Five-course specialization plus a peer-graded capstone. 150+ hours. Covers deterministic, probabilistic, and optimization modeling. Culminates in a max-Sharpe portfolio strategy built from raw monthly returns.
VerifyI can build, in Excel, end-to-end, no add-ins, the three families of quantitative models companies actually use to make money decisions: deterministic forecasting, probabilistic risk simulation, and constrained optimization. I've calibrated a defensible portfolio strategy on real return data, benchmarked it against a real single-stock concentration, and produced both the workbook and the deck that defended the result.
Why it matters. Most business questions look messier than they are because people skip the modeling-type question. Before you reach for a tool, you have to ask whether you are forecasting (deterministic), pricing in uncertainty (probabilistic), or making a choice under constraints (optimization). That triage is the whole foundation of the rest of the specialization.
What I learned applying it. When I size a tuck-in acquisition, the revenue forecast is deterministic, the synergy realization is probabilistic, and the payment-mix decision is optimization. Same workbook, three different model types. Naming them correctly is what stops the model from quietly contradicting itself.
Why it matters. Most analysts can do NPV. Few can structure a workbook that survives more than one analyst's tenure. This course is about discipline. Input, calculation, and output separation. Named ranges. Single source of truth for assumptions.
What I learned applying it. My deal underwriting workbooks now have one input tab, several calculation tabs, and one output tab. When the lender asks "what if we re-flex pricing 50 basis points," I change one cell, not seventeen. Structural discipline is the difference between a model the sponsor trusts and a model that becomes obsolete the day the analyst who built it leaves.
Why it matters. Single-point estimates are wrong almost always. The real question is by how much and in which direction. Monte Carlo simulation forces the model to answer that, in Excel, with no add-ins. Distributions for the inputs, thousands of trials, a confidence band for the output.
What I learned applying it. The point is not precision. It is communicating uncertainty in a way that does not get rounded back to a single number on the way to the investment committee. When I deliver a deal model with a risk envelope around NPV, the conversation shifts from "is this number right" to "are we comfortable with the spread." Different conversation, better outcomes.
Why it matters. Most operating decisions are constrained optimization problems that we solve with intuition. Solver lets you solve them properly. Objective function, decision variables, constraints, and one number out the back end that tells you the best you can do given what you have.
What I learned applying it. The shadow price of a constraint is where the negotiation should happen. If labor is binding and the shadow price is high, that is the line item to argue about. I built a product-mix model that surfaced two service lines we were over-investing in. The intuition said keep them. The model said reallocate. The model was right.
Why it matters. The capstone is the proof that the framework works on real data, against a real benchmark, with a real outcome. I calibrated monthly returns from 2012 through 2015 for two Vanguard funds, solved for max-Sharpe weights with Excel Solver (roughly 68 percent VBTLX, 32 percent VFIAX), and benchmarked the result over the first half of 2016 against a single-stock AAPL concentration on a 5 million dollar base.
What I learned applying it. The two-asset portfolio that no retail investor would touch beat the AAPL concentration on both return and variance over the test window. Diversification is boring and correct. When the model and the gut disagree, calibrate the model harder before overriding it. The 68/32 split looked conservative. It also produced the better risk-adjusted return.