Show the method, not just the number.

Average price and cost basis

The Average Down calculator supports purchase lots, not sales or tax-lot accounting. Each lot contributes price × quantity plus any entered buy fees in the quote currency. Total cost divided by total quantity gives the average paid per unit.

For a proposed additional buy, the calculator solves for the quantity required to meet a selected average price, taking the proposed buy's percentage and fixed fees into account. Impossible targets are reported as such rather than silently coerced.

Precision and rounding

Calculations use exact fractions based on base-10 inputs, with BigInt numerators and denominators. Rounding is applied only to the displayed answer (half-up). The target calculation may show an approximate quantity; real exchanges enforce lot sizes and tick sizes, so the actual result can differ slightly.

Market price and unrealized P&L

When you enter a market price, the calculator estimates gross liquidation value minus historical purchase cost. It does not subtract a future sale commission or account for spread, slippage, financing, taxation, withdrawals or corporate actions.

Risk-based position sizing

The linear position-sizing model converts a chosen account balance and risk percentage (or absolute risk amount) into a risk budget. For a long position, the stop is below the entry; for a short, it is above. The assumed adverse stop fill adjusts the stop by the entered slippage percentage. The model calculates loss per unit as the adverse price difference plus both entry and stop-exit fees.

For long positions, you can optionally cap the quantity so that entry notional plus entry fees do not exceed the available account balance. Stops can fill worse than estimated; inverse contracts, margin requirements, borrowing, funding and liquidation are not modeled. The displayed risk budget is a user input, not a suggested amount.

Inverse COIN-M contract risk sizing

COIN-M uses a fixed USD contract face value, while payoff, fees and margin settle in the collateral coin. Long P&L in coin = contracts × USD face × (1 / entry − 1 / exit); short P&L reverses the sign. Sizing divides a coin risk budget by the coin loss per whole contract, including modeled execution fees, then limits by indicative coin margin plus entry fee and rounds down to the contract increment. USD equivalents are snapshots at specified prices, not constant-dollar risk. Real exchange contract face values, tick sizes, funding, cross margin, maintenance tiers and liquidation must be verified separately. View COIN-M Position Size ↗

Risk Lab — Monte Carlo simulation

The Monte Carlo Simulator is a seeded, deterministic pseudorandom experiment (Mulberry32, not cryptographic or sourced market randomness) over user-defined independent binary trade outcomes. With a fixed fraction f of current capital at risk and a gross winner payout R, a win multiplies equity by (1 + f × R) and a loss by (1 − f). The input win rate is the unchanged Bernoulli probability on every trade; results make no claim about actual trade outcomes.

For each of 100–5,000 paths and up to 500 trades, the engine records ending capital, worst peak-to-trough drawdown on that path, whether it ever touched a user-defined floor relative to STARTING capital, and the capital at up to 31 timeline checkpoints. Cross-sectional P10/P50/P90 percentiles use linear interpolation of sorted observations. The fan chart uses a logarithmic capital axis. Its three lines are not necessarily individual paths.

The sampled first-passage probability is the proportion of paths that touched the floor at least once. The displayed 95% Wilson interval describes only binomial sampling variation given fixed model assumptions. We independently compute an exact absorbing-binomial dynamic-programming benchmark for this same binary setup; it is not an independently verified real-world risk estimate. Monetary calculations within Monte Carlo use floating-point arithmetic, unlike the separate BigInt exact-decimal cost basis tools. No fee, slippage, stop gap, partial close, variable payoff, market-dependent probability, funding, liquidation or real exchange series is modeled. Explore Monte Carlo ↗

Scenario & Stress Lab: assumptions versus robustness

Scenario & Stress Lab uses the exact absorbing first-passage Risk of Ruin model for a 5×5 sensitivity matrix and four deliberately adverse stress assumptions. The matrix varies win rate by −10, −5, 0, +5 and +10 percentage points, clipped to 0–100%; percentage risk is scaled by 0.5×, 0.75×, 1×, 1.25× and 1.5×. Reward/risk, horizon and the loss floor stay fixed. Stress rows reduce win rate by 5 percentage points, winner payout by 20%, increase stake risk by 50%, or combine all three. These are illustrative conditional scenarios rather than empirical confidence bands. The expected log growth per trade equals p×ln(1+fR) + (1−p)×ln(1−f), multiplied by 100 for display.

A separate absorbing run-length dynamic program calculates the chance of at least one K-loss streak anywhere in N independent trade outcomes. The deterministic sizing illustration displays K losses then M wins using either a fraction f of CURRENT equity or fixed dollar risk A=f×starting capital. Fixed-dollar sizing halts when it cannot fund the original stake. This ordered path is not a probability distribution or forecast. No trading costs, liquidations, margin maintenance, correlation, fat tails or changes in market regime are included. Explore Scenario & Stress Lab ↗

Research standards

Any future backtest will disclose its execution timestamp rule, source, coverage, fees, missing-data policy and algorithm version. Historical performance will not be represented as a forward-looking guarantee.

Known limitations of the current models

No exchange API or live price feed, account synchronization, portfolio storage, automatic currency conversion, exchange-specific liquidation engines or tax-lot accounting. Some tools model derivatives mathematically but do not place or manage real orders. Results are educational scenarios, not instructions to trade.