Brazil market snapshot
Is the market cheap? Stocks against cash
Why this reading
Did it matter? What the next three years paid
Each dot is a month since 2011: how far the earnings yield stood above cash after inflation, and how much every eligible stock (equal weights, dividends reinvested) beat or trailed the CDI per year over the next three years.
By how cheap stocks were, in thirds
The same months, split into thirds by the gap at the start.
This week's shortlist
Full screener →Magic Formula baseline (Greenblatt: earnings yield + return on capital):
Screener
Every listed company, scored on four pillars, each a percentile within the eligible universe (0–100; blue above the median, red below). Growth is shown, not scored. Click a column to sort, a row for why it scores what it does, a ticker for its history.
Backtest: buy the list once a year and hold
Loading…
R$100 invested in May 2012
Does the score separate winners from losers?
Each May, every eligible company split into fifths by each score (Q1 = best), each fifth held as an equal-weight portfolio for the year. Bars: return per year, 2012–today (growth: 2013–today, it needs three years of statements). A working factor steps down from Q1 to Q5; the gray line is all eligible stocks. Growth failed this test, so it is shown but not counted in the score.
Year by year
Bought on the first session of May, held twelve months, dividends reinvested. Cells shaded by return.
Buy once, hold for 1, 3 or 5 years
Each year's list bought and never touched again. * = still running (return to date).
What the backtest held
The top 10 by score each May, and how each did over the following year.
How much to trust the backtest
- Point in time: each list uses only quotes, filings and events public on its date, through the same code as the live screen; share counts are the ones filed at the time.
- Not in the numbers: brokerage, taxes, slippage. A yearly rebalance trades little, but taxes on gains are real.
- Survivors: companies that went bankrupt or into recovery stay in, and their quotes follow them into the special segments. Companies that delisted keep their quotes but lose their dividends (B3 only serves listed companies' history), which slightly understates returns.
- Approximations before 2020: share counts come from the yearly Reference Form, dated by when CVM received it; registration status is today's, so it is not used as a filter in the past.
- One path: fifteen yearly decisions are a small sample. A strategy that beat the market here can lag for years; compare it with "all eligible stocks" to see what the picking itself added.
Track record
How it is kept
- Every Monday the top 10 by score (and by the Magic Formula) is appended to
data/picks/picks.csvwith its prices. Nothing is ever removed, and a recorded list is never replaced (a re-run on the same session keeps the first). - Each list is held as an equal-weight portfolio until the next one, on dividend- and split-adjusted prices, against BOVA11 over the same days.
- Lists are stamped with the scoring rules' fingerprint: a rule change ends one record and starts another, it never rewrites the past.
The live lists against BOVA11
Start = 100.
Every list so far
Return to date of each list (composite), vs BOVA11 over the same days.