than computation is the largest unpriced decision most traders make.

There is an optimum, and it cuts both ways

Optimal f is the fraction that maximises the geometric mean of returns for a given outcome distribution. The curve around it is not symmetric. Bet meaningfully below the optimum and you give up growth — a real cost, slowly paid. Bet meaningfully above it and the geometric mean falls off a cliff, and past a certain point the expected terminal wealth of a positive-expectancy system goes to zero. The arithmetic mean stays positive the whole way down. This is the single most counter-intuitive result in the field: a system with a genuine edge, sized aggressively enough, is a losing business.

The drawdown constraint is the real problem

The later work's central move. Growth-optimal sizing produces drawdowns that are mathematically survivable and psychologically not — the optimum for a typical trading distribution routinely implies peak-to-trough declines that no human operator and no client relationship endures. So the honest formulation is not 'what maximises growth' but 'what maximises growth subject to a drawdown ceiling I will not breach'. Once drawdown is the binding constraint rather than an output, the whole allocation problem changes shape.

Ruin is a price, and it should be quoted

Risk of ruin is treated as a number to be computed and traded against, not a vague hazard. Every increment of leverage buys growth and sells survival probability, and the exchange rate is calculable from the outcome distribution. A sizing scheme that does not state its ruin probability has not been evaluated, only chosen.

Mechanics

How it actually runs.

The operating detail, stated the way a reference describes a technique.

Input

The empirical distribution of trade outcomes — not a win rate and an average, but the full set of results including the worst one.

The worst case anchors it

Optimal f is computed relative to the largest losing outcome in the sample, which is why the figure is unstable on small samples and why a new worst trade rewrites it.

Output

A fraction of capital per unit of risk, from which position size follows mechanically.

Drawdown constraint

In the later model, the allocation is solved within a stated maximum drawdown rather than maximising growth unconditionally.

Portfolio dimension

Leverage space extends the problem to multiple simultaneous market systems, where the joint distribution — not each system alone — determines the allocation.

Honest limitation

Every output depends on the sampled distribution being representative of the future one. Vince is explicit about this; most people quoting optimal f are not.

Divergence

Where MARS does something else.

Listed first, and at length, because a mapping that only claims similarity is a poster. Some of these are scale limits; at least one on every dossier is a deliberate refusal.

Continuous fraction versus discrete ladder

Ralph Vince

A computed fraction of capital, recalculated as the outcome distribution updates. Continuous, precise, and sensitive to the worst trade in the sample.

MARS

A seven-tier discrete ladder with fixed per-trade averages inside each tier. Coarser on purpose: a tier is legible, auditable and cannot drift a decimal at a time, and it does not get rewritten by one bad trade.

What sets the level

Ralph Vince

The outcome distribution sets the fraction. Capital state enters only through the drawdown constraint in the later model.

MARS

Drawdown from peak routes the tier directly. The gate reads one objective number and sets a ceiling, and expectancy cannot lift it. Evidence informs; capital state decides.

Aggression

Ralph Vince

Optimal f, taken literally, is far more aggressive than almost any operator can hold. Vince says so himself — it is the reason for the later constrained framing.

MARS

Explicitly rejects the unconstrained optimum. The ladder's ceiling exists to make the aggressive branch of the geometric curve unreachable rather than merely inadvisable.

Portfolio joint distribution

Ralph Vince

Leverage space solves allocation across correlated market systems simultaneously, using the joint distribution.

MARS

The exposure pool sums gross and does not model correlation — stated as an honest limit on the site. It errs toward forgone deployment, never toward hidden leverage.

Replication

What the rail carries over.

Each mapping names the module that performs the function, so the claim can be checked against the product rather than taken on trust.

Sizing as a first-class object

Throttle Control Panel

The Throttle exists because of the argument in Vince's first two books. Position size is not derived at the end of a trade plan; it is the output of a dedicated panel with its own inputs, its own authority and its own governance. Every deployment figure in MARS comes from there, and no other layer may raise it.

The drawdown constraint, made structural

The gate ladder · drawdown routing

This is the deepest borrowing and it comes from late Vince, not early. The seven-tier ladder is a growth-within-a-drawdown-ceiling scheme implemented as discrete states: as drawdown from peak deepens, the ceiling falls, mechanically and without negotiation. Vince argued the constraint should bind. MARS makes it bind by construction, which is the same idea with the discretion removed.

Ruin priced by simulation

Monte Carlo Lab · Dynamic 7-Tier Benchmark

Where Vince computes ruin analytically from the outcome distribution, MARS resamples it. The Monte Carlo layer produces the distribution of drawdowns and terminal outcomes a given deployment profile implies, and the seven-tier benchmark states what each tier is expected to look like. Different mathematics, identical purpose: quote the survival price before paying it.

The geometric framing

Risk & Equity Accelerator · compounding geometry

MARS reasons in R multiples and compounding terms rather than in per-trade currency, which is Vince's frame. The accelerator metrics ask whether risk is converting into equity growth efficiently — a geometric question, not an arithmetic one.

Integration rate

Scored, with the shortfall shown.

Each dimension is judged separately and the headline is their mean — recomputed at render, so it cannot be hand-set. Every dossier in this library carries at least one dimension below 35%. Four uniformly high scores would be marketing.

Integration rate

What MARS actually reproduces

Throttle Control Panel — the 7-tier deployment ladder

67%

mean of 5 dimensions

  • Sizing as a strategy variable

    95%

    A dedicated panel with its own authority — the whole premise of the Throttle

  • Drawdown as the binding constraint

    90%

    The gate ladder is late Vince implemented as discrete, non-negotiable states

  • Geometric growth framing

    70%

    R multiples and compounding, but no explicit geometric-mean maximisation step

  • Risk of ruin priced explicitly

    65%

    Resampled by the Monte Carlo Lab rather than computed analytically from the distribution

  • Optimal f / continuous fractional sizing

    15%

    Deliberately rejected — the discrete ladder exists to make the aggressive branch unreachable

Why it is not higher

The 15 is a doctrinal refusal rather than a gap. MARS does not implement optimal f and will not: the figure is anchored on the worst trade in the sample, so it is unstable on retail sample sizes, and its unconstrained form implies drawdowns that end trading careers. The system takes Vince's problem statement and his later constrained answer, and declines his most famous formula.

Instrument 04 / 04

The result, in one control.

Vince's central claim is that the same sequence sized two ways is two different businesses. It is easier to believe after you have moved the slider yourself.

Instrument 04 / 04

Optimal-f versus the 7-tier throttle

One outcome sequence, two sizing regimes. The unconstrained fraction compounds without authority; the throttle routes the same fraction through drawdown. Move the risk slider until the two diverge.

6.0%
conservativepast the optimum
45%
2570
2.0R
1.04.0

Expectancy per trade

+0.35R

Unchanged by either slider below it. The edge is identical in both regimes — only the sizing differs.

Unconstrained fraction 7-tier throttlelog scale · 240 trades · fixed sequence

Unconstrained — terminal

5867

max drawdown -46.7% · started at 100

Throttled — terminal

1085

max drawdown -28.5% · 0 trades at System Lock

Below the optimum — governance costs growth here

Below the optimum the unconstrained curve wins, and the throttle's ceiling is a real cost paid in forgone growth. That cost is the premium. Raise the risk slider and watch what it buys.

Demonstration of a sizing property, not a backtest. Both curves run the same fixed synthetic sequence generated from the inputs above with a constant seed. It is not a MARS track record, contains no market data, and is not a comparison against any individual's performance.

This runs one fixed sequence at one fraction. The Throttle solves the same problem against your live cycle pool, open exposure and tier ceiling — The Throttle Control Panel

Open the full Foundry Lab sandbox ↗

Computed locally in your browser. Nothing is uploaded.

Ralph Vince built the governance function privately, because nothing off the shelf existed to buy. So did every other trader in this library. That private apparatus — the constraint stack, the sizing authority, the rule about when to stop — is the part that never gets published, and it is the part that separates a documented edge from a surviving account.

The claim on this page is not that MARS outperforms anyone. It is narrower and considerably more useful: their method is public, their apparatus was not, and this is the apparatus — at a scale one person can actually run.

Sealed · nearest neighbours

Three dossiers sit next to Vince's.

These mappings are not in the bundle and not rendered anywhere on this site. They are the closest methodological neighbours to the dossier you have just read — which is precisely why they are the ones held back.

Eleven dossiers remain sealed, each at this depth — the framework, the record, the divergences, and the integration rate with its shortfall shown. Waitlist registration unlocks all fifteen.

Join the waitlist — unlock all 15

Also declassified

The other three open dossiers.

Sources and standing notice

Ralph Vince has no affiliation with Aura Logic Systems or the Montex AlphaRail System, and nothing on this page constitutes an endorsement. This dossier summarises publicly documented method and publicly reported career facts, in the way a reference work describes a technique. It contains no quotations. Performance figures are as reported by the sources listed below, are historical, and are not audited by us; past performance of any trader, fund or method does not indicate future results. Nothing here is investment advice.

The doctrine, made executable.

Eleven governed modules that turn documented method into arithmetic you can actually run.