aprise

Backtest

Run a screen through the past. On each rebalance date the backtester rebuilds every company’s screener figures from what had been filed before that day and its closing prices, applies your filters and ranking with the screener’s own code, and holds the top-ranked names until the next rebalance, net of trading costs. The results are hypothetical.

How this backtest works

Point-in-time. On every date the screener’s figures are rebuilt from what was knowable that day: fundamentals as they stood in the SEC filings available by then — as first reported until a restatement was filed, restated after that — with each filing first used on the trading day after it was filed, because a filing can land after the close. Price is the close as it traded that day, and share counts are put on that same day’s basis across splits, so market value and every ratio compare like with like. Returns are measured on closes adjusted for splits and dividends. Nothing published later reaches an earlier date.

Next-close execution. Names are picked at a signal day’s close and traded at a later close — one trading day later by default, the execution lag — and they earn returns from the day after they are bought. Between rebalances the weights drift with prices, and cash earns nothing.

Costs. Every rebalance is charged the cost you set, in basis points of the value traded — everything bought plus everything sold — on the day it trades. Turnover is reported one-way, half of that traded value, so a rebalance that turns over 50% of the portfolio at 10 bp costs 10 bp. There is no separate bid-ask spread or market-impact model.

Missing and broken prices. A pick with no price on its trading day is skipped and the next-ranked name takes its place; the portfolio holds fewer names only when the ranking runs out. A holding whose trading is halted earns nothing until it resumes, and a rebalance that falls in the halt cannot trade it, so it keeps its drifted weight and the new picks share what is left. A holding whose price history ends is closed at a 0% return and held as cash until the next rebalance, and the result says what a total loss there would have done instead. A stored history that breaks — the same ticker splicing in a different security — is closed the same way on the day it breaks. A one-day move of more than threefold up or two-thirds down that no corporate event explains is kept as stored, and any result that held the name through it lists it.

Universe. On each date, U.S.-listed companies worth at least $50M with a median of at least $100K a day in dollar volume over the previous 60 sessions. It is built from the companies listed today, so those delisted earlier are missing.

Short interest and 20-day average volume have no stored history, so screens that use them cannot be backtested yet.

Each preset fills in a screen and portfolio settings you can change. Starting points, not recommendations.

Screen
Portfolio

Leave max weight empty for no cap.

Period and costs

Stored history: — to —.

Subtracted from returns before Sharpe, Sortino, beta and alpha are computed.

Administrator access

Kept in this tab’s memory and sent with each backtest request. It is never saved and never put in a link.

Survivorship bias. The universe is today’s listed companies; companies delisted before today are missing, which biases results upward.

Hypothetical, backtested performance. Results reflect simulated trades at end-of-day prices with estimated costs, do not reflect actual trading, and are not a guarantee of future results. Fundamentals are used as they stood in the filings available on each date; the universe excludes companies delisted before today, which biases results upward. Not investment advice.

Every result on this page is hypothetical. Aprise presents data and its sources, not recommendations; this is not investment advice or an offer to buy or sell any security.