# Kelly Criterion, Initial Margin & Liquidation Formula Guide

> Master institutional position sizing: fractional Kelly optimal leverage, CME SPAN 2 & ISDA SIMM initial margin, cross vs isolated liquidation, and ruin math.

Canonical: https://quantcalctools.com/guides/kelly-criterion-initial-variation-margin-liquidation-sizing-guide
Last updated: 2026-11-14T19:30:02.871Z

**Publication:** QuantCalc Tools
**Category:** Kelly Margin Leverage Sizing
**Published:** 2026-11-14T19:16:25.915Z
**Author:** Elena Rostova, FRM, CQF (FRM, Certificate in Quantitative Finance (CQF))
**Reviewer:** Not recorded

## Answer-first summary

Position sizing and margin mechanics dictate whether a positive-expectancy edge compounds exponentially or hits forced liquidation; this quantitative hub unifies Kelly math, CCP margin, and close-out thresholds.

## Key takeaways

- Fractional Half-Kelly (0.5 * f*) captures 75% of theoretical maximum exponential growth while cutting portfolio variance by 75% and slashing the probability of a -50% drawdown from 50.0% down to 12.5%.
- Betting strictly above 2.0 * f* (Double-Kelly) turns expected log growth negative even on a positive-expectancy strategy, guaranteeing asymptotic ruin as variance drain overwhelms arithmetic edge.
- Unlike equity Reg T maintenance calls (which only top equity back up to the 25%–30% maintenance line), breaching CME futures Maintenance Margin triggers a Variation Margin call all the way back to full Initial Margin!
- Under UK FCA COBS 22.5 and ESMA rules, retail leveraged accounts combine asset-class initial margin floors (3.33% major FX, 5% indices/gold) with mandatory automated close-out at 50% of total Initial Margin.

## Related financial entities

- **CME SPAN 2 Historical & Stress Value-at-Risk (VaR) Model** (CME-SPAN2-VAR) — product: CME SPAN 2 Historical & Stress Value-at-Risk (VaR) Model (CME-SPAN2-VAR) refers to an institutional financial benchmark and regulatory standard monitored by CME Clearing (US), currently calibrated at 6.26% (Initial Margin / Notional) as of October 2026.
- **Continuous Fractional Half-Kelly Portfolio Leverage Criterion** (HALF-KELLY-LEVERAGE) — product: Continuous Fractional Half-Kelly Portfolio Leverage Criterion (HALF-KELLY-LEVERAGE) refers to an institutional financial benchmark and regulatory standard monitored by BIS Quantitative Working Paper (INTL), currently calibrated at 1.635x (Optimal Gross Leverage) as of October 2026.
- **Continuous Fractional Half-Kelly Portfolio Leverage Criterion — 2026 Standard** (HALF-KELLY-LEVERAGE-STD) — product: HALF-KELLY-LEVERAGE 2026 Institutional Benchmark Framework is defined as the standardized quantitative and compliance methodology published by BIS Quantitative Working Paper governing Continuous Fractional Half-Kelly Portfolio Leverage Criterion across INTL capital markets (1.635x Optimal Gross Leverage, October 2026).
- **ISDA Standard Initial Margin Model (SIMM) v2.6 Sensitivity Grid** (ISDA-SIMM-MARGIN) — license: ISDA Standard Initial Margin Model (SIMM) v2.6 Sensitivity Grid (ISDA-SIMM-MARGIN) refers to an institutional financial benchmark and regulatory standard monitored by ISDA SIMM Governance Forum (INTL), currently calibrated at $50M (IM Threshold) as of October 2026.

## Analysis

Positive expected value is a necessary condition for trading profitability, never a sufficient one: over-betting a genuinely profitable strategy beyond **twice the Kelly Criterion** ($f > 2f^*$) drives terminal wealth to **$0.00$ with $100\%$ mathematical probability**, and miscalculating exchange Maintenance Margin or FX quote-to-account pip conversion triggers automated broker liquidation right at the trough of an otherwise recoverable wick. Institutional capital allocation sits at the intersection of **utility-theoretic optimal sizing** (John Kelly 1956, Edward Thorp 1969, Robert Merton 1971) and **statutory clearinghouse margin rules** (CME SPAN 2, ISDA SIMM v2.6, FINRA Rule 4210, and UK FCA COBS 22.5). This pillar hub connects our live [Kelly Criterion & Fractional Leverage Calculator](/calculators/kelly-criterion-calculator) and [Value-at-Risk Calculator](/calculators/value-at-risk-calculator) with 19 rigorous derivations spanning Kelly covariance shrinkage, bankruptcy prices, CCP liquidity spirals, and option margin expansion.

## Discrete & Continuous Kelly Sizing, Covariance Shrinkage & Ruin Bounds

Maximizing expected logarithmic growth $g(f) = \mathbb{E}[\ln(1 + fR)]$ balances linear return drift against quadratic variance drain. Yet unconstrained full-Kelly sizing suffers a brutal flaw in live trading: under continuous Brownian motion, a full-Kelly bettor faces a **$50.0\%$ probability of suffering a $-50\%$ drawdown** before doubling equity, and feeding noisy sample means $\hat{\mu}$ into multi-asset $\Sigma^{-1}(\hat{\mu} - r_f\mathbf{1})$ wildly over-levers estimation errors! Our [Continuous Fractional Half-Kelly Engine](/entities/half-kelly-leverage-continuous-fractional-half-kelly-portfolio-leverage-criterion) tames both risks:
- **Core Kelly & Empirical Trade Sizing**: Derive binary and asymmetric payoff fractions $f^* = p - q/b$ in [discrete binomial Kelly payoff asymmetry math](/guides/discrete-binomial-kelly-criterion-win-rate-payoff-asymmetry-proof), bridge Ito calculus to $f^* = (\mu - r_f)/\sigma^2$ in [continuous Merton-Kelly and Half-Kelly leverage](/guides/continuous-merton-kelly-leverage-mu-over-sigma-squared-half-kelly), generalize non-parametric trade logs in [Ralph Vince Optimal f largest-loss scaling](/guides/ralph-vince-optimal-f-empirical-trade-distribution-largest-loss), and compute exact absorption probabilities in [Markov gambler's ruin vs fixed-fractional sizing](/guides/risk-of-ruin-gamblers-ruin-markov-absorption-fixed-fractional-math).
- **Multi-Asset Shrinkage, Drawdown Caps & Volatility Targeting**: Stabilize ill-conditioned $\Sigma^{-1}$ matrices via Ledoit-Wolf and Bayesian priors in [multi-asset Kelly inverse-covariance shrinkage](/guides/multi-asset-kelly-portfolio-inverse-covariance-matrix-shrinkage) and [Black-Litterman posterior Bayesian Kelly sizing](/guides/black-litterman-posterior-expected-return-kelly-bayesian-sizing), enforce exact tail drawdown ceilings in [drawdown-constrained fractional Kelly math](/guides/drawdown-constrained-fractional-kelly-probability-of-50pct-loss), and size daily ATR / Equal-Risk-Contribution tickets in [volatility-target ATR & ERC position sizing](/guides/volatility-target-atr-stop-loss-equal-risk-contribution-sizing).

```formula
Unified Optimal Position Sizing, Exchange Margin & Forced Liquidation Equations:
 1. Discrete Asymmetric Kelly : f* = p / a - q / b (Win +b w/ prob p, Lose -a w/ prob q = 1 - p; Full Loss a=1 -> p - q/b)
 2. Continuous Merton-Kelly : f* = (mu - r_f) / sigma^2; Fractional c-Kelly keeps c*(2 - c) of max growth (c=0.5 -> 75% growth, 25% var!)
 3. Futures / Equity Margin : Margin Call Price (Long, Maint m) => P_call = P_0 * (1 - IM_pct) / (1 - m)
 4. Isolated Long Liquidation : P_liq = P_entry * [ 1 - (1 / L) + MMR + Fee_close ] (Bankruptcy P_bank = P_entry * (1 - 1/L))
 5. FX Lot Size (Account CCY) : Lots = (Equity * Risk_pct) / (Stop_Pips * Pip_Value_per_Lot_in_Account_CCY)

Worked Full-Kelly vs Half-Kelly vs Over-Kelly Audit (mu - r_f = 12.0%, sigma = 20.0%, r_f = 4.5%):
 * Full-Kelly Leverage f* = 0.12 / (0.20)^2 = 3.00x -> Excess Log Growth = 0.5*(0.12^2/0.04) = +18.00% (Vol = 60%, P(50% DD) = 50.0%!)
 * Half-Kelly Leverage 0.5*f* = 1.50x -> Excess Log Growth = 0.75 * 18.00% = +13.50% (Vol = 30%, P(50% DD) = 12.5%!)
 * Double-Kelly Overbet 2.0*f*= 6.00x -> Excess Log Growth = 6*(12%) - 0.5*(36)*(4%) = 0.00% (120% vol for ZERO excess growth!)
```

![Institutional position sizing and margin survival pipeline integrating Bayesian Kelly sizing, exchange initial margin, liquidation buffers, and cash spirals](/api/figure?type=flow&title=Four-Layer%20Position%20Sizing%20%26%20Margin%20Survival%20Architecture&subtitle=Ensuring%20trade%20sizing%20clears%20both%20statistical%20log-growth%20optimality%20and%20clearinghouse%20margin%20gates.&steps=1.%20Signal%20%26%20Prior%20Shrink%3A%3AShrink%20noisy%20alpha%20mu%20%26%20covariance%20Sigma%3B%20cap%20at%200.25x-0.50x%20Half-Kelly%7C2.%20Ticket%20Risk%20Normalization%3A%3AConvert%20stop%20pips%20%26%20quote%20FX%20rates%20into%20exact%20base-currency%20risk%20units%7C3.%20Initial%20Margin%20Gate%3A%3AVerify%20CME%20SPAN%202%2C%20ISDA%20SIMM%20v2.6%2C%20Reg%20T%2C%20or%20FCA%2030%3A1%20initial%20margin%7C4.%20Liquidation%20%26%20VM%20Buffer%3A%3AHold%203x%20Maintenance%2FVM%20cash%20buffer%20above%20cross%2Fisolated%20close-out%20price!&metrics=HALF-KELLY%20EFFICIENCY%3A%3A75%25%20Growth%20%2F%2050%25%20Vol%3A%3ACuts%20P%2850%25%20DD%29%20from%2050%25%20to%2012.5%25!%7COVER-KELLY%20CLIFF%3A%3Af%20%3E%202%20*%20f*%20%3D%3E%20Ruin%3A%3ANegative%20drift%20despite%20positive%20edge%7CFCA%20RETAIL%20CLOSE-OUT%3A%3A50%25%20Total%20Margin%3A%3AMandatory%20account-level%20COBS%2022.5%20rule)

## Institutional Clearinghouse Margin: CME SPAN 2, ISDA SIMM, Reg T & Repo Haircuts

Professional derivatives books do not post flat retail percentage margins; they post risk-sensitive **Initial Margin (`IM`)** calibrated to 99% 1-to-10-day tail stress plus intraday cash **Variation Margin (`VM`)**:
- **Cleared Futures, Uncleared OTC & Prime Brokerage**: Contrast legacy 16-scenario grids with Historical VaR in [CME SPAN 2 Initial vs Maintenance Margin top-up math](/guides/cme-span-2-historical-var-initial-margin-maintenance-top-up-math) ([CME SPAN 2 Model](/entities/cme-span2-var-cme-span-2-historical-stress-value-at-risk-var-model)), aggregate Delta, Vega, and Curvature sensitivities in [ISDA SIMM v2.6 uncleared OTC Initial Margin](/guides/isda-simm-v26-uncleared-otc-initial-margin-delta-vega-curvature) ([ISDA SIMM v2.6 Grid](/entities/isda-simm-margin-isda-standard-initial-margin-model-simm-v2-6-sensitivity-grid)), compare static `50%` Reg T against TIMS stress grids in [Reg T vs FINRA 4210 Portfolio Margin house calls](/guides/reg-t-50pct-equity-margin-vs-finra-4210-portfolio-margin-house-call), model option crash margin spikes in [short naked option Reg T 20% OTM gamma expansion](/guides/short-naked-option-reg-t-20pct-otm-minimum-margin-gamma-expansion), and bound fixed-income repo balance sheets in [repo haircut & rehypothecation leverage multipliers](/guides/repo-haircut-rehypothecation-multiplier-maximum-balance-leverage).

## Retail Leverage Caps, Liquidation Bankruptcy Prices & Liquidity Spirals

When market volatility spikes $3\times$ overnight, margin requirements jump *simultaneously* with mark-to-market losses—triggering Brunnermeier-Pedersen margin spirals unless unencumbered cash buffers are pre-funded:
- **Close-Out Mechanics, Inverse Perps & GSLOs**: Audit statutory retail leverage caps (`30:1` major FX, `20:1` gold/indices) and account-wide `50%` margin close-out under [UK FCA COBS 22.5 & ESMA retail margin math](/guides/fca-cobs-225-esma-retail-50pct-margin-close-out-leverage-caps-math), derive exact bankruptcy vs maintenance prices in [isolated margin vs cross-margin liquidation math](/guides/isolated-margin-vs-cross-margin-liquidation-bankruptcy-price-math), expose non-linear $1/P$ payoff convexity in [inverse coin-margined vs linear USDT perpetual liquidation](/guides/inverse-coin-margined-vs-linear-usdt-perpetual-liquidation-math), convert cross-currency stops accurately in [FX pip-value lot sizing and quote conversion](/guides/fx-pip-value-lot-sizing-quote-account-currency-conversion-math), survive CCP margin calls in [central-clearing Variation Margin liquidity buffer sizing](/guides/central-clearing-variation-margin-liquidity-spiral-buffer-sizing), and price gap insurance in [Guaranteed Stop-Loss Order (GSLO) premium vs slippage math](/guides/guaranteed-stop-loss-order-gslo-premium-vs-slippage-deficit-math).

## Institutional & Retail Margin Regimes Comparison Matrix

Compare margin methodology, holding horizon, top-up mechanics, and close-out rules across major global regimes:

| Margin / Sizing Framework | Calibration Horizon & Confidence | Maintenance / Close-Out Trigger | Portfolio Cross-Offsetting? | Primary Venue / Regulator |
| :--- | :--- | :--- | :--- | :--- |
| **Half-Kelly Optimal ($0.5 f^*$)** | Continuous infinite-horizon log utility | Voluntary rebalance band | Full covariance $\Sigma^{-1}$ | Quant funds ([Kelly Tool](/calculators/kelly-criterion-calculator)) |
| **CME SPAN 2 (Cleared Futures)** | $99\%$ 1–5d HVaR + Stress + Liquidity | Equity $<$ Maint $\to$ **Top up to full IM!** | Cross-product pod offsets | CME / Clearinghouses ([benchmarks](/benchmarks)) |
| **ISDA SIMM v2.6 (Uncleared OTC)** | $99\%$ **10-day** 3y+1y stress sensitivities | Bilateral CSA threshold + daily VM | Within 6 risk classes only | Prudential ([UK PRA](/regulator-index/prudential-regulation-authority-uk-173) / BCBS) |
| **FINRA 4210 Portfolio Margin** | $\pm 15\%$ index / $\pm 15\%\text{--}30\%$ stock TIMS | Real-time house deficit liquidation | Broad delta/vega offsets | Institutional / Prime US accounts |
| **FCA COBS 22.5 / ESMA Retail** | Statutory `3.33%–50%` asset notional caps | **Mandatory `50%` Initial Margin close-out** | Account-level net equity | [UK FCA](/regulator-index/fca) / [CBI](/regulator-index/cbi) ([brokers](/broker-lookup/activtrades)) |

![Parabolic Kelly growth curve showing expected excess log growth peaking at 1.0x Full Kelly, retaining 75 percent at 0.5x Half Kelly, and turning negative past 2.0x](/api/figure?type=bars&title=The%20Parabolic%20Kelly%20Cliff%3A%20Growth%20vs%20Drawdown%20Across%20f%20%2F%20f*&subtitle=Comparing%20expected%20excess%20log%20growth%20%28%25%20of%20max%29%20and%20P%2850%25%20Drawdown%29%20from%20Quarter-Kelly%20to%202.2x%20Over-Kelly.&steps=0.25x%20Quarter-Kelly%3A%3A43.8%25%20of%20max%20growth%20%20%20Just%2025%25%20of%20Full-Kelly%20vol%20%20%20P%2850%25%20DD%29%20%3D%201.6%25!%7C0.50x%20Half-Kelly%3A%3A75.0%25%20of%20max%20growth%20%20%2050%25%20of%20Full-Kelly%20vol%20%20%20P%2850%25%20DD%29%20%3D%2012.5%25%7C1.00x%20Full-Kelly%3A%3A100.0%25%20max%20growth%20peak%20%20%20100%25%20vol%20%20%20P%2850%25%20DD%29%20%3D%2050.0%25%20coin%20flip!%7C2.00x%20Double-Kelly%3A%3A0.0%25%20excess%20growth%20%20%20200%25%20vol%20%20%20P%2850%25%20DD%29%20%3D%2087.5%25%20-%3E%20Ruin%20cliff%20beyond%202x!&metrics=SWEET%20SPOT%20%280.35x-0.5x%29%3A%3AHalf-Kelly%3A%3ASacrifices%2025%25%20growth%20to%20cut%20variance%2075%25!%7CFUTURES%20TOP-UP%20TRAP%3A%3ARestore%20to%20IM!%3A%3AFutures%20call%20demands%20full%20IM%2C%20not%20Maint%7CSIMM%20HORIZON%3A%3A10-Day%2099%25%20VaR%3A%3ABilateral%20gross%20IM%20%28no%20netting%20across%20dealers%29%7CNEGATIVE%20BALANCE%20SHIELD%3A%3AFCA%20COBS%2022.5%3A%3ACaps%20retail%20loss%20at%20deposited%20equity)

## Execution Infrastructure & Regulatory Margin Verification

Test exact ticket sizes and margin buffers before routing live orders on [MT4](/software/mt4), [MT5](/software/mt5), [cTrader](/software/ctrader), or [FIX API](/software/fix-api). Browse all 20 position-sizing guides on the [Kelly, Margin & Liquidation Topic Hub](/topics/kelly-margin-leverage-sizing), audit regulated margin schedules across [ActivTrades](/broker-lookup/activtrades) and [Admirals](/broker-lookup/admirals), and verify firm permissions on the [UK FCA register](/regulator-index/fca) and [UK PRA register](/regulator-index/prudential-regulation-authority-uk-173).

## Frequently asked questions

### Why do institutional quantitative funds almost never trade at 1.0x Full Kelly?

Two mathematical reasons: first, Full Kelly assumes you know true expected return mu and volatility sigma with zero estimation error—whereas overestimating mu by 2x accidentally pushes you onto the 2.0*f* zero-growth cliff! Second, even with known parameters, Full Kelly endures a 1/k probability of shrinking equity to 1/k (a 50% chance of a -50% drawdown).

### What is the difference between Initial Margin and Maintenance Margin on futures versus equities?

Initial Margin (IM) is the collateral required to open a position; Maintenance Margin (MM) is the minimum equity floor required to keep it open. If a futures account dips $1 below MM, you must wire cash all the way back up to full IM; if a Reg T equity margin account dips below 25%/30% MM, you only top back up to MM.

### Does cross-margining lower or raise my probability of account wipeout compared to isolated margin?

Cross-margining lowers the probability that a single noisy wick liquidates one small position (by sharing free equity across the whole account), BUT if an un-stopped runaway trade breaches cross-margin maintenance, it liquidates your ENTIRE account balance rather than capping loss at one isolated margin bucket!

### Why is ISDA SIMM Initial Margin posted gross by both OTC counterparties without netting against each other?

Under post-2008 BCBS-IOSCO uncleared margin rules, both Dealer A and Institutional Client B must post 99% 10-day SIMM Initial Margin into segregated third-party custodians so that if either party defaults, surviving close-out replacement risk is funded without creating systemic contagion.


## Source URLs

- https://bell-labs.co/pub/kelly-1956.pdf
- https://www.cmegroup.com/clearing/risk-management/span-overview.html
- https://www.isda.org/category/margin/isda-simm/
- https://www.handbook.fca.org.uk/handbook/COBS/22/5.html

## Citation

- APA: Elena Rostova, FRM, CQF. (2026). Kelly Criterion, Initial/Variation Margin & Liquidation Math: 2026 Guide. QuantCalc Tools. https://quantcalctools.com/guides/kelly-criterion-initial-variation-margin-liquidation-sizing-guide
- Canonical article: [Kelly Criterion, Initial/Variation Margin & Liquidation Math: 2026 Guide](https://quantcalctools.com/guides/kelly-criterion-initial-variation-margin-liquidation-sizing-guide)
- Site research index: [QuantCalc Tools](https://quantcalctools.com/llms-full.txt)
