Skip to content

Release Notes

Last updated 2026-10-26

What changed in each version of QuantAscent, newest first.


v1.0.0-beta — Coming winter 2026

The public beta. The release date and what it includes will appear here when it ships.

  • Research is its own page in the sidebar's Strategy Lab section, between the Library and the Adaptive Builder. It was the Static Builder's Metrics tab; its steps and every number on them are unchanged. The Static Strategy Builder is now the editor alone, and its toolbar gains an Open button that lists the portfolio's saved static strategies, so a strategy can be opened for editing without going through the Library.
  • Research is organized as three steps: Universe, Signals and Combine.
  • On the Combine step, picking an adaptive strategy shows a line above the strategy panel saying that the analysis covers the rules its process derived last, with the date, and that the test of the process is its walk-forward record. Add as filter and Add as scoring criterion on the Signals step now offer static strategies and adopted built-ins only, since a rule written into an adaptive strategy would be overwritten at its next derivation. The Universe step adds a profile of the universe you define (names per quarter, base rate against the whole market and SPY, single-pick odds, dispersion, stable signals, data quality) and a sector-by-size grid you can click to pick a universe.
  • The Signals step's Quintiles card now shows each fifth held as an equal-weight portfolio, compounded per year (the base rate's measure), instead of the plain average of every name, which a few extreme winners could carry the wrong way. The table adds the typical name's return, the share up and the worst period; a sentence under the chart says whether the fifths line up in order and what to do about it, and a warning names a fifth whose plain average is carried by its ten biggest winners.
  • The Static Strategy Builder's metric insight panel, in the Add and Edit Criterion dialogs, shows the same held-portfolio fifths with the same reading sentence and warning, so the two pages agree on a metric.
  • When a weekly data pack installs while the app is open, Research reloads the matrix it had open with the pack's successor file (the pack names its matrices for their window, so the old file is removed), and the pages that read the matrix follow. Before, the page kept last week's matrix until a different one was picked by hand.
  • The Signals step shows a metric's detail in place of the old Metric Inspector window: its IC on every rebalance date, its quintiles, where it works by sector and size, and a rule preview with a lift interval that can be added to a strategy as a filter, a universe floor or a scoring criterion. The IC table gains a Stable column. The Threshold Scan, Conditional IC and Filter Tester tools are retired.
  • The Combine step adds a filter funnel that applies a saved strategy's filters one at a time and reports what each leaves (names per quarter, share of the matrix, equal-weight return, hit rate), and a redundancy grid of the metrics its composite score reads, each turned the way the score reads it, with a count of the pairs that rank the companies alike and of the pairs that rank in opposite orders; pointing at a cell names the pair, its correlation and the share of picks the two have in common. Score Analysis and Score Breakdown stay as they were.
  • Research opens on the Strategy Builder matrix, and each matrix's label counts its years from its own dates. The mouse wheel scrolls the page over the IC table; click the table first to scroll the table.
  • The matrix row sits directly under the Research page's title, and the strategy type row directly under the Static Strategy Builder's.
  • The rule preview on the Signals step states the rule in one sentence and shows its numbers as four tiles (companies that pass, whole universe, the difference with its 95% range, and how many rebalance dates the passing companies were ahead), with a line that says whether the range includes zero. Its controls read as a sentence (Share of the universe · keep the top · 17% of the universe, or Fixed value · greater than · 0.163).
  • The rule preview adds a chart of how the difference changes as the cutoff moves along the metric's own percentiles, highest values at the right, with the cutoff marked and the kept side shaded (keeping the top 10% is a line at the 90th percentile). The per-date chart keeps one scale as the slider moves, so a weaker cutoff shows smaller bars instead of a rescaled picture.
  • On the Universe step the two profile charts share the full width of their card, the grid card is named "Sector × size grid", a sentence under its Measure dropdown says what the chosen measure is, and the pointer becomes a hand over a grid cell. Clicking a cell shows a notice naming the universe it set, with Undo and Show editor buttons; the notice goes away on its own.
  • The Signals step adds an Independent signals card under the IC table: the strongest metrics that pass the table's two reliability checks, one row per signal with the metrics that rank the companies like it listed beside it, next to a grid of how alike the signals are. It answers how many separate signals a universe carries, and which metrics are the same signal under another name, before a strategy exists.
  • The Independent signals headline also says how many unrelated metrics would carry the same ranking information as the signals, so a set of signals that are alike without reaching the 0.7 rule reads as fewer signals than rows.
  • A strategy can set a minimum average daily dollar volume (Min Dollar Volume in the Static Strategy Builder, Dollar volume in the Adaptive Builder, Dollar Volume in Research's universe editor), so a universe excludes names too thinly traded to buy in size. The floor applies only to data that carries trading volume, which the data packs do from late October 2026; on an older matrix the control says so.
  • Fresh Filings is on at 135 days by default in Research and for new strategies in the Static Strategy Builder, so stale financial statements stay out unless you turn the screen off. Saved strategies keep their own setting.
  • In the Library, an adaptive strategy's Open in Backtester button now rebuilds the Backtester replay of its walk-forward record when the file is missing, for example after you deleted it on the Backtester page, and opens it once it is written. Deleting a replay on the Backtester page now says that it comes back priced on today's closes the next time you open the record or the Library.
  • In the Strategy Manager, View on a strategy now gives every stock in its list a button that opens the company in Company Analysis and, for scoring strategies, an Exclude checkbox that adds it to that strategy's Excluded Tickers immediately. An excluded stock stays in the list, crossed out, until the next rebalance rebuilds the list without it.
  • In Research's universe editor, each metric filter now shows the ✕ that removes it. The control was there but painted nothing, so a filter could be edited but never deleted; the ✕ that closes the onboarding tour had the same problem and is visible again.
  • In the same editor, the + Add Metric Filter link now stays in place above the filters instead of moving along after each one you add.
  • The minimum share price in every universe now uses the price a stock actually traded at on each historical date, not a price adjusted for splits that happened later. Backtests no longer include stocks that only cleared the floor because of a later reverse split, so results for universes near the floor can change.
  • The data now covers every actively traded stock on the NYSE, NASDAQ and NYSE American, including stocks under $1. Before this, stocks under $1 and NYSE American listings were left out of the data entirely, and a stock that fell under $1 stopped updating and looked delisted in backtests, so a universe with no minimum share price will now hold more companies than it did. New universes start with a $1 minimum share price; set it to 0 to include those stocks.
  • A new metric, DollarVolume_3m, gives each company's average daily dollar value traded over the past three months. It appears in Research and the Company Screener with the first data pack built after this change.
  • Leverage now follows your broker. If IBKR reports a cash account (buying power equal to settled funds, including IRAs), rebalances size buys at 1x whatever Max Leverage says, and Settings shows which kind of account IBKR reports under a new Broker Account line.
  • Wording: the app and docs now say rules wherever they used to say "recipe", and the Adaptive Builder's third tab is called Records everywhere (no more "Your Records"). The rebalance approval dialog shows the date an adaptive strategy's rules were derived, next to its Eastern-time stamps.
  • In the Adaptive Builder and the Sweep, Equal Weighting is now a complete design on its own: choosing it clears the tilt templates, and sweep draws never blend it with them. A structure named Equal Weighting is always equal-weighted.
  • In the Adaptive Builder, Open full record on a sweep candidate now takes you straight to its record, and every record shows where it stands, Recorded, Locked, Adopted, Deployed, with the next step as its one button. Adopting a process continues into choosing the sleeve that will run it, or keeping it in your Library for now.
  • The Library now shows a strategy adopted or saved earlier in the session on your first visit, instead of only after leaving the page and coming back. The Static Builder had the same gap.
  • When a subscription ends, the app now removes the licensed market data snapshot (about 3 GB) from your computer and downloads it again when you resubscribe. Your portfolios, trades, strategies and settings are never touched, and Settings → Database shows when and why the data was removed.
  • The ETF trackers gain T-Bills (SGOV), a 0-3 month Treasury bill fund you can use as the cash leg of a fixed-allocation basket. It arrives with the next weekly data pack.
  • Scoring backtests are now valued at every trading day between rebalances, the same basis ETF trackers and baskets already used, so max drawdown, Calmar, the 30D/90D/6M windows and every chart compare like for like across strategy types. Expect deeper, more honest drawdowns when you re-run a saved scoring strategy; saved results keep their old numbers until then.
  • Fixed-allocation strategies can now be backtested. The Backtester lists your saved baskets and deployed fixed-allocation strategies alongside scoring strategies and ETF trackers, and refuses a basket holding a ticker the data pack doesn't cover rather than quietly dropping it.
  • The Fixed Allocation editor no longer has a Rebalance Schedule box. Cadence is set on the strategy in the Strategy Manager, the same as for scoring strategies.
  • The Percentile editor no longer shows the threshold-only Score Thresholds, Tiebreaker, Direction, and Min Stocks fields before a strategy is loaded.
  • Updates now check the downloaded installer's own version before installing it. A mismatch is refused and reported instead of installed.
  • The Interactive Brokers connection now runs on the actively maintained successor of the library IBKR flagged as no longer updated. Nothing changes in how you connect or trade.
  • In the Static Strategy Builder, a note under Lookback lists which of the strategy's metrics use it (the means, growth rates and current-versus-mean metrics) or says that none do, since a strategy built on current-period ratios and momentum scores the same at any lookback. Lookback now starts at 8 quarters in the builder and in Research's matrix build, because under eight quarters the trailing-twelve-month and year-over-year metrics lose their inputs.
  • The Backtester's launcher no longer has its own Lookback. A backtest scores with the lookback saved on the strategy, the same one live rebalances use, so a value changed in the builder can no longer be overridden by a forgotten setting on the launcher.
  • A backtest's default run name now states its date window and rebalance cadence, plus any cost setting you changed from its default, instead of repeating the strategy's lookback.