How Options Simulator runs Monte Carlo over its 7 stochastic models for portfolio risk analytics, strategy stress testing, and single-option MC pricing — with real outputs.
Key takeaways This article explains how Options Simulator implements Monte Carlo simulation over its seven stochastic price models — GBM (Steady Drift), Merton jump diffusion (Sudden Shock), Heston (Shifting Volatility), fBm (Deep Tide), GARCH (Calm & Storm), Bates (Perfect Storm) and SABR (Volatility Smile). Rather than a separate add-on, the Monte Carlo engine uses the same models and parameters driving the simulated market. Three use cases are covered with real outputs from the simulator: portfolio risk analytics (Risk metrics with expected P/L, probability of profit, VaR/CVaR at 95% and 99%, worst and best case; a P/L fan; a terminal P/L distribution; rewind-aware), strategy stress testing of multi-leg positions from the Custom Strategy panel (illustrated with a bimodal short iron condor), and MC Price — Monte Carlo pricing of a single option with Longstaff-Schwartz early exercise, a 95% confidence interval and the difference versus the chain price. The implementation separates real-world drift for the underlying from risk-neutral pricing for options, uses variance reduction (antithetic variates, control variates, Longstaff-Schwartz, Sobol QMC), runs 10,000 paths by default (up to 50,000 paths and 500 steps on Premium), and is validated against Black-Scholes and binomial references. The tool is for understanding risk, not predicting markets or generating trading signals.
Monte Carlo in Options Simulator runs over the same seven stochastic models that drive the simulated market — from GBM (Steady Drift) to SABR (Volatility Smile) — not as a separate add-on with different assumptions Portfolio risk analytics (Analytics & Volatility → Monte Carlo) reports expected P/L, probability of profit, VaR and CVaR at 95% and 99%, worst and best case, a P/L fan and a terminal P/L distribution, and is rewind-aware The strategy stress test simulates a custom multi-leg strategy before it enters the portfolio, with options settling at their own expiration dates MC Price in the trade ticket prices a single option by Monte Carlo with Longstaff-Schwartz early exercise, a 95% confidence interval and the difference versus the chain price; it runs on demand rather than for the whole chain at every candle The underlying evolves under real-world drift while options are repriced risk-neutrally at each horizon — a technically correct separation The engine is validated: a European put Monte Carlo price of 5.54 agrees with Black-Scholes 5.57 to within about half a percent, and an American LSM put of 6.10 matches a binomial reference of about 6.09 Variance reduction uses antithetic variates, control variates, the Longstaff-Schwartz method for early exercise, and Sobol quasi-Monte Carlo sequences The default run is 10,000 paths with one time step per day of horizon; Premium allows up to 50,000 paths and 500 steps; the tool is educational and does not predict markets or generate trading signals From One Path to Thousands
Until now, Options Simulator has shown you one possible future. You generate a market, press play, and watch a single sequence of candles unfold — a concrete "what if this happens?" scenario. Monte Carlo simulation changes the question entirely. Instead of one path, it generates thousands of possible futures at once (10,000 by default) and computes the probability distribution of outcomes — moving you from "what if?" to "what is the probability?"
This is the final article in our three-part Monte Carlo series. The first article covered the method's foundations; the second covered its applications in pricing and risk. Here we show exactly how Options Simulator implements Monte Carlo — what it does, how it connects to the rest of the platform, and, just as importantly, what it does not claim to do.
The most important thing to understand up front: Monte Carlo in Options Simulator is not a separate mathematical add-on bolted on the side . It runs over the same seven stochastic price models that already drive your simulated market. It is a natural extension of the same logic into the probabilistic dimension.
Built on the Same Models You Already Use
When you choose a price model in Model Settings , you are configuring how your simulated market behaves. The seven stochastic models appear in the simulator under their in-app names:
Tier Mathematical model In-app name Chain pricer
Essential Geometric Brownian Motion (GBM) Essential 1 (Steady Drift) Black-Scholes Essential Merton Jump Diffusion Essential 2 (Sudden Shock) Black-Scholes Advanced Dynamics Heston stochastic volatility Advanced Dynamics 1 (Shifting Volatility) Bjerksund-Stensland Advanced Dynamics Fractional Brownian Motion (fBm) Advanced Dynamics 2 (Deep Tide) Bjerksund-Stensland Professional Grade GARCH Professional Grade 1 (Calm & Storm) Binomial Tree Professional Grade Bates (stochastic volatility + jumps) Professional Grade 2 (Perfect Storm) Binomial Tree Professional Grade SABR Professional Grade 3 (Volatility Smile) Binomial Tree
Monte Carlo uses that same model and same parameters to generate its thousands of paths. If you set up a Heston (Shifting Volatility) market, the Monte Carlo engine simulates Heston paths; if you choose Bates (Perfect Storm), it simulates Bates paths with stochastic volatility and sudden jumps. When you load an S&P 500 ticker, its real price history becomes the starting point of your simulation, and the volatility of the generated market — and therefore of the Monte Carlo paths — is derived from that underlying. You can see the link directly in Analytics & Volatility → IV Analysis : our simulated AAPL market showed a base implied volatility of about 17–19%, in line with AAPL's own history, while a more volatile stock such as TSLA starts from a correspondingly higher level under the same model and settings.
This consistency matters. A common weakness of bolt-on risk tools is that they use a different, simpler model than the one driving the analysis — producing risk numbers that don't match the market you're actually testing in. Options Simulator avoids this by design: the market you see and the Monte Carlo paths that assess it share the same mathematical engine. (For a deep dive into the seven models and three pricers, see The Mathematics Behind the Simulation .)
Use Case 1: Portfolio Risk Analytics
The first and most powerful application reads your live portfolio — every stock and option position you currently hold — and simulates how it might evolve. You find it under Analytics & Volatility → Monte Carlo : pick a horizon (Auto runs to your nearest option expiration; 1W, 1M, 3M, 6M and 1Y are also available) and press Run Monte Carlo . From 10,000 simulated futures, it produces a complete risk picture across three tabs:
P/L fan — your portfolio's profit/loss over the chosen horizon, day by day, drawn as percentile bands: the median line, the 25–75% band, and the wider 5–95% band. The widening fan shows how uncertainty grows with time.VaR and CVaR at 95% and 99% — Value at Risk tells you the loss threshold you shouldn't exceed at each confidence level; Conditional Value at Risk tells you the average loss in the worst scenarios beyond that threshold. (For the full explanation of these metrics, see the previous article .)Expected P/L, probability of profit, worst and best case — shown alongside VaR and CVaR on the Risk metrics tab.Distribution — a histogram of terminal P/L across all paths (losses in red, gains in green), revealing the true shape of your risk: where the mass of likely results sits, and how heavy the loss tail is.
The P/L fan: median and percentile bands of portfolio profit/loss over the horizon (10,000 paths, Perfect Storm model) Here is what that looks like on a real simulated portfolio: 100 shares of AAPL, two long December 345 calls and one short December 330 put, on a Perfect Storm (Bates) market, with the nearest expiration 49 days away. The run shown in the screenshots reported an expected P/L of +$223, a probability of profit of 40.3%, a 95% VaR of $7,273 and a 99% VaR of $10,100, a 95% CVaR of $9,064 and a 99% CVaR of $11,400, with a worst simulated outcome of −$13,800 and a best of +$24,400. The distribution shows a tall pile of moderate losses just below zero (the long calls losing time value if the stock goes nowhere) and a long right tail of gains — the asymmetric shape you would expect from a bullish, call-heavy position. Note what the two headline numbers say together: the average outcome is slightly positive, yet most paths lose a little. That gap between expected value and probability of profit is exactly the kind of insight a single price path can never show you.
Crucially, this analysis is rewind-aware . If you step back in time within your simulation, the risk analytics recompute for the portfolio state as it was at that candle — not the final state. Rewinding our example by five trading days, for instance, automatically lengthened the horizon from 49 to 56 days and recalculated every metric from the earlier prices. This lets you ask "what did my risk look like at the moment I opened this position?" and learn from the answer.
Risk metrics and terminal P/L distribution from the same 10,000-path run Use Case 2: Strategy Stress Testing
The second application works at the strategy level, before a position ever enters your portfolio. Add legs to the Custom Strategy panel with the + buttons in the option chain, then press Run MC Stress Test . The Monte Carlo engine takes the individual legs and simulates how the entire structure would behave across 10,000 possible market futures, with the same Risk metrics, P/L fan and Distribution views as the portfolio analysis.
This is an ad-hoc "what would this strategy do?" analysis. You construct an iron condor, a calendar spread, or any custom multi-leg position, and the engine shows you its outcome distribution before you commit. You see the probability of profit, the range of outcomes, and the tail risk — all while the position is still hypothetical.
For example, a short iron condor on simulated AAPL with 65 days to expiration (short 335 put / 360 call, long 320 put / 375 call, $705.85 credit) produced a strongly bimodal distribution: a bar of paths at the maximum loss of about $794 and a taller one at the full credit, with little in between. The probability of profit was 60.2% and the expected P/L +$108, while the 95% and 99% VaR and CVaR all equalled the maximum loss of $794.15. That is the honest profile of a defined-risk credit strategy: you win more often than you lose, but when you lose, you lose the whole width minus the credit — and the Monte Carlo run shows both facts at once instead of letting the 60% headline hide the tail.
An important technical detail makes this realistic: when the simulation reaches an option's expiration, that option settles at its own expiration — its intrinsic value is locked in, and it does not continue tracking the spot price afterward. This correctly models how a real multi-leg strategy behaves when its legs have different expiration dates, such as in calendar and diagonal spreads.
Strategy stress test: a short iron condor's bimodal outcome distribution and risk metrics before it enters the portfolio Use Case 3: Checking a Single Option's Price
The third application zooms in on one option . The option chain is calculated at every candle with fast pricers — Black-Scholes, Bjerksund-Stensland or a binomial tree, depending on the model tier — because hundreds of strikes have to update smoothly during playback. When one specific contract matters to your trade, you can price it independently: open the trade ticket for that option and press MC Price . The engine simulates thousands of paths for that single contract and returns:
A Monte Carlo price , with American-style early exercise handled by the Longstaff-Schwartz (LSM) method
A 95% confidence interval — how precise the estimate is with the number of paths used
The difference versus the chain price , so you see immediately whether the two methods agree
In the screenshot below, a December 345 call on the same simulated AAPL market (65 days to expiration) is quoted at $11.30 in the chain. MC Price returned $11.1353 ± $0.3201 from 10,000 paths — a difference of −$0.16, comfortably inside the confidence interval. In other words, the two independent methods agree within the statistical precision of the simulation, which is exactly what you want to see.
MC Price in the trade ticket: a Monte Carlo price with its 95% confidence interval and the difference versus the chain How should you read the result? If the chain price sits inside the confidence interval, the difference is just sampling noise — there is no hidden mispricing to exploit. If it falls well outside, the two methods are capturing something differently (for example, early-exercise value on a deep in-the-money put), and that is worth understanding before you trade. Either way, the confidence interval keeps you honest: a Monte Carlo price without one is a number with unknown precision.
This is deliberately an on-demand tool, not a replacement for the chain. Running full Monte Carlo on every strike and expiration at every candle would make real-time playback far too slow; fast pricers for the live chain plus Monte Carlo verification for the one option you care about is the sound trade-off.
The Technical Honesty: Real-World Drift vs. Risk-Neutral Pricing
Here is a subtle but important point that separates a rigorous implementation from a naive one. When Options Simulator runs portfolio Monte Carlo, it treats two things differently — and deliberately so:
The underlying asset evolves under the real-world drift of the chosen model. This reflects how the asset might actually move — its genuine expected return and volatility dynamics. This is the right choice for assessing the realistic distribution of future portfolio values.
The options are repriced risk-neutrally at each future horizon. As explained in the previous article , risk-neutral valuation is the only arbitrage-free way to price an option. At each simulated future point, the option's fair value is computed using risk-neutral dynamics, even though the underlying path that led there was generated under real-world drift.
This separation is not a contradiction — it is the technically correct approach. Mixing them up (pricing options under real-world drift, or evolving the underlying risk-neutrally for risk assessment) is a common error that produces subtly wrong risk numbers. Options Simulator keeps them properly distinct: real-world dynamics for where the asset goes , risk-neutral dynamics for what the option is worth when it gets there .
💡 Why this matters: A portfolio risk simulation should answer "given how this asset actually tends to move, what is the distribution of my portfolio's value?" — while still pricing each option correctly at every step. Conflating the two measures is one of the most common mistakes in homemade risk tools.Validation: How We Know It's Accurate
A Monte Carlo engine is only trustworthy if it reproduces known answers where those answers exist. Options Simulator's implementation is validated against analytical and lattice references:
Test Case Monte Carlo Result Reference Reference Method
European put 5.54 5.57 Black-Scholes (analytical) American put (LSM) 6.10 ~6.09 Binomial Tree (lattice)
The European result agrees with Black-Scholes to within about half a percent — as it should, since both price the same instrument and the only difference is Monte Carlo's sampling error. The American result, computed with the Longstaff-Schwartz Least Squares Monte Carlo method (which handles early exercise), matches the binomial tree reference closely. These agreements are the evidence that the engine is implemented correctly: it is validated against the established methods, not asserted to be accurate.
Under the hood, the engine uses textbook variance reduction techniques to achieve smooth, reliable distributions efficiently: antithetic variates , control variates , Longstaff-Schwartz for American-style early exercise, and Sobol quasi-Monte Carlo sequences. Every run reports its settings at the bottom of the panel — for example "Paths: 10 000 · Steps: 49 · Horizon: 49d", with one time step per day of horizon. The default run uses 10,000 paths — enough for smooth distributions — and the Premium plan allows up to 50,000 paths and 500 time steps, with built-in safety caps on the total computation (paths × steps) so a single calculation never overloads the system.
You can also see the sampling error for yourself. Running the same portfolio twice gave a 95% VaR of $7,273 and then $7,266, and a probability of profit of 40.3% and then 39.9%. Differences of this size are normal for 10,000 paths; if two runs disagree by much more, something other than sampling is going on.
Connected to the Whole Platform
What makes the implementation genuinely useful is how tightly it integrates with the rest of the simulator:
With generated data: Monte Carlo paths come from the same model and parameters set for your market, including volatility and starting price derived from real history when you load an S&P 500 ticker.With the option chain: The engine takes volatility from the live chain — including per-strike implied volatility — so it respects the volatility smile. When you run a SABR (Volatility Smile) market, the smile structure flows through into the Monte Carlo analysis. Time to expiration is computed from the real expiration date and the current simulation date, never from a stale relative value.With your strategies: Each leg is repriced independently, and options settle correctly at their own expirations.With your portfolio: The live portfolio drives the risk analytics, with full rewind-awareness.
What It Is — and What It Isn't
This deserves to be stated plainly, because it is easy to misunderstand a probability tool.
Monte Carlo in Options Simulator is an educational tool for understanding risk. It shows you the range and likelihood of outcomes under a chosen model's assumptions. It helps you see the shape of your risk — especially the tail — before you would ever face it with real money.
It is not a prediction of the market. The simulated paths are not forecasts of where prices will actually go. It is not a trading signal. A "70% probability of profit" is a statement about a model under its assumptions — not a recommendation to trade, and not a guarantee of any outcome. The value is in building intuition about probability and risk, not in being told what to do.
Used this way — as a lens for understanding the distribution of outcomes rather than a crystal ball — Monte Carlo is one of the most powerful learning tools available to an options trader. It turns the abstract idea of "risk" into concrete, visual, explorable numbers.
Availability
The complete Monte Carlo suite — portfolio risk analytics, single-option Monte Carlo pricing and the strategy stress test — is included in the Premium and Lifetime plans, with up to 50,000 paths and 500 time steps per run. The stress test works from the Custom Strategy Builder, which is also a Premium feature. See the pricing page for the current breakdown across all plans.
What This Means for Your Trading
Test your risk before you take it. Run portfolio risk analytics on any position to see its VaR, CVaR, and outcome distribution. Understand the tail before the market shows it to you.Stress-test strategies while they're still hypothetical. Use the strategy stress test to see an iron condor's or calendar spread's full outcome distribution before committing.Verify the prices that matter. When one specific option is central to your thesis, press MC Price in the trade ticket to check it — with a confidence interval that tells you how precise the estimate is.Think in probabilities, not predictions. Let the distributions train your intuition. The goal is not to be told the future, but to understand the range of what could happen and how likely each outcome is.
Try Monte Carlo Risk Analysis
Frequently Asked Questions
Does the Monte Carlo feature predict where the market will go?
No. It generates a distribution of possible outcomes based on your selected model's assumptions — it does not forecast actual prices or provide trading signals. A probability of profit figure describes the model's behavior under its assumptions, not a guarantee or recommendation. It is an educational tool for understanding the range and likelihood of outcomes.
Does Monte Carlo use the same model as my simulated market?
Yes. This is a core design principle. The Monte Carlo engine runs over the same seven stochastic models — from GBM (Steady Drift) and Heston (Shifting Volatility) to Bates (Perfect Storm) and SABR (Volatility Smile) — and the same parameters that drive your simulated market. It is not a separate tool with different assumptions — it is an extension of the same engine into the probabilistic dimension.
How accurate is the Monte Carlo pricing?
It is validated against established references. A European put priced by Monte Carlo (5.54) agrees with the Black-Scholes analytical price (5.57) to within about half a percent, and an American put via Longstaff-Schwartz (6.10) matches the binomial tree reference (~6.09). MC Price in the trade ticket also reports a 95% confidence interval and the difference versus the chain price, giving an honest measure of its statistical precision.
Why does the underlying use real-world drift but options use risk-neutral pricing?
This separation is technically correct. To assess realistic portfolio risk, the underlying should evolve under real-world dynamics (how the asset actually tends to move). But options must always be priced risk-neutrally, because that is the only arbitrage-free valuation. So at each simulated future point, the asset's position comes from real-world drift, while the option's value at that point is computed risk-neutrally. Conflating the two is a common error that Options Simulator deliberately avoids.
How many simulation paths does it use, and which Monte Carlo techniques?
The default is 10,000 paths, with safety caps on total computation (paths × steps). The engine uses antithetic variates, control variates, the Longstaff-Schwartz method for American-style early exercise, and Sobol quasi-Monte Carlo sequences — standard, validated variance reduction techniques. Each run uses one time step per day of the horizon, and the Premium plan allows up to 50,000 paths and 500 steps.
References & Sources
Boyle, P.P. (1977). "Options: A Monte Carlo Approach."
Journal of Financial Economics , 4(3), 323-338.
DOI
Longstaff, F.A., Schwartz, E.S. (2001). "Valuing American Options by Simulation: A Simple Least-Squares Approach."
Review of Financial Studies , 14(1), 113-147.
DOI
Rockafellar, R.T., Uryasev, S. (2000). "Optimization of Conditional Value-at-Risk."
Journal of Risk , 2(3), 21-41.
DOI
Glasserman, P. (2003). Monte Carlo Methods in Financial Engineering .
Springer-Verlag. ISBN 0-387-00451-3.
Link
Heston, S.L. (1993). "A Closed-Form Solution for Options with Stochastic Volatility with Applications to Bond and Currency Options."
Review of Financial Studies , 6(2), 327-343.
DOI
Try It Yourself
Ready to move from one path to thousands? Run Monte Carlo risk analytics on your own portfolio and see the full distribution of possible outcomes — with virtual money and zero risk.
Open Simulator
What to Read Next
Start the series: Monte Carlo Methods Explained
Previous: Monte Carlo in Options Trading — Pricing, Greeks, and Risk Analysis
Related: The Mathematics Behind the Simulation — 7 Price Models and 3 Option Pricers Explained
Frequently asked questions Does the Monte Carlo feature predict where the market will go? No. It generates a distribution of possible outcomes based on your selected model's assumptions — it does not forecast actual prices or provide trading signals. A probability of profit figure describes the model's behavior under its assumptions, not a guarantee or recommendation. It is an educational tool for understanding the range and likelihood of outcomes.
Does Monte Carlo use the same model as my simulated market? Yes. This is a core design principle. The Monte Carlo engine runs over the same seven stochastic models — from GBM (Steady Drift) and Heston (Shifting Volatility) to Bates (Perfect Storm) and SABR (Volatility Smile) — and the same parameters that drive your simulated market. It is not a separate tool with different assumptions — it is an extension of the same engine into the probabilistic dimension.
How accurate is the Monte Carlo pricing? It is validated against established references. A European put priced by Monte Carlo (5.54) agrees with the Black-Scholes analytical price (5.57) to within about half a percent, and an American put via Longstaff-Schwartz (6.10) matches the binomial tree reference (~6.09). MC Price in the trade ticket also reports a 95% confidence interval and the difference versus the chain price, giving an honest measure of its statistical precision.
Why does the underlying use real-world drift but options use risk-neutral pricing? This separation is technically correct. To assess realistic portfolio risk, the underlying evolves under real-world dynamics (how the asset actually tends to move). But options must be priced risk-neutrally, the only arbitrage-free valuation. So at each simulated point, the asset's position comes from real-world drift, while the option's value is computed risk-neutrally. Conflating the two is a common error the simulator avoids.
How many simulation paths does it use, and which Monte Carlo techniques? The default is 10,000 paths, with safety caps on total computation (paths × steps). The engine uses antithetic variates, control variates, the Longstaff-Schwartz method for American-style early exercise, and Sobol quasi-Monte Carlo sequences — standard, validated variance reduction techniques. Each run uses one time step per day of the horizon, and the Premium plan allows up to 50,000 paths and 500 steps.
