MATHEMATICAL MODELING & CAPITAL ACCUMULATION

Risk glide and the capacity ceiling

Position sizing steps down a reverse-exponential glide as equity grows, and participation is capped at 50 units so the cohort's own order flow never moves the quote. The arithmetic on this page is the losing sequences, not projections of the winning ones.

THE GLIDE BASELINE

Certified Glide Model Baseline

Audited Monte Carlo parameters across 14 orthogonal streams, inside the risk governor's capital protection boundary.

-11.9%
Max drawdown, published record
68 closed deals, as of 17 Sep 2026 - carried unedited
44.1%
Win rate, same record
A loss is the more likely outcome of any single signal
-9.6%
Cost of a 10-loss streak
At the 1% desk default: 1 - 0.99¹⁰ compounded, not -10%
-40.0%
Hard equity brake
The desk halts when realized drawdown reaches it; members set their own risk below it

Source: Risk glide model (audited multi-factor portfolio validation and capital protection framework)

RISK GOVERNOR

The Reverse-Exponential Multiplier Curve

The risk multiplier m contracts smoothly as equity grows, so order flow never outruns liquidity:

Risk Multiplier vs Portfolio Balance CALCULATED MODEL PATH
m(Balance) = clip(1.0 - 0.8 × (log10(Balance) - 4) / 2, 0.2, 1.0)
1.00x0.70x0.50x0.35x0.20x$10k$31.6k$100k$316k$1.0M+1.00x~0.70x0.50x~0.35x0.20x Floor
Scale Parameter: E_scale = $100,000 Decay Exponent: γ = 0.75 Base Bounds: 1.26% ≤ r ≤ 14.32% Terminal Floor: 0.20x

Source: Certified risk-glide model (reverse-exponential glide parameterization and the systematic allocation framework)

THE LOSING SEQUENCES

What a bad run actually costs

The projection this page once carried - milestone targets with probabilities that never priced a loss - was withdrawn: a model that cannot ruin cannot inform a decision. What follows is arithmetic on the published record, checkable by hand.

  • The bracket. Every signal risks 1R and targets +3R, so the breakeven win rate is 25%. At the record's 44.1% win rate the expectancy per signal is +0.76R; that edge is what the losing sequences spend.
  • Streaks are routine, not anomalies. At 44.1% wins, ten consecutive losses cost 9.6% of the account at 1% risk (compounded). The record's own worst drawdown, -11.9% over 68 closed deals, is the scale a member should expect to sit through.
  • What a bad quarter looks like. A quarter in which the 3R winners do not arrive can carry the account down double digits even with every signal taken exactly as delivered. The desk halts at its 40% equity brake; a member copying at 1% is asked to size so that halving would still be survivable.
  • Members set their own risk. The 1% default is a starting point; the cost figures above scale linearly with it - at 0.5% risk the same ten-loss streak costs 4.9%.
-11.9% Record max drawdown Realized, published, 68 closed deals
-9.6% 10-loss streak at 1% risk Exact compounding: 1 - 0.99¹⁰
25% Breakeven win rate 1R risk against the +3R target

Source: the published track record and the bracket the desk trades.

WHY 50 IS THE CEILING

Market Depth & Aggregate Capacity Governance

Why the ceiling is 50 Liquidity Units: systematic models cannot expand forever without fills degrading.

  • Order book absorption. Metals order books deepen through the London and New York overlap. Sizes up to 10 standard lots fill inside sub-pip spreads without moving the quote.
  • Market impact dampening. As aggregate member equity scales, the risk glide compresses fractional risk automatically, so collective order flow never trips broker toxic-flow classifiers or earns adverse execution selection.
  • The 50-unit equilibrium. Capping total platform participation at 50 units keeps the cohort inside the modeled slippage envelope at the cap - about $0.02 per ounce against $2.10 at retail scale. Fills still vary member to member; nothing is guaranteed beyond the model.
  • Dynamic capacity accounting. Telegram seats weigh 0.5 units because manual execution staggers naturally. The planned Stream API remains non-purchasable; its 2.0-unit weight is reserved for a future gated launch.
50 Liquidity units, total pool Every tier draws from the same cap
0.15 Pips max fill deviation Canonical entry against member fill
0.5 Units per Telegram seat Staggered manual execution
2.0 Planned stream seat Reserved for a future gated launch

Source: Capacity & Execution Reliability at Scale (Level-2 order book depth and the aggregate 50-unit liquidity allocation framework)

Fifty units is the ceiling, because past it the cohort's own order flow moves the quote.

Every tier draws from the same 50-unit pool. A future planned Stream API entitlement would contract it four times faster than a Telegram seat; none is active while the gateway is contained. That is why the price steps up as seats fill instead of the seat count growing.

See the seat ladder Watch the fills on the tape

Canonical sourced documents

Sourced documents for the glide equation and the capacity ceiling. Drawdown figures on this page are the record's own or exact arithmetic on the stated risk; nothing is a modeled upside.