Launching Simulation Lab — 50k-iteration Monte Carlo for tax teams

Why we built it

Cross-border tax decisions are made under uncertainty. The effective tax rate of a proposed structure depends on a chain of inputs — treaty rates, withholding outcomes, transfer-pricing ranges, Pillar Two top-ups, jurisdictional ETRs, FX assumptions, regulatory stability — and each of those inputs is itself a distribution, not a number. The standard tool for modeling this on the tax-function's desk is a deterministic Excel cascade with one or two sensitivity pages. It is good for a base case. It is not good for the question the CFO is actually asking, which is: what is the range of outcomes, and what is the probability of each?

Our enterprise customers have been doing Monte Carlo modeling on the side, in spreadsheets and bespoke Python notebooks, for years. The work is good. The tooling is poor: every analysis is rebuilt from scratch, the assumption library is held in an analyst's head, and the results are not reusable across transactions. After the third request from a customer asking whether we could productize this, we did.

Simulation Lab is the result. It is a 50,000-iteration Monte Carlo engine that runs against the same verified jurisdiction and treaty data that powers the rest of FiscalEyes, and produces board-ready output: distribution charts, percentile ETRs, sensitivity rankings, and the underlying assumptions in a form that survives audit.

What it does

Simulation Lab takes a defined structure — a chain of legal entities, a set of payment flows, an income-source map — and runs 50,000 simulated outcomes against:

• Jurisdictional ETR distributions.Each jurisdiction's effective rate is modeled as a distribution rather than a point, calibrated against the statistical history of that jurisdiction's headline rate, deductions, and elective regimes.
• Treaty outcome probabilities. Treaty positions that depend on PPT, LoB or beneficial-ownership analyses are modeled with assigned outcome probabilities, based on our internal database of recent decisions in the relevant jurisdictions.
• Transfer-pricing inter-quartile ranges.Where TP outcomes drive the result, the IQR is sampled rather than centered, with a tail for adjustment risk.
• Pillar Two top-up scenarios. The full QDMTT/IIR/UTPR cascade is modeled, with safe-harbour eligibility evaluated in each iteration.
• Regulatory drift. Jurisdictions modeled with announced or expected rate changes apply the change in a stochastic share of iterations consistent with our legislative-tracking confidence.

The output is a distribution of consolidated effective tax rates over the modelled horizon, with percentile reporting (p5, p25, p50, p75, p95), per-jurisdiction contribution analysis, and sensitivity rankings showing which inputs are driving the variance. The whole thing runs in 30–90 seconds depending on the structure complexity.

How customers are using it

Three patterns have emerged in the eight months since the private beta. Each represents a question that was previously answered with a deterministic point estimate and is now answered with a distribution.

1. M&A pre-deal modeling

For deal teams, Simulation Lab has become the standard tool for the post-acquisition tax model. Where a typical deal memo previously contained a single “projected effective tax rate” line, the new format contains a percentile band: “p25 ETR 21.4%, p50 22.7%, p75 24.1%”. The bid teams have found this materially more useful than a point estimate, because it lets them price uncertainty into the model — and lets them defend a higher bid where the tax variance is genuinely low.

2. Internal restructuring

For groups considering a structural change (an IP migration, a holding-company relocation, a principal-company consolidation), Simulation Lab models the post-restructuring ETR distribution under varying counterfactuals. The most useful output here is the sensitivity ranking: which input moves the ETR most? In nearly half of the structures we have modeled in 2025, the dominant variance driver was not the restructuring decision itself but a single transfer-pricing assumption — an answer that would have been invisible in a deterministic model.

3. Capex location decisions

For groups choosing between two or three jurisdictions for a new manufacturing or R&D site, Simulation Lab compares the multi-year ETR distributions under each option, including the tax-incentive regimes and their conditionality. This is the use case that has generated the most board-level engagement: a CFO who can show the board a probability-weighted comparison of three site options is having a different conversation from one who is presenting three point estimates.

Audit trail and assumption discipline

A Monte Carlo engine that produces a distribution without an audit trail is unusable in a regulated environment. Every iteration in Simulation Lab is reconstructable: the assumption set, the random seeds, the per-jurisdiction draws, the cascade outcome, the contribution to the consolidated ETR. Customers can lock an assumption set, version it, and re-run the simulation against it months later for an audit response. The engine is deterministic given a fixed seed.

The assumption library itself is curated. Distribution parameters for each jurisdictional ETR, treaty outcome and Pillar Two scenario are maintained by our research team and versioned monthly. Customers can override any parameter, and every override is logged with attribution. The default assumption set is the one we would defend in front of a tax authority, and it is the one we use ourselves when we consult on customer projects.

Pricing and access

Simulation Lab is included in the FiscalEyes Enterprise tier at no additional cost, and Professional-tier customers can access it on a per-simulation basis. The fastest way to try it is to create a free account — Simulation Lab ships with a sample structure preloaded so you can run your first 50,000-iteration simulation before you wire up your own data.

The bottom line

Tax decisions are decisions under uncertainty. The standard tooling treats them as decisions under certainty and produces answers with false precision. Simulation Lab is our attempt to put the uncertainty back into the model — at enterprise scale, with the audit discipline that a regulated environment requires. It is the tool we wish had existed when we were on the customer side; we are pleased it now does.