Do Kalshi NFL favorites win as often as their price says?
Collect every settled game market, bucket by closing price, and compare implied odds with what actually happened.
AI spreadsheet for trading research
Trading research you can check, cell by cell.
Ask Jordan a question about the markets. It pulls the data onto a sheet, writes the formulas, and shows you every change before it lands. You keep the model. Jordan never places a trade.
Private early access. Public sign-up opens soon.
E8fx
=C8-B8| A | B | C | D | E | |
|---|---|---|---|---|---|
| 1 | Price bin | Implied | Actual | Markets | Gap |
| 2 | 0.10–0.20 | 0.15 | 0.13 | 212 | −0.02 |
| 3 | 0.20–0.30 | 0.25 | 0.24 | 236 | −0.01 |
| 4 | 0.30–0.40 | 0.35 | 0.33 | 241 | −0.02 |
| 5 | 0.40–0.50 | 0.45 | 0.46 | 259 | +0.01 |
| 6 | 0.50–0.60 | 0.55 | 0.56 | 268 | +0.01 |
| 7 | 0.60–0.70 | 0.65 | 0.67 | 247 | +0.02 |
| 8 | 0.70–0.80 | 0.75 | 0.78 | 233 | +0.03 |
| 9 | 0.80–0.90 | 0.85 | 0.88 | 197 | +0.03 |
CalibrationMarketsChart
Start with a question you already ask
Each of these runs on public Kalshi market data, so it works before you connect an account.
What does the market think the Fed will do next?
Turn each outcome’s bid and ask into an implied probability, normalize them, and chart the distribution.
What’s priced into this week’s NFL games?
Moneylines, spreads and totals side by side, with the implied probability behind every price.
How it works
Ask in plain English
“Backtest buying NFL favorites under 70¢.” Reference a sheet or range with
@and Jordan reads the whole workbook for context.Jordan builds the sheet
It collects the market history onto a sheet, then writes formulas, tables and charts on top of it. The data lives in cells, not in the model’s memory.
You review the diff
Every edit arrives as a changeset you can preview cell by cell. Accept all, accept some, or reject. Accepted changes undo like any other edit.
Keep the model
Formulas stay live, so you can change an assumption and watch it flow through. Export to Excel whenever you like.
Why a spreadsheet, not a chatbot
A chat answer is a claim. A spreadsheet is evidence. If Jordan reports that favorites priced near 80¢ won more often than their price implied, you can click the cell, read the COUNTIFS behind it and trace it back to the rows it counted.
That matters most in trading research, where the expensive mistakes are quiet ones: a lookahead bias, a missing fee, a sample of twelve. Jordan puts the whole chain where you can audit it.
- The sheet is the dataset. Collected history is written to cells, so the analysis can use every row.
- Formulas, not pasted values. Derived numbers stay connected to their inputs.
- Excel-compatible. Import and export
.xlsx, with familiar functions and shortcuts. - Nothing changes without you. The agent proposes; you accept.
Read-only by design
Jordan is built to study markets, not to trade them.
No trading, ever
The connection layer allows a fixed list of read operations and refuses everything else before a credential is loaded. No setting, prompt or argument can extend it.
Your keys never reach the AI
Credentials sit in a separate service backed by Google Secret Manager. The agent asks for a named operation and receives data, never a key or token.
Public data needs no account
Market data, price history and settlements come from Kalshi’s public API. Connect your own account only if you want Jordan to analyze your positions and fills.
Learn the method
All articles →Plain-English guides to the ideas behind good trading research.
Foundations · 5 min read
What is algorithmic trading? A plain-English guide
Algorithmic trading means making trading decisions with explicit, testable rules. Here is how it works, the main strategy families, and how to start researching without writing code.
Backtesting · 6 min read
How to backtest a trading strategy without fooling yourself
A step-by-step backtesting method, the six biases that make bad strategies look good, and the spreadsheet layout and formulas to run one honestly.
Prediction markets · 5 min read
How prediction market prices work, and how to read them as probabilities
A Kalshi contract that pays $1 trading at 62¢ implies about a 62% chance. Here is how bids, asks and multi-outcome markets turn into clean implied probabilities, with formulas.
Prediction markets · 5 min read
Market calibration: do favorites win as often as their price says?
How to test whether prediction-market prices are well calibrated, with price bins, a reliability chart, the Brier score and the sample-size check most studies skip.
Risk and sizing · 4 min read
Expected value and edge: the math behind every profitable trade
How to compute the expected value of a binary event contract, include fees and spreads, find your breakeven probability, and tell edge apart from variance.
Risk and sizing · 4 min read
Position sizing with the Kelly criterion (and why most traders use a fraction of it)
The Kelly criterion tells you what fraction of your bankroll to risk on a bet with an edge. Here is the formula for binary event contracts, a worked example, and why fractional Kelly is the practical choice.
Questions
What is Jordan?
Jordan is a spreadsheet with an AI agent built in. You describe the analysis you want, for example a backtest or a calibration study of prediction-market prices, and Jordan collects the data onto sheets, writes the formulas and charts, and shows you each change for review before it is applied.
Can Jordan place trades for me?
No. Jordan is read-only by design. Trading operations are not part of the product: the connection layer only allows a fixed list of read operations, and it refuses everything else before any credential is loaded.
What data can Jordan use?
Public Kalshi market data (markets, events, series, price history) works without an account. If you connect your own Kalshi account, Jordan can also read your balance, positions, fills and settlements so it can analyze your own trading.
Does the AI see my exchange credentials?
No. Credentials are stored in a separate service backed by Google Secret Manager. The AI agent asks that service for a named operation and gets data back; it never receives a key, token or account number.
Do I need to know how to code?
No. If you can read a spreadsheet, you can check Jordan’s work. Every number lives in a cell, and every derived number is a formula you can inspect, edit or export to Excel.
Is Jordan financial advice?
No. Jordan is a research tool. It helps you test ideas against data. It does not recommend trades, and a backtest or calibration study describes the past, not the future.
How do I get access?
Jordan is in private early access and public sign-up is closed for now. Request early access on this page and we will reach out when a spot opens.
Get early access
Jordan is in private early access. Public sign-up opens soon. In the meantime, start with the Learn guides.