What’s new
Changes to Studio on the web and how to use them. Dates use Taipei time.
Nov 15, 2026
Comparing a few settings keeps your strategy
When you ask the AI in AI optimization to compare a few settings, such as holding counts or rebalance timing, it shows the comparison and keeps your strategy instead of registering the best one as a new backtest result. To adopt one, ask the AI to change the strategy to it.
Open feature →Resume unfinished older research on its chosen period
If a removed exam setting leaves unfinished older research starting after its cutoff, reopening keeps your chosen start and removes that obsolete restriction. Saved research figures stay as recorded.
Open feature →
Nov 14, 2026
Keep your program when a research direction is unavailable
If a conversation cannot find its research direction, the AI explains how to preserve its program and start a separate direction from the original strategy version you choose, without treating an unknown starting program as empty or claiming it changed.
Open feature →A one-off comparison ends with the requested result
When you ask for a one-off backtest comparison in a strategy conversation, the AI keeps those versions and reports the result even when the new factor loses; ask it to optimize when you want to explore further.
Open feature →Read conditions and measure baselines in large packed strategies
The research assistant can read conditions and parameters in large packed strategies in pages and measure the baseline using the complete original program. Reload the page and continue the existing research.
Open feature →The backtest period asks two things: how many years, and how much of the latest to keep from the AI
In the research goal, one tap picks the years to backtest and another keeps the latest 6 months, 1 year or 2 years from the AI; the bar beneath shows the dates the AI researches and those it never sees, and your main strategy’s report draws its line on the same day. New research backtests the last 10 years by default. For exact dates, open Custom dates.
Open feature →The research flow reads as three sections at a glance
The research flow now sets its summary, its goal and each direction on cards over a light grey ground, with more room between the sections. The goal’s Apply sits at the foot of its own card, each summary row says Report, and a line under Directions says what Research does.
Open feature →After you change the backtest period, the original strategy’s figures follow it
Once you change the backtest period in the research goal, the AI measures the original strategy again on the new period at its next comparison; from then on Original strategy in the research list, and the summary’s best figures and comparisons, use the new period, the same as the other versions. With the original as your main strategy, its report also draws the line at the end date, splitting AI researched from Unseen by AI.
Open feature →
Nov 13, 2026
A main strategy whose report isn’t in yet no longer shows the earlier one
If you leave or reload the page before a version you set as main has its report, Report and Holdings now say its report is not in yet instead of showing the earlier main strategy’s figures as if they were the new one’s; Run backtest on that card brings it in about a minute. With an end date on the research period, models the AI retrains each year also keep running to the latest day, so Unseen by AI shows that model’s own results.
Open feature →Research narration follows your interface language
Research openings, step explanations and directions now explicitly follow your interface language, reducing unexpected English in Traditional Chinese conversations.
Open feature →Reading a main strategy’s chart keeps its AI researched and Unseen by AI columns in place
Sliding across the chart at the top of the report, or resting on a year, keeps both columns in place and the card the same height: a day reads what each side returned up to it, a year what each returned that year, and a side with none of those days keeps its dates over a dash.
Open feature →Comparing or rerunning no longer adds a “No better” result
When you ask the AI in AI optimization to compare a few options, or to rerun the same rules, and the strategy stays as it was, it records the comparison and keeps your strategy instead of registering it again as a “No better” backtest result, and the chat no longer shows an unfinished Save validation findings step. Asked for rules an existing version already has, it points you to that version.
Open feature →Set as main no longer shows the old version’s report first
After you press Set as main in AI optimization, Report and Holdings say they are switching to the new main strategy’s report and show it once the new version is run up to the latest data, instead of showing the old version’s figures as if they were new. With an end date on the research period, or after a reload, it no longer waits for the notebook to run again and is usually ready in a few seconds.
Open feature →No more exam: to keep a stretch from the AI, choose an end date
The research goal no longer holds out recent years as an exam: the AI sees results for the backtest period you choose, and each result shows that period alone. To keep a stretch from the AI, choose Custom under the backtest period and set an end date; once it is your main strategy, the report’s right of the line is what the AI never saw.
Open feature →A main strategy’s annual return says which years it covers
Research scores only its own years, while a main strategy’s annual return runs to the latest day, so the two differ. Your strategy list now shows the year the annual return starts (Ann. 2020–), and the preview card, the report’s check-up and the professional view name the whole backtest’s period, apart from the AI research stretch at the top of the report.
Open feature →Uploaded reports that stop early point to your own program
A report you uploaded with the finlab package that stops on a past day now says, above its report and holdings, to delete the end date in your own program and run it again, instead of opening an empty code tab in Studio.
Open feature →Folders for your strategies
Your strategies in the sidebar now have New folder: drag a strategy onto a folder, or choose Move to folder from its menu; to sort many at once, press Select, tick them and press Move. Folders can be closed, renamed and dragged into order, and deleting one keeps the strategies in it.
Open feature →Saved research stays saved when live updates are interrupted
Once research is saved, an interruption to live updates no longer reports a failed save or asks the AI to close it again or repeat the backtest; reopen the strategy to read the saved result.
Open feature →Status lists only open issues, with fixes pending release in one row
Service status on the Status page now lists one by one only issues still being looked into or fixed; fixes waiting to go live fold into one row you can open for each one’s details. The count on the Support page leaves them out too, so it reads All normal when only those remain.
Open feature →A main strategy’s report splits what the AI researched from what it never saw
Set an AI-researched version as your main strategy and its report draws a line where research ended, with each side’s return and how it did against the market above it: the left is what the AI could see and pick from, the right the days it never saw, growing each day and closer to how the strategy does in use.
Open feature →Fixed-end warnings leave upcoming rebalances alone
Strategies waiting for an upcoming rebalance no longer show a fixed-end warning. When a program stops at the end of 2026 and its report records newer prices, the warning can appear on the first trading day of 2027, without waiting two weeks.
Open feature →A report stuck on a past day runs to the latest in one press
A strategy whose program sets a backtest end date keeps its report and holdings on that day, however often it runs. Report and Holdings now say which day at the top; Run to latest removes the end date, saves the program and runs it once, keeping a backup of the previous one. When Studio can’t change the end for you, it opens the code.
Open feature →Ask the AI for a chart and get an interactive one in the chat
Ask the AI in the chat for a chart, such as the cumulative return of this strategy against the market, and it draws in that run’s output in your light or dark theme. Hover or tap to read the values, and tap a legend entry to hide or show a line; it reads well on a phone too.
Open feature →
Nov 12, 2026
Running saves the program, and the report is always this run’s
Pressing Run or Backtest first saves your edited program, so the report is the one this program made; upload=False no longer keeps the report and holdings from updating. If the run stops on an error or its report isn’t saved, Analysis and Holdings both say so and show which run the report below is from; on an error you can ask AI to fix it, and if it runs out of memory, they point you to FinLab Desktop.
Open feature →Research goals decide whether the AI may use market filters and stops
Adjust on the research goal now has Allow the AI to use, with Market filter and Stop-loss and take-profit switches. Both start off: the AI neither adds nor proposes them and keeps any the strategy already has; turn one on and apply to let the AI try it. Goal words are no longer turned into rules: Apply saves them as typed, and the rows below decide what counts as success.
Open feature →Rolling report metrics use the Python formulas
Rolling Sharpe, volatility and tail ratios in Analysis now use the same formulas as Python. Reopen the report to see the update; strategy return curves are unchanged.
Open feature →Start a new model after completing an analysis
After an analysis, tell the AI your new model direction and backtest period in the same conversation to start a separate direction with those dates; the earlier analysis and main strategy are kept.
Open feature →Reasoning effort moves to the top of the AI menu
With your own Claude Code or Codex on the desktop, press the AI name beside the input to pick its reasoning effort near the top of the menu: medium, high or xhigh, also shown beside the AI name. Research depth now sits under Advanced and sets only how long and how thoroughly a research report goes.
Open feature →
Nov 11, 2026
Analysis says why a run left the report unchanged
When a strategy already had a report and its program carried upload=False or ran no backtest, running it looked like nothing happened. The Code tab now warns beforehand that running the program will not update the report or holdings, and a note above the report says this run saved none and the one below is the earlier report.
Open feature →Strategies the AI writes update their report when run
The AI no longer adds upload=False to a strategy’s sim(), so running it after setting it as main updates the report and holdings; for an existing strategy, delete upload=False to do the same.
Open feature →Factor checks stop at the research period’s end
When the research period has an end date, the card the AI uses to check a stock score or condition also measures only to that day.
Open feature →AI research judges each direction by the research goal
AI research judges whether a direction holds by the research goal’s success conditions and Check again on earlier years, the same rules that mark a version better.
Open feature →Open the fixed version from beside the error
Once the research AI has built the fixed version, the note beside the error says so, and See the fixed version opens it in the research flow instead of building another.
Open feature →
Nov 10, 2026
Continue a large older research flow in a new strategy
Older research flows over the 12 MB storage limit now show their size in AI optimization; Continue in a new strategy opens a new strategy from the chosen version and leaves the original research as it is.
Open feature →When a strategy fails, AI can look into it
When a strategy fails, Ask AI to fix it now sits beside the error. Once AI finds a fix, the research AI builds it as a version beside the original, so you can compare the returns before making it your main strategy.
Open feature →Chat allowance hint moves beside the AI menu
When your allowance runs low, the AI menu shows the percentage left and Your AI on desktop beside it, keeping research content clear.
Open feature →Understand which filters drive your strategy
What drives this strategy now names filters from the full program in everyday words, lets you explain saved results without rerunning backtests, and wraps long overview actions.
Open feature →See what Ask AI is checking on your phone
In AI optimization, Start check or Start test scrolls smoothly to the check card and shows what is being checked, automatically retries temporary connection failures before execution starts and keeps your reading position as results and the AI explanation arrive.
Open feature →
Nov 9, 2026
A used-up download allowance now says when it renews
When a strategy stops because today’s data download allowance is used up, the screen now says it renews at 8:00 every morning, Taipei time, instead of showing a program error.
Open feature →Research save errors now say why
When a research result can’t be saved to the cloud, the notice now says why and what to do next. If the research flow is full, delete historical directions you no longer need and this round is saved with that change. If another window updated it, choose the version to keep.
Open feature →Practice results stay when you change your mix
Changing your practice mix no longer wipes your earlier results. The account page now shows both Since this mix and Since day one against 0050, along with how each earlier mix did. If you do not name a new amount, the new mix starts from what the account is worth now. Rise and fall text across the site is also darker in the light theme, so it is easier to read.
Open feature →Schedules move up when memory runs out
When a cloud schedule runs out of memory, it moves to a larger size if your plan has one. At your plan’s limit you’ll see which plan to upgrade to, and after 3 failures in a row the schedule pauses and we email you.
Open feature →Cloud schedules get twice the memory
S and M cloud schedules now run on 8 GiB and 2 CPUs, and your cloud allowance is charged at the new size. Strategies that run in AI research now run on a schedule too.
Open feature →Open a practice account in one press
If you have no practice account yet, Portfolio now offers one of NT$1M on your best strategy, or on one of FinLab's if you have none that fits. Press Start practising and it is open: no form and no real money, and it follows the strategy at real prices every day. While Taiwan is trading, press it after the closing prices are out.
Open feature →Ask AI questions and actions line up
In the AI optimization overview, each Ask AI question now has its own row with consistent places for checks, results and version details; on phones, actions sit below the explanation.
Open feature →Clear next steps when memory runs out
When a strategy needs more memory than the web version or a cloud schedule allows, you now get three simple steps to finish it in FinLab Desktop instead of a program error. No code changes needed.
Open feature →How free and trial chats are retained
Free and trial chat composers show the 30-day retention notice, with a privacy policy link explaining its purpose and scope.
Open feature →
Nov 8, 2026
Research conclusions in plain words
New research writes each version's conclusion in everyday words: what changed, whether the result got better and what is still uncertain. Research terms and measured detail stay under More details. Conclusions already saved are not rewritten.
Open feature →Version figures say what they mean, and History says View result
Open a version in History and the annual return, Sharpe and drawdown each have a few plain words under them, with no need to hover, and Sharpe says above 1.3 is good. Versions with a result say View result in the list. While the AI is busy, a spinner under the Ask AI heading says you can ask once it finishes and whether you can close the page.
Open feature →Notices show over the large window, and a failed Set as main can be tried again
While the large window in AI optimization is open, failure notices and the warning about deleted backups now show over it, on a phone too, and stay after it closes. If saving fails on Set as main, the button stays so you can press it again. A next step in an Ask AI answer that goes to another tab or starts the AI closes the window, so you see the result right away.
Open feature →Holdings details no longer leave a dot at the end of a line
On a phone, the dot between the entry, signal date and current price no longer hangs at the end of a line when the text wraps, and the rows now line up with the Next rebalance card above.
Open feature →Holdings list a stock a stop is waiting to sell
A stock that hit its stop-loss or take-profit and that the next rebalance keeps now stays under Sell at the close in Next rebalance until it is sold, instead of disappearing once the signal day has closed; when a stop is among the sales, the sell list says which each one is, stop-loss, take-profit or rebalance, and the weight to buy back to afterward.
Open feature →When a notebook run leaves no report, the page says why
When Run the notebook on a strategy's Report tab leaves no report, the page now gives the actual reason: an error in the program (with the name you may have meant if one was misspelt), a setting that turns saving off, no backtest in the program, or a report that did not arrive, with buttons to open the code or run it again. A missing backtest line can be copied with one press.
Open feature →Strategy chats count an experiment only when research starts
Asking the AI on a strategy page to place orders, explain code or help with controls no longer adds a failed research direction or increases the number of strategies tried; requests to change, backtest or analyze a strategy still start research and save its results in the same conversation.
Open feature →Plans and checkout say prices are in TWD and cards from abroad are accepted
The plan window and checkout now say under the price that it is in New Taiwan dollars (NT$, TWD), tax included, and that Visa, Mastercard and JCB cards issued outside Taiwan are accepted, with your bank converting the charge; the card field lists the three cards. The invoice field is now Receipt, and its default emails you the e-invoice, which readers outside Taiwan can simply keep.
Open feature →Versions and Ask AI answers open in a large window
Clicking a version in History, or View result on an Ask AI question in the overview, opens a large window: the annual return, Sharpe, drawdown and conclusion on the left, the settings grid or the four Ask AI questions on the right, all in view without scrolling on a 1440×810 screen. An answer opened from a version replaces it in the same window, with Back at the top left. History keeps a compact list. On a phone the window fills the screen.
Open feature →The AI’s steps are named in plain words
The FinLab guides the AI reads in a row while building a strategy are one step, Read the FinLab guide, with topics such as backtesting and data tables under it instead of file names; a command the desktop AI runs on your computer shows as Run a command, with the command itself under its code.
Open feature →“Will it hold up as well as the backtest?” also reads the earlier years
A version research found better than the original only over the research period, and no better over the earlier years, is no longer called ready to schedule when you ask “Will it hold up as well as the backtest?”: the card leads with a cell such as Better than the original over 2015–2020 too, with this version’s and the original’s figures over those years, and concludes it may only fit the research period’s market; a version that was better over the earlier years too shows Better too there.
Open feature →“How does this strategy make money?” names undescribed filters in plain words
When code written by AI research or optimization has no description for a filter, the AI reading the answer names each filter and stock score from what the code computes, such as “revenue at a new high” or “revenue growth”, so the card, its buttons and the Ask AI entry no longer show code names like consecutive_records or yoy; Details says which names the AI gave and shows the code behind them.
Open feature →
Nov 7, 2026
“Why has it been losing lately?” finds the cause with a fixed method
When a strategy falls deeper than expected from its high, the strategy page’s Ask AI why, the alert email’s link and Ask AI in AI optimization open the same card: it runs the main strategy again to the latest day and compares the fall with its own range, the market over the same days, momentum, value and small caps, and its five worst holdings, to tell normal swings, the market falling with it, strategies like it out of favor, or a few stocks dragging it down, with a chart since the high and the next steps. The method and the verdict are the same every time.
Open feature →When an Ask AI check does not finish, the conversation stays and you can check again
If an Ask AI check in a strategy’s AI optimization drops its connection or hits an error, the question and the AI’s first words stay in the conversation, the check’s step says why in plain words, and Check again beside it checks the same version once more.
Open feature →Scrolling back through a chat on a phone, nothing covers the text
Scrolling back through a strategy’s AI optimization chat on a phone, the strategy name and the Overview / Chat / Research flow switch now sit on a solid backdrop, so text passing beneath no longer runs into them; Jump to latest and Next now rest just above the message box, with the chat fading out behind them, so they never cover a card’s figures or buttons.
Open feature →Ask AI answers the moment you press
Press any Ask AI question in a strategy’s AI optimization and the conversation shows the question, the AI’s first sentence and the check’s progress (with seconds) at once. When the check is in, the AI lists the directions it added to the research flow (each starts only when you press Research), then tells you in two or three plain sentences what the result means and what to do next.
Open feature →“How does this strategy make money?” names its filters in plain words
The card, its buttons and the directions it adds now call each filter and stock score by the strategy’s own description or the data it reads, such as “revenue at a new high”, instead of a code name like keep_high; the code itself is under Details.
Open feature →Settings that fall short stand out in the settings grid
In “Will it hold up as well as the backtest?” and a re-tune’s settings grid, a setting that falls short is now a framed grey cell that stands out from the nearby settings’ backdrop in light and dark themes; each legend swatch looks like its cell, and untested cells are keyed too.
Open feature →A strategy you just copied no longer opens a blank report
A strategy copied from the library to My Strategies that has not run yet now says it has no report yet in the Report tab; press Run the notebook below to create one.
Open feature →Start researching right after signing in
A new account back from Google sign-in no longer sees “Choose a plan” first: after a second or two of “Getting your account ready…”, it can start. A question brought in from an article now has a single Start research button, signed in or not.
Open feature →Comparison tables show names and figures together on a phone
In the AI’s run output, a long variant name in a comparison table now wraps so annual return and Sharpe ratio sit beside it; swipe sideways for the other columns and the names stay in place.
Open feature →One AI menu beside the input
Press the AI name beside the input to choose the AI, the research depth on Home and whether to continue the conversation; models and other details are under Advanced. When your FinLab AI allowance is almost used up, a line above the input suggests using your own Claude Code or Codex in FinLab Desktop.
Open feature →Report and Code open with what the strategy does, in one sentence
Open one of your strategies and its Report and Code tabs now begin with its rule, such as “Each month-end, from stocks with a P/E of 8 to 20, buy the 60 cheapest.” It updates by itself after you change and save the program.
Open feature →Ask AI: can I trust this backtest?
In a strategy’s AI optimization, press “Can I trust this backtest?” to check four things: whether its data was out on the day it was used (no peeking at the future), whether fees and taxes are counted, whether its stocks could be bought, and whether stocks that later delisted are included. You get “Looks trustworthy”, “Things to watch” or “It peeks at the future”, and when something is off the AI adds a “Fix it and run again” direction under that version, one press to start.
Open feature →Ask AI: what makes this strategy money?
Open a strategy’s AI optimization and press “What makes this strategy money?” under Ask AI: the AI backtests it once without each of its filters and checks whether its stock score tells good stocks from bad, then tells you which filter it relies on and which make no difference, with next steps you can start researching right away. The answer stays with that version and can be run again; asking the AI whether a stock-picking condition works shows the same card.
Open feature →The AI writes to you in everyday words
The directions the AI proposes after a research now read as questions anyone would ask, without jargon; its comparison tables name variants and columns in everyday words; and the names it gives strategies are short phrases, without underscores or settings.
Open feature →Plainer buttons under a research, technical warnings tucked away
Under a research, the ways to try next now read Find the best settings, Drop extra conditions and Pick stocks another way, with plain descriptions. Technical warnings from a run no longer sit between its results: they wait under Technical notes at the top right of the run output, to open when you want them.
Open feature →Pick the model and rounds right on auto-optimization
Auto-optimization at the top of a strategy’s AI overview now lets you pick the model (the same setting as the chat box, so changing one changes the other) and the rounds (5, 10, 30, 100 or 300) before pressing the blue Start. While it runs you see the round it is on, a cell per round, with the rounds that improved lit up.
Open feature →Re-tuned settings now show as a heatmap of how steady results are
When AI optimization re-tunes a strategy’s settings, that direction in the research flow now charts every setting it tested: settings that beat the original are blue, bluer the better, and the chosen one is outlined. If the settings around it are blue too, it sits on a steady plateau rather than a lucky peak; switch between annual return, Sharpe ratio and drawdown to compare. Ask AI’s “Will it hold up as well as the backtest?” uses the same chart for whether it holds up when settings change a little, and a version not yet re-tuned offers Try re-tuning its settings to add one.
Open feature →Strategy questions finish as completed analyses
Ask about factors, possible future losses, Alpha/Beta or why a strategy weakened on its page: the AI answers using its original rules and backtest period and marks the analysis complete with the strategy unchanged; requests to change a strategy still compare whether it improved.
Open feature →
Nov 6, 2026
Daily updates are offered only when a strategy has code
Strategies uploaded from Python or with no code yet no longer show a daily update button that could not work. For a Taiwan stock strategy with code, press Turn on daily updates above its holdings and the cloud keeps them up to date.
Open feature →Ask AI: will this strategy hold up as well as its backtest?
Open a strategy’s AI optimization and press the card under Ask AI: the AI tests how deep a bad year could fall, whether the result looks like luck, and how much is left once the versions tried during research are taken into account, then tells you whether it is ready to schedule, worth starting small, or most likely luck, with next steps. The answer stays with that version on any device and can be run again; you can also ask from a backtest result or an auto-optimization recap.
Open feature →Usage from the account menu, and the report says how current it is
The account menu now has Usage, with AI and daily data usage on one page, and the notice when an allowance runs out links to it. A strategy’s report now says through which day it runs, such as 2026 – Nov 5. Opened inside Facebook or LINE, where Google does not allow signing in, Sign in now explains how to open the page in your phone’s browser.
Open feature →The strategy menu stays on screen on phones
On a phone, the ⋯ beside a strategy now opens its menu fully on screen, to backtest, schedule, rename or delete it. On a research result, +N now lists the rest of its holdings, and pressing the research goal opens Adjust.
Open feature →AI optimization opens on its overview, auto-optimization first
A strategy’s AI optimization now opens on its overview: auto-optimization at the top, one press away, then the best version so far, directions to pick yourself and your last conversation on this device. The Overview | Chat switch at the top is on phones too, and AI keeps working whichever you look at.
Open feature →Segmented choices show clearly which one is picked
Segmented choices such as 3M | 1Y | 3Y | All in Market Pulse, strategies, schedules and research goals now outline the picked option clearly in light and dark themes; with a keyboard, Tab stops once, on the picked option, and the arrow keys switch between options.
Open feature →Research goal settings are shorter
Adjust on the research goal now lists the metrics to improve and those that must not worsen on one line each: add one with Add metric, tap one to take it off. The backtest period and rebalancing follow, while extending, checking on earlier years, card metrics and templates sit under More settings, so the settings are less than half as long.
Open feature →A strategy’s Analysis tab is now Report
The Analysis tab at the top of a strategy is now called Report: it holds the strategy’s backtest report, unchanged. In AI optimization’s research flow, Backtest report on the ★ main strategy now opens this tab straight away instead of running the program again.
Open feature →Kept holdings count their return from the first buy
For a stock a rebalance kept, the holdings view’s Total now counts its return from when it was first bought, on adjusted prices with dividends, and lists the return since the rebalance beside it as This period, which stops and targets still use.
Open feature →Download the presentations and spreadsheets the research AI makes
Ask the research AI on the web for a PowerPoint or Excel file and it appears under the run’s result; select Download to save it. Each file is up to 5 MB, downloadable for 7 days, by your account only.
Open feature →AI optimization’s best version always beats the original
The large Best version figure in a strategy’s AI optimization overview is now always a version that beats the original on your goal’s metric, such as annual return, and the count of experiments that beat the original uses the same test; until one does, the overview says so instead of showing a version that trails the original.
Open feature →Cloud schedule costs show as a share of your budget
On the Schedule page, each cloud run now shows its share of this month’s budget instead of a US$ amount, and the plan comparison shows each plan’s cloud budget as a multiple of VIP’s.
Open feature →The home page on a phone opens on the web app
On a phone, the Studio home page opens on Open the web app, without first showing Download the app and jumping; the demo’s five tabs show in full and scroll sideways. On a computer it opens on the download for your system.
Open feature →The research overview stays within reach in AI optimization
Once a strategy’s AI optimization has a conversation, a Chat | Overview switch sits at the top: Overview brings back the best version, every experiment’s outcome and the directions to try next, and Chat returns to the conversation; research in progress keeps running.
Open feature →Analysis shows how bad the next year could get
A strategy’s analysis adds How bad the next year could get: 2,000 trials on its past daily returns show how far it typically falls from a high, how deep about 1 year in 10 falls, how long it could stay below a high, and where its Sharpe ratio could be and whether the result may be luck; in Professional, open Risk, then Next year. These are trials on past data; the future may differ.
Open feature →Comparing strategies shows whether they move alike
Add two or three strategies to a comparison with + in the library, and the comparison now ends by saying how alike each two move day to day, such as Move almost as one or Move independently; the more alike two are, the less holding both spreads your risk.
Open feature →
Nov 5, 2026
Holdings tell a kept stock from a new buy
In a strategy that rebalances on a schedule, a stock kept through a rebalance now shows in Holdings the day it was first held and the price its return restarts from at that rebalance, instead of reading as bought that day; the holdings map’s card says the same.
Open feature →The rebalancing you pick stays through AI research
Pick weekly, monthly or quarterly rebalancing on Home, then press Start or type your own idea: the new strategy’s first research, later research and auto-optimization all keep it. The research goal shows Fixed weekly rebalancing; to let the AI choose, open Adjust on the research goal, set Rebalancing to Auto and apply.
Open feature →Paid plans’ web research no longer pauses at 600 runs a day
Programs that research and auto-optimization run on the web for paid plans no longer pause after 600 a day, and paid users on the same network no longer use up each other’s runs.
Open feature →Schedules show each strategy’s next rebalance
Each strategy on Schedules now shows its next rebalance date, the same one its holdings show; click its name to open the strategy.
Open feature →When a direction doesn’t work out, the chat says why and suggests the next
When an AI optimization direction is set aside, doesn’t beat the original or doesn’t finish, the chat now says what was tried and how it ended, then suggests one direction to Continue research. When a strategy gets its first backtest result, Let AI auto-optimize leads the next steps.
Open feature →Taiwan holdings show their progress and whether the data is current
Open a Taiwan strategy’s holdings and Next rebalance shows how many sessions the holdings have run, when this period ends and whether the data behind the result is current; it says so when newer data has arrived, and Data details lists when each kind of data updates.
Open feature →The research flow opens on a summary you can act on
The Research flow in the AI tab now opens on a summary: how many tries were made and improved, and the best version on each goal figure, such as annual return, Sharpe ratio and max drawdown. Click one to see its backtest report or set it as main beside it. When auto-optimization finishes or stops, the conversation keeps a recap card too.
Open feature →Schedules now say what you get
Above your holdings, on Schedules and in New schedule, the same words now say what a schedule does for you: holdings update on time without a rerun, a Taiwan strategy’s rebalance brings an 8 a.m. email listing what to buy and sell, and nothing places orders for you.
Open feature →AI optimization is easier to find, and Auto experiment is now Auto-optimization
A strategy’s views now show an icon beside Holdings, Analysis, AI and Code, and AI stands out in blue even while another view is open. Auto experiment in the research flow is now called Auto-optimization, the name the plans use; it works the same way.
Open feature →
Nov 4, 2026
FinLab Desktop starts on FinLab AI and keeps your model
FinLab Desktop now starts on FinLab AI’s Claude Haiku 5.5, so you can research before connecting Claude Code or Codex; every plan can use it, sharing the allowance with the web. Pick another model in the input’s model menu and the app keeps it next time.
Open feature →The strategy library in the navigation
The left navigation and the phone menu now include Library, which opens FinLab’s strategies as cards you can sort by performance, filter by market and tag, and compare.
Open feature →A verified strategy links its own research
Open a verified FinLab strategy and Research above its holdings and analysis opens its article in a new tab; strategy library cards are marked Verified with the same link.
Open feature →Delete several research directions at once
In the AI tab’s Research flow, choose Select, tick the directions to remove and press the red Delete to remove them after one confirmation; ticking a direction ticks those grown from it too. The original strategy and directions under research cannot be ticked.
Open feature →Support’s answer to a problem the research AI reported comes back to the research
When the AI reports a problem to support while researching, reopen that strategy once support has looked into it: the conversation shows what they found and the fix, and Continue with this fix lets the research AI carry on with it.
Open feature →Easier-to-read Taiwan research pages
On the Taiwan research pages the chart period now sits under the main chart instead of over the cards, each page opens on the question it answers with its method one tap away, and figures read as percentages, percentage points and z-scores in both themes.
Open feature →App sign-in says why it didn’t finish
When you sign in to the FinLab app through the browser and cancel, or the link has expired, the page now says why in your language and points you back to the app to start again, with a link home.
Open feature →A new home page that shows Studio at work
The home page is now dark and opens on FinLab Desktop at work: writing a strategy from one sentence, finding a better version, running on your own Claude or ChatGPT, reading the backtest and holdings. The data section draws every S&P 500 company at the latest close as a point of light, with real recordings below. The download button picks the right version for your computer.
Open feature →Upgrade on your current card and pay only the difference
When you upgrade a monthly card subscription to a higher plan, you no longer enter your card again: the payment page shows what is due today (the monthly price difference for the days left this period), and Upgrade and pay charges your current card. The new plan applies at once and your next payment date stays the same.
Open feature →
Nov 3, 2026
Shorter stock research links, with a preview when shared
Researching a stock in Studio from its finlab.finance page now uses a short link such as studio.finlab.finance/en/stocks/us/AAPL, and pasting it into LINE, Facebook or Discord shows a preview of that stock’s research.
Open feature →Support shows how many questions you have left
Under the support composer you can always see how many of this month’s questions you have used, how many are left and when they reset. The website and the CLI share 200 questions a month, reset on the 1st in Taipei time.
Open feature →An interactive level on every docs page
Every docs page now has a level to play: drag the days, build conditions, replay a backtest, and see how finlab works as you go. Clear its missions to earn a star in the contents and on the docs home’s level map.
Open feature →FinLab docs move to Studio, rewritten
The finlab package docs now live in Studio: open them from Docs in Help, How to use data.get in the data catalog, or API docs beside a strategy’s code. Pages are rewritten with diagrams, / or ⌘K searches every page, and code copies in one click.
Open feature →
Nov 2, 2026
A plainer holdings map that opens a stock in one tap
Each tile on the holdings map now shows its move in one colour, without logos or the path along its foot, and a holding closed by a stop says so in words. On a phone one tap opens the stock panel, as a click does on a computer.
Open feature →The strategy list says how far a backtest reaches
The strategy list now shows how far each backtest reaches, in place of Last run, so an old report reads as old. The holdings map names its 5- to 240-day figures Holdings-weighted: the recent move of the stocks held, not the strategy’s backtest return, with the time of the prices.
Open feature →Clearer labels on research results
Holdings maps now group funds under ETF, US strategies name their benchmark as the S&P 500 without dividends, and comparison tables title seconds in your language.
Open feature →Article reruns use the published backtest period
Press Rerun on a research article with a published backtest period to use that period in Studio; you can still change it in the research settings.
Open feature →Looking up data no longer uses a visitor’s free research
Signed out, your free research is used only when a backtest or report comes back. After that, Start backtest takes you to Google sign-in, and a new account sends your question on its return. The notice now says only a new account keeps the strategy you just made.
Open feature →
Nov 1, 2026
Subscribe from Schedules after your trial
When a trial ends and cloud schedules stop or only one stays on, the top of Schedules offers Subscribe to resume, which opens the plans.
Open feature →Plans links keep your language and referral details
Plans links from Studio now keep referral details in the address, and links with lang=en open the English plans page directly.
Open feature →Some accounts keep one strategy updating after their trial
Some trials ending from November 2 keep their most recently kept Taiwan strategy updating on weekdays, with rebalance emails, on the free plan. Strategy and Schedules pages show this benefit, with options to pause updates or turn off rebalance emails; if it is withdrawn after the experiment, that strategy will stop updating.
Open feature →The data catalog opens on what’s new
The top of the data catalog lists recently added tables and fields; tap one to see what it holds and the line that loads it.
Open feature →The plans page now answers common questions up front
The top of the plans page now answers whether you can buy one month, whether your price changes, how the AI allowance works and what U.S. data is included. Each plan also says who it suits.
Open feature →Results and holdings now show when a strategy updates
After you keep a strategy, its result card and the top of its holdings show when the cloud updates it and when a rebalance email would go out. A Taiwan strategy with no schedule can start with Update automatically. The result card now names the same rebalance day as the holdings, including at month end.
Open feature →Holdings now use the same map as Industries
A strategy’s holdings map now groups stocks by industry with each industry’s share, and larger tiles carry the company’s logo and its path since entry. Switch between the total return, periods from one day to the year, or figures such as P/E and institutional net buys, each beside the whole market. Point at or tap a holding for its weight, impact, cost, days held, entry point and latest filings; select an industry’s name to zoom in.
Open feature →
Oct 31, 2026
A reminder to save anonymous strategies
If you have strategies without signing in, the sidebar and workspace remind you they are only accessible in this browser; use a Google account new to FinLab and choose Register with Google to save to keep them and try VIP for 7 days. Signing in to an existing account does not transfer the strategies.
Open feature →Holdings show the day they were computed, and say when a rebalance has gone by
The holdings view now says the day its holdings were computed. When a strategy has not been rerun since its rebalance date, Next rebalance shows that date and the day the holdings are from, instead of No trades needed, with a way to update. After a trial ends, the sidebar and Schedules list the cloud strategies that stopped updating daily, which resume when you subscribe.
Open feature →AI research checks earlier years before calling a version better
When AI research finds a version better than the one it grew from, the same backtest also looks at the same length of earlier years. If it was no better than the original strategy there, it is marked No better, with the years in its first line, and auto research only extends versions that held. To judge as before, turn off “Check again on earlier years” under Research goal → Adjust.
Open feature →Clearer Taiwan market signal definitions
Expand Trend breadth, Weekly MACD or Margin maintenance ratio in Taiwan Market pulse to see the moving-average periods, the 0050 benchmark and the market-wide estimate; indicator values and scoring stay the same.
Open feature →
Oct 30, 2026
Verified FinLab strategies are marked, with their research a click away
FinLab examples backed by published research now lead the sidebar and search, marked Verified. Below a strategy's analysis, other verified strategies in its market are listed; open one directly, or press Research to read its article in a new tab.
Open feature →All four plans side by side, with FinLab AI as a multiple of free
Compare plans now shows VIP, AI and both AI MAX versions (your own AI, more FinLab AI) side by side, with nothing to switch. Plan cards, the comparison, checkout and AI usage in Settings state FinLab AI as a multiple of the free plan; on a paid plan, AI usage shows your plan's multiple and how much you've used this month. When your trial ends, the own-AI prompt says your 7-day trial has ended.
Open feature →The payment page lists every way to start, and strategies kept before paying update daily
After you pay, the page lists Studio, Python (two snippets to copy; login() signs you in through the browser), the Claude Code / Codex setup prompt and the FinLab Desktop download; strategies you kept on the free plan are scheduled in the cloud to rerun daily and email you on rebalance days. Settings → Data download has the same Python and AI entries, and the API key in General copies with one click.
Open feature →Pay again right away when a card is declined
When your bank declines a card, the payment result page now says you can pay again, and to use another card or ask your bank to allow online payments if the same one fails; press Pay again to return to checkout for the same plan instead of finding it again from the plans.
Open feature →Research goes on past the cloud 60-second limit, and stops show the next step
When one cloud run passes 60 seconds, the output before the stop stays with the reason, and the AI carries on with smaller runs instead of ending the direction; for longer calculations, press Get FinLab Desktop. When an experiment stops for memory or a repeated error, press Rerun on its stop card, which also links to FinLab Desktop when it names it.
Open feature →
Oct 29, 2026
Signing in starts a 7-day VIP trial, and the prompt after research says so
After a first research without signing in, the prompt now says that signing in starts a 7-day VIP trial and keeps the strategy you made in My Strategies; press Sign in with Google. On phones it takes one line, so more of the result stays in view. The home page pricing and the payment page use the same wording.
Open feature →Strategy chat keeps only the Continue context switch
With a strategy open there is no Create strategy switch: every message starts a research direction and backtests it. Continue context keeps its name and is off by default; turn it on before sending to ask the previous AI more or have it save the strategy.
Open feature →Six more U.S. strategies in the English library
The English strategy library now includes six validated U.S. strategies; open the library to view their analysis and compare performance.
Open feature →Where to go next sits right under a research result
The three directions the AI lists after a research each have a short title and a line on the idea; press Research beside one to start it. Optimize parameters, Remove redundant conditions and Compare ranking methods are now small buttons under them. Directions follow the result card directly, without note-taking steps or a repeated performance summary.
Open feature →Research catches up when you return to the tab
When you switch to another tab or minimise the window during a research in a desktop browser, the AI keeps going. Back in Studio, the conversation shows the latest progress and backtest results at once instead of replaying from where you left. While you watch, text still appears step by step.
Open feature →
Oct 28, 2026
Shared strategy links show a performance preview
Links to strategies whose performance is shared with everyone now unfold on LINE, Facebook, Threads and Discord into a card with the name, author, annual return, Sharpe ratio, drawdown and past-year curve. Strategies shared only with invitees or signed-in members, and private ones, show FinLab Studio without their name or author. Holdings and code never appear in a preview.
Open feature →Run an AI research without signing in first
On the web you can press Start backtest or send a question before signing in: the research starts right away and shows its backtest, with no Google sign-in first. Sign in with Google to keep researching, keep the strategy or schedule it. A new account keeps the research; an account you already had does not bring it over, so run it again after signing in.
Open feature →More translucent phone navigation
Phone headers and bottom tabs are more translucent, revealing soft shapes from the content as you scroll; reduced transparency and increased contrast still use solid backgrounds.
Open feature →Content scrolls behind phone navigation
On phones, content can scroll behind the header and bottom tabs with softly faded edges and space to read the first and last items; conversation inputs and research queues stay above the tabs.
Open feature →Round controls and floating tabs on phones
Phone headers now have clearer titles and separate round controls, with space between the Back arrow and the strategy name; floating rounded tabs use a soft highlight to show your current view.
Open feature →Bottom tabs and a clearer Back button on phones
On phones, switch views within Strategies, Markets, Portfolio and Help using the tabs at the bottom; content and chat inputs stay above them, the strategy’s larger Back button is separate from its name, and desktop tabs remain at the top.
Open feature →
Oct 27, 2026
Recover interrupted strategy access in a new tab
If strategy access is interrupted, Recover in new tab opens your saved strategies. Unsaved code and research stay in the original tab; keep it open and back them up there.
Open feature →New strategies run in the cloud and email you on rebalance days
When a trial or paid account keeps a Taiwan research result as a strategy, it is scheduled in the cloud and reruns every weekday once the data is updated, so on a rebalance day you get a morning email listing what to buy and sell. If it cannot be scheduled, for example because your cloud slots are full, you are told why. Pause it on Schedules, or turn off Strategy Signal Notification in Settings.
Open feature →Jump to the next research directions when research ends
When a research round ends with new directions from the AI, the Jump to latest button below the conversation becomes Next: N research directions and scrolls straight to them, so you can pick one and keep going. Without new directions it stays Jump to latest.
Open feature →Research picks up again when you come back
If you switch apps, your phone reclaims the tab or you close it while research is under way, the research continues from its saved progress when you return to Studio or reopen the report, showing Research continues. Research you stopped yourself stays stopped.
Open feature →Free plan AI quota counts researches
The free plan includes 30 AI researches a month, and stopped or failed research is not counted. The usage page in Settings shows how many you have used this month, and the notice when they run out says how many you had.
Open feature →A cleaner preview when you rest on a strategy
Resting the pointer on a sidebar strategy shows a cleaner card: the annual return in large type beside the last 20 trading days, three key figures in a row, and holdings with company logos. Run, Schedule and Rename are text buttons, Delete is the red trash icon at the end, and long names show on up to two lines.
Open feature →Code keeps its size when you click to edit
Clicking code on a strategy's Code tab to edit it keeps the same font, size and line spacing, so the text no longer narrows or shifts, in light and dark mode alike.
Open feature →Email when your strategy trades
When your strategy buys or sells at that day’s open, you get an email at 8:00 Taipei time on the trading day listing the stocks to buy and sell, with a link to the strategy’s holdings. No trades, no email. It goes to every plan, and you can turn off Strategy Signal Notification under Settings › Notifications.
Open feature →
Oct 26, 2026
Research a direction without opening its card
In the research directions list, a direction not yet researched has a Research button on its card that adds it to the research queue without opening the card, and you stay on the research flow when its research starts. While it waits, Cancel in the same place takes it back out.
Open feature →The research flow is a tree of cards that stays put
Each direction in the research directions list is now a card, with connectors showing which version it grew from. A card you open stays where you tapped it, and research starting never folds the one you are reading. The main strategy's card has an ink edge on its left, and tries that came to nothing with nothing grown from them wait behind one row of tries that did not work out until you open it.
Open feature →Setting a main strategy updates Analysis and Holdings
Pressing Set as main in the research directions list now also switches the backtest report to that version, so Analysis and Holdings show the new main strategy.
Open feature →Complete research while keeping child directions
Measured research that keeps the original strategy can now save its validation conclusion while preserving existing child directions and research notes.
Open feature →See at a glance which direction the AI is researching
In the research directions list, the one the AI is researching now has a light blue background and a blue line along its left edge. Opening it no longer repeats "The AI is researching" and shows only the current step.
Open feature →Clearer pop-up notices
Pop-up notices at the bottom of the screen now have a background and border, and a thin line along the bottom, colored for success, warnings and errors, shrinks as time runs out.
Open feature →Old page addresses lead to their new places
Old Taiwan market, strategy embed, portfolio and video course addresses now open Markets, the strategy library and Home.
Open feature →Refreshed checkout and card update forms
Checkout and card update use the same rounded buttons, invoice fields take the card update form's style, and the company invoice fields span the full width.
Open feature →Refreshed switches and buttons in Settings
Settings switches, usage bars and buttons now match the rest of the workspace, and tapping anywhere on a switch row flips it.
Open feature →
Oct 25, 2026
Find the official Facebook page in Support
Open FinLab official Facebook page from Support to find the FinLab community.
Open feature →Smoother moves between pages
On phones, every page opened from the menu now slides in, and moving between pages keeps the navigation and sidebar loaded instead of rebuilding them.
Open feature →One sign-in for finlab.finance and Studio
Sign in on finlab.finance or Studio and the other opens signed in to the same account; signing out of either signs you out of both.
Open feature →The phone menu slides in and out
On phones, opening the menu slides the page aside to reveal it, and choosing a page slides it back; you can tap right away without waiting for the motion.
Open feature →Replacement directions recover from duplicate feedback
When a replacement repeats a parent introduction, AI feedback guides it to submit the new hypothesis on its own while preserving measured research evidence.
Open feature →
Oct 24, 2026
Dataset status stays inside Help
On the Chinese site, opening the dataset update status from Status keeps the navigation and Help's tabs, so you can go straight back to Status or anywhere else.
Open feature →Finished reports unwrap in the chat
A finished report appears as a cover at the end of the conversation: open it any time with Open report, and tap a follow-up question under it to research next.
Open feature →Stock page questions are researched as chosen
A question picked on a stock page, such as the P/E backtest or the peer comparison, is researched as asked and names the company.
Open feature →The screen stays on while research runs
While research runs on a phone or in a browser, the screen stays on so automatic locking no longer pauses it; switching apps or closing the tab still pauses it.
Open feature →
Oct 23, 2026
Research errors name missing packages
When a research program cannot import a module, its tool summary shows the missing module name.
Open feature →
Oct 22, 2026
Research directions focus on the new hypothesis
AI follow-up directions describe the new hypothesis and parameters while keeping their parent strategy link, reducing false duplicate refusals.
Open feature →Start AI research from a stock page
Open research from a stock page and sign in to start a report automatically; signed-in readers start immediately.
Open feature →Ask a question, then build a strategy
Enter a research question or strategy request on Home; the same AI researches and creates a backtested strategy when explicitly requested.
Open feature →
Oct 21, 2026
Check-up matches the Sortino calculation
Strategy check-up shows and applies the threshold for the Sortino calculation used by each report; saved reports keep their original verdict.
Open feature →AI supports Taiwan convertible bond backtests
Ask AI to build a Taiwan convertible bond strategy using daily or lower-frequency data and the existing backtest and analysis reports.
Open feature →Research report AI clarified in plan comparison
The research-report row in the plan comparison now says “With your own AI” for VIP and AI; MAX Local and AI MAX continue to show “Unlimited”.
Open feature →Python CSV example for broker details
Broker detail downloads in the data catalog now offer a Python example for saving a date range as CSV; broker indicators still download directly.
Open feature →
Oct 20, 2026
Resume an unfinished strategy from a blank notebook
Restarting an unfinished direction that began with a blank notebook keeps its candidate program, including a saved candidate after reload; when you add notebook code, research starts from that code.
Open feature →Plans and checkout in English
Arriving in English, plan comparison, the upgrade dialog, checkout, card updates and the post-payment page now read only in English; prices and plans are unchanged.
Open feature →Studio in English, U.S. first
Arriving in English, the home page, every page, sign-in and the desktop download now read only in English, with U.S. examples and markets first; the language control is a globe that lists the languages when you open it.
Open feature →Holdings map includes pending buys
The strategy holdings map shows pending buys at their next rebalance weights with a Buy label alongside pending sells; holding counts and returns still reflect existing positions.
Open feature →
Oct 19, 2026
Retry when your strategy list cannot load
My Strategies waits for cloud confirmation before showing an empty list; if reading fails, it keeps the strategies already shown and offers Reload strategies.
Open feature →
Oct 17, 2026
Your own AI now runs in FinLab Desktop
WebMCP is no longer offered on the web: to research strategies with your own Claude Code or Codex, download FinLab Desktop for macOS, Windows or Linux, with the same monthly own-AI tool calls; research reports are written by FinLab AI: quick research on every plan, standard and deep research with MAX or a trial.
Open feature →Chat writes research reports for you
On Home, a question that needs data analysis gets a research report the AI researches and writes right in the chat, telling you what it finds as it goes; the report opens beside the chat when done, and you ask about it or change it right there. Pick quick, standard or deep research in the composer; on a report page, Discussion keeps talking with its AI, saved with the report. Home from another page returns to your conversation and report; press it again at Home for a new conversation.
Open feature →Keep successful measurements after report diagnostics
After a successful research backtest, report diagnostics can reuse the measurement so findings can be saved while retaining the initial strategy.
Open feature →Save research findings when outer blank lines differ
When research retains the initial strategy, successful backtests can still be used to save findings if only outer blank lines differ.
Open feature →Save findings from parameter checks inside functions
The research assistant can save findings that retain the initial strategy after parameter backtests inside functions. Execution results now explain when no valid performance measurement was obtained.
Open feature →
Oct 16, 2026
Browse trade records by page
The professional backtest report shows 20 trades per page. Choose a page or use the arrows to browse; changing the sort order or period returns to the first page.
Open feature →Attach screenshots to support questions
Choose Add images or paste a screenshot in support, preview it, then send it with your question. Images stay in the conversation so support can help diagnose the problem.
Open feature →Finish research with an unmeasured starting strategy
When research keeps the initial strategy but lacks starting metrics, the assistant can reuse its successful unchanged backtest on that direction to record the measurement and save the findings.
Open feature →Complete robustness checks while keeping the strategy
After a parameter or robustness check, the assistant can save measured evidence and a conclusion to keep the initial strategy; automatic research then continues to the next direction.
Open feature →Plain-language data catalog descriptions
Open a table to see what it holds and when to use it. Revenue and statement tables also say which one to use for backtests or corrected figures, and what to check before backtesting.
Open feature →
Oct 15, 2026
Pages open faster from the side menu
On a computer, Reports, Schedule, Data Catalog, Markets and Help in the left menu are now ordinary links: a page starts loading as soon as you point at it and opens the moment you click, and ⌘/Ctrl-click or a middle click opens it in a new tab. Opening Help’s Status directly now loads the service log together with the page.
Open feature →Home opens a fresh conversation at once
Pressing Home from another page now opens a fresh conversation straight away, without flashing the strategy you last viewed; it also works from the strategy library, and Back returns to the page you came from.
Open feature →Schedules and the data catalog get their own addresses
Schedules now live at /schedule and the data catalog at /database, and old links still lead there; opening Schedules again shows your cloud schedules at once while they refresh in the background.
Open feature →Spot the main strategy in the research flow
In the research flow, the main strategy’s row now has a dark line at its left edge and a ★ Main strategy tag after its name, which a long name no longer hides; on phones the tag shrinks to a star.
Open feature →Holdings separate signal and entry dates
Each holding now pairs its execution date with the entry price and labels the signal date separately; the next rebalance reads Not set when there is no pending plan.
Open feature →A professional layout for strategy analysis
At the foot of a strategy’s analysis, choose Switch to Professional: annual return, max drawdown, Sharpe ratio and trade win rate up top, then tabs for profit, risk, ratios, win rate and liquidity, each listing its figures against their targets, with one chart at a time. This device remembers your choice; Switch to Guided goes back.
Open feature →Service status marks fixes pending release
In Support’s service status, issues already fixed that take effect with the next release or data update are now marked Pending release instead of Monitoring.
Open feature →
Oct 14, 2026
A clearer payment confirmation page
If your plan isn’t active yet after paying, the confirmation page lists the order, transaction and member numbers for support in full beside the status, and buttons and explanations no longer break mid-phrase; use the email button to send them.
Open feature →Free AI quota renews monthly
Free accounts receive a fresh US$0.50 FinLab AI allowance on the 1st of each month, Taipei time; sign in to continue researching, with no rollover of unused quota.
Open feature →Cloud research stops after repeated memory limits
If a reduced attempt still stops for memory limits in the same direction, AI records the unfinished measurement and guides you to continue with a complete program in FinLab Desktop.
Open feature →Code output keeps line breaks and all printed text
Run your strategy on the Code tab to see every printed line, with blank lines and explicitly unbroken output preserved.
Open feature →
Oct 13, 2026
Holdings open faster from the sidebar
Hover over a sidebar strategy to prepare its holdings so the map and list open sooner when selected, with Taiwan live quotes loading alongside them.
Open feature →Color-blind friendly rise and fall colors
Turn on Color-blind friendly colors in Settings → General to show rises and falls in orange and blue instead of red and green on the heatmap, candles and returns; your red-up or green-up choice still applies.
Open feature →Navigation now groups Markets and Help
Market Pulse and Industries now sit under Markets, and Status, What’s new and Support under Help, which opens on Status; switch with the tabs at the top of the page.
Open feature →FinLab 2.2.11: faster daily and quarterly conditions
Web backtests now align, combine, and filter daily and quarterly conditions faster, preserving results and unknown values before disclosure.
Open feature →Redesigned Analysis and Holdings windows
Open a strategy's Analysis to see the growth curve span the whole card with its total growth multiple and hover a year to mark it on the curve; the Holdings map now reaches the card's edges, and reports you have opened before appear at once.
Open feature →Strategy tabs: Holdings first, no longer overlapping the buttons
The tabs now run Holdings, Analysis, AI and Code, so a strategy opens on its holdings; in a narrow window they make room for the buttons on the right, and the AI and History labels are shorter.
Open feature →Strategy conditions work on mobile Safari
Tap a condition slot on the home screen to open the condition picker, then select Done to return to your strategy.
Open feature →