Every backtester on the internet (Portfolio Visualizer and friends) simulates a make-believe "$X every month." Divvy is different: it replays the exact money you actually invested — the real dates, the real dollar amounts, straight from your broker — into whatever portfolio you're curious about. So instead of a hypothetical, you get your answer:

💡 "If I'd put the exact money I actually invested into this basket of ETFs/stocks instead, how much more — in dividends specifically, and in total — would I have made?"

Dividends get reinvested (DRIP) into the same holding, so it compounds exactly like a real brokerage account.

Why it exists

No off-the-shelf tool does this. They all assume a clean, synthetic contribution schedule. Real investing is lumpy — you skip months, you add lump sums, you rebalance. Divvy replays your actual cash-flow history against a counterfactual portfolio, and reports on dividend income as a first-class metric (not just total return) — the thing dividend investors actually care about.

✨ Features

• 🔁 Real-ledger replay — your actual contribution calendar, not a synthetic assumption
• 🧺 Compare any number of "buckets" (portfolios) side by side from simple YAML
• 💵 Dividends as a first-class metric — lifetime and trailing-12-month run-rate
• 📈 DRIP compounding, money-weighted return (XIRR), equity & dividend charts
• 🧪 Interactive Experiment Lab — a local web app to tweak weights with sliders and compare live
• 🚀 Zero-data quickstart — try it in 10 seconds with synthetic mode
• 🔌 Bring your own data — generic CSV, or a Fidelity ledger / 1099 importer
• 🔒 Private by default — your financial data never leaves your machine

Install

pip install divvy-backtest          # core CLI (import + command are 'divvy')
pip install 'divvy-backtest[ui]'    # + interactive Experiment Lab

Or run from source with uv: uv sync.

⚡ Quickstart (no data, no files)

Backtest a hypothetical "$500/month since 2019" into a couple of dividend baskets — define them inline with --holdings (weights are auto-normalized), so this runs from anywhere:

pip install divvy-backtest

divvy compare --synthetic-monthly 500 --synthetic-start 2019-01-01 \
  --holdings "SCHD=45,DGRO=25,VYM=15,SDY=15" \
  --holdings "SCHD=40,VYM=20,SDY=20,SPYD=20"

Prefer files? Pass --bucket path/to/bucket.yaml instead (see Define a bucket).

…and out comes a side-by-side comparison, plus equity & dividend charts in results/<date>/:

          variant  total_contributed  total_dividends  trailing_12mo_dividends  ending_value  total_return_pct  xirr_pct
dividend_etf_core             7200.0           764.12                   294.02      11641.27             61.68     12.47
  high_yield_tilt             7200.0           958.87                   336.69      11277.28             56.63     11.61

(illustrative output from the bundled example data)

🧪 Interactive Experiment Lab

Prefer sliders to flags? Launch the local web app and tweak portfolios live — add/remove tickers, drag weights, and watch the comparison table and charts update:

pip install 'divvy-backtest[ui]'   # or: uv sync --extra ui
uv run divvy ui

It opens in your browser (running 100% locally — no data leaves your machine): pick a contribution source, edit one or more portfolios in the sidebar, hit Run comparison, and get headline metrics, a sortable table, and interactive value & dividend charts.

Bring your own contributions

Any broker can give you a list of what you invested and when. Put it in a two-column CSV:

date,amount
2021-01-04,200
2021-02-01,200

divvy compare --contributions-csv my_contributions.csv \
  --holdings "SCHD=45,DGRO=25,VYM=15,SDY=15"

See examples/contributions.csv for a full sample.

Define a "bucket" (a candidate portfolio)

A bucket is just a YAML file of tickers and target weights (must sum to 1.0):

name: My dividend basket
weights:
  SCHD: 0.50
  VYM: 0.15
  SDY: 0.20
  ABBV: 0.15

Pass as many --bucket flags as you like to compare them side by side — or skip files entirely with inline --holdings "SCHD=45,DGRO=25,VYM=15,SDY=15" (repeatable, weights auto-normalized).

Compare against your actual Fidelity account

If you export your Fidelity transaction history CSVs, Divvy can auto-derive both your real contribution calendar and the real dividends you received, and add your actual account as a comparison row:

divvy compare --ledger path/to/fidelity_history_csvs/ \
  --holdings "SCHD=45,DGRO=25,VYM=15,SDY=15" \
  --real-value 12345.67 --real-as-of 2026-07-03

Import dividends from a Fidelity 1099 (optional)

To reconstruct the dividend income you actually received from a Consolidated 1099 PDF (as a comparison baseline):

pip install 'divvy-backtest[pdf]'
uv run divvy import-1099 --pdf 2025-Consolidated-1099.pdf --out dividends_2025.csv

Note: a 1099 records dividends received, not what you bought — so it can't drive a backtest on its own (that needs your contribution calendar). It's a baseline helper. The parser targets Fidelity's 1099 layout; other brokers differ.

What the numbers mean

Every comparison also includes an SPY benchmark row by default (disable with --benchmark none, or pick another ticker with --benchmark VTI), and a dividend-income-by-year chart so you can see income growth, not just a lifetime total.

Realism knobs (optional)

Model a taxable account, periodic rebalancing, and fund fees:

uv run divvy compare --contributions-csv my.csv --bucket buckets/mine.yaml \
  --dividend-tax-rate 0.20 \   # reinvest only after-tax dividends; adds a net-income column
  --rebalance annual \         # annual | quarterly | monthly (default: none, DRIP drifts)
  --expense-ratio 0.0006       # annual fund fee applied to every holding (0.06%)

All three are also available in the Experiment Lab under Realism (optional). Defaults leave behavior unchanged (tax-free DRIP, no rebalancing, no fees). The tax model is a flat-rate estimate — real dividend taxation (qualified vs. ordinary, brackets, state) is more nuanced.

Project a future income goal

How much must you invest monthly to reach a target dividend income? divvy project gives both a deterministic estimate and a Monte Carlo range (returns are random, not a single guess):

uv run divvy project --income 2000 --current-value 7500 --monthly 1000 --years 25

Outputs the required contribution across return assumptions, plus p10/p50/p90 ending-value and after-tax-income scenarios. Like all projections, these are assumption-driven ranges, not predictions.

⚠️ This is a backtester, not a crystal ball

Data sources

• Prices & dividend history: yfinance (free, no key), cached to data/cache/.
• Finviz Elite (optional): if you have a key, copy .env.example to .env and add it — used only for ad-hoc yield/screening lookups, not required for the core backtest.

🔒 Privacy

Your financial data never leaves your machine and is never committed: data/, results/, your personal buckets/, and .env are all gitignored. Only code and the fake examples/ data live in the repo.

Install / dev

uv sync --extra dev   # or: pip install -e '.[dev]'
uv run pytest

See CONTRIBUTING.md — new broker adapters and 1099 importers are especially welcome.

License

MIT — see LICENSE.