Model the math on a decision in minutes, not hours of spreadsheet work.
Most people who lose money on predictions do not lose because they lacked information. They lose because they never converted their beliefs into numbers. They had a hunch, they liked the story, and they sized the position by feel. The Probability Engine lab is built to close that gap: you work through expected-value analysis, Kelly sizing, and Monte Carlo simulation on real decisions, with an AI instructor that corrects your reasoning as you go, not after you have already committed capital.
By the end of 16 hours of hands-on work, you will be able to:
Step 1 — You build the model, not watch someone build it
Each module puts a real decision in front of you: a Kalshi contract, a portfolio allocation thesis, a season-outcome forecast. You use the built-in tools — the probability workup template, the EV calculator, the simulation engine — to work through the problem. There is no passive video. The interface is a live environment.
Step 2 — The AI instructor listens and pushes back
As you work, the AI instructor monitors your inputs and reasoning. If you anchor on a round number probability without justification, it asks you to defend it. If your Kelly output implies a position size that ignores variance, it flags the assumption. The feedback is specific to what you actually typed, not generic hints.
Step 3 — You leave with output you can act on
At the end of each exercise you have a completed probability workup, a sized position or bet, and simulation output with confidence intervals. The certification is awarded on performance — the AI grades the quality of your reasoning, not a multiple-choice quiz. That means the credential reflects something you actually did.
All tools run in the browser. There is nothing to install.
| Probability Engine | A video course on probability or forecasting | |
| Format | Browser-based, hands-on environment | Passive video, typically with PDFs |
| Feedback | AI instructor corrects you in real time as you work | None during the lesson; forums at best |
| Tools included | Live EV calculator, Kelly tool, Monte Carlo engine, Kalshi feed | Usually screenshots or external spreadsheet links |
| Certification | Awarded on demonstrated performance | Typically awarded for completing watch-hours |
| Genuine strengths of the alternative | Video courses from experienced forecasters (Philip Tetlock's associated materials, for instance) offer deep conceptual grounding and are genuinely useful for building a mental model | — |
| Price | $39 one-time, or $19/mo for every lab | Varies; structured courses often $100–$500 |
Video courses taught by credible practitioners are worth watching. The honest gap is that watching someone run a Kelly calculation is not the same as running one yourself on a live market price while an instructor checks your inputs.
Do I need a statistics or math background to use this lab?
No. The lab is rated all levels. The AI instructor explains each concept at the point you encounter it, and the tools handle the arithmetic. What you need is the willingness to think carefully about numbers.
How does the Monte Carlo simulation work in the browser without any software installation?
The simulation engine runs entirely client-side. You set the input distributions and number of trials through a guided interface, and the engine returns a percentile table and distribution chart. You do not need Python, R, Excel, or any other tool.
Can I use the Kelly criterion calculator for investment portfolio sizing, not just sports betting?
Yes. The calculator is built for any decision with a quantifiable edge and variance. The exercises cover prediction market contracts, season-outcome bets, and equity position sizing. The underlying math is identical across those contexts.
How does the live Kalshi market feed work, and do I need a Kalshi account?
The lab pulls publicly available Kalshi contract prices into the comparison tool. You do not need a Kalshi account to use the feed for analysis inside the lab. If you want to place actual trades on Kalshi, that requires your own account on their platform, which is separate.
What does the AI instructor actually correct — does it just check whether my final answer is right?
The instructor monitors your reasoning at each step, not just the final output. If you assign a probability without decomposing the question, it asks how you got there. If your EV calculation treats correlated outcomes as independent, it flags the dependency. The corrections happen while you are working.
How is the performance-based certification different from a standard course completion certificate?
A completion certificate records that you watched or clicked through content. The Probability Engine certification is issued after the AI instructor evaluates the quality of your probability workups, your position sizing rationale, and your simulation interpretation across the exercises. You can review the evaluation criteria before you start.