MORE STRATEGIES, MORE DIVERSITY [MAJOR]

[PITCH] Campaigns now produce more strategies, with more variety and higher average scores.

[WHY] The breeder now optimises stop loss, take profit and entry distance properly, not just the signals.

[DO] Nothing to switch on.

[PROOF] Tested side by side on the same markets and settings: average score up in every run, top five up to eight points higher, drawdown the same.

NEW FEATURES

NEW IN THIS RELEASE

The higher-timeframe control is now a single switch, on the launch page and on automation cards, with nothing to pick. Switched on, half of every 15 and 30 minute campaign starts on the chart alone and half checks the hourly first, and the breeding decides which shape survives on that market. H1, H2 and H4 strategies always train on their own. Strategies you already have keep exactly the shape they were bred with.
The 5 minute timeframe is no longer offered for training. Strategies you already have on M5 are untouched, and an automation saved with M5 keeps running on its other timeframes.
The symbol picker shows only the markets your broker offers, so every market you tick is one your campaign can run.
The TF column now shows a strategy's context timeframe as well, so one bred on the 15 minute chart against the hourly reads M15+H1 instead of just M15, in the vault, the portfolios and the classic dashboard. Filtering by M15 still shows both kinds.
The breeding engine now spends a fixed share of every generation on each strategy's entry distance, stop distance and take-profit distance, so those settings are bred to fit the market as hard as the signals are. In paired tests on gold, a major, a cross, an index and the hourly timeframes on their own, it held the strategy count and lifted the median score in every run, with drawdown unchanged.
The markets and timeframes you breed on now sit in the first panel of the Command tab, top left, so the one choice you make before a campaign is the first thing you see. It is the same panel as on the Settings tab, higher timeframe switch, select all and browse included, and a change on either side is the same change. Everything else on Settings is already tuned and can stay as it is. The Latest Campaign panel made way for it; the Campaign tab carries that view in full.
Algory now names its partnered brokers beside your own broker on the Command tab and on the deploy screen, with the partner's current offer and the date it was last checked, and the name opens their page in your browser.
The Appearance menu shows a locked SIMPLE MODE entry. A one-screen Algory that talks you through it is being built, and the entry says so until it is ready.
FULL DEPLOY has a new MAX EXPOSURE guardrail: set a multiple of your equity and a bot will not open a trade that would take the total size of every open position on the account past it, which is the number prop firms mean by leverage. Off by default.
The live trade record now says which of a bot's own rules closed a trade: profit target, daily drawdown limit, Friday close, session end or news close, instead of one generic "closed by EA" label. MetaTrader itself only reports that the bot closed it, so the bot now leaves a note the Algory Pilot picks up as it records the trade. Applies to bots deployed from this version on; the Pilot updates itself on your next deploy.
There is a PARTNERS tab now. It lists every broker and prop firm Algory works with, one card each with their current offer, the date the offer was last checked, the small print in full and the firm's own banner, and one button that opens their page in your browser.
The Command tab now has a NEXT STEP panel that says, in one sentence, the one thing worth doing right now, with a button that does it. It reads your setup, any running campaign, your strategies, your portfolios and your live bots, so it says press LAUNCH on a fresh install, build a portfolio once you have strategies, put that portfolio on a demo account, or replace a bot whose edge has faded. When there is nothing to do it says so. It took the place of the Strategy Vault panel, which showed the same count as the bottom bar and the same rows as the Vault tab.
Popups now use larger text. The title, message and button labels in every alert, confirm and upgrade prompt are a clear step bigger and easier to read at a glance.
The Command tab now has a LIVE BOTS panel, bottom left, listing every bot you have deployed: its market and timeframe, the rung it is on and the size it trades, what it has made or lost live against what it made in testing, and whether its edge is HOLDING, WEAKENING or gone. A bot whose edge has gone gets a REPLACE button that opens the portfolio it came from. It took the place of the Genome Synthesis panel, which showed the same genome list as the Map tab.
Settings now has a SEND BUG REPORT button. Write what happened, add a way to reach you if you want a reply, and press it. Algory sends the same report that COPY SUPPORT REPORT builds straight to us and shows you a reference number. No Discord needed.
[TOP: Algory now learns from every user's breeding through an anonymous summary; a PRIVACY switch in Settings turns it off in one click.] Algory can now learn from every user's breeding, not just its own. Settings has a new PRIVACY switch, on by default, that shares a small anonymous summary of each strategy you breed: its building blocks, its market and its scores. Never your account, your files, your trades or anything that names you. It is what lets Algory tell which building blocks actually work across thousands of strategies instead of one machine's, and you can turn it off in one click.
Automations now carry the higher-timeframe control on the campaign card itself, with the same on and off switch and the same choices as the launch page, so you can ask an automation for strategies that check a slower chart without hunting through a settings fold. The card also states the size of what you have set up before it runs: your markets against your timeframes, the higher-timeframe layer when it is on, and a warning when the shape is large enough to run for days without producing anything.
An automation can now skip dead markets. A campaign step can leave out any market that produced no strategies in its last few runs, so a night's breeding goes to the markets that actually produce something instead of grinding the same empty ones again. Nothing is dropped for good: each skipped market is tried again after a few runs, or after 30 days, and the activity log names every market it left out and why. If every market qualified, none are skipped.
Incubation now manages every MetaTrader you have deployed to, not just the one currently picked in Settings. Deploy to a second terminal once and it appears in the incubation tab with its own ladder; bots on every terminal graduate on their own account's live results, with no switching needed. New deploys still go to the terminal you have picked.
Every strategy now carries a GOOD or WEAK grade, in both interfaces. It is measured against real live results across every Algory user, not against a backtest: GOOD means the strategy scores 55 or more, the one mark that actually separated live winners from losers. It holds on every timeframe we can test, and it is the single biggest separator on the slower ones. Scores above 55 all performed alike live, so the grade is the honest signal and the number is for comparing strategies within a campaign.
Start Algory without any window: launch it with --nohud (or --headless) and only the background service starts - then use Algory from any browser at the local address. For people who drive Algory from a browser tab and never wanted the app window, including anyone whose machine shows that window as a black screen.
Incubation now watches for a bad start, not just a good one. A bot whose early live record is one a healthy bot would rarely produce, or whose live losing streak is longer than the worst run in its own backtest, is flagged COLD on the Portfolios incubation panel with the evidence spelled out, and cannot graduate while it stays cold. Nothing is changed automatically: stepping the bot down a level or rotating it out stays your click, with the numbers in front of you.
TRUE SIM runs your whole portfolio on one account before you risk a penny on it. Press TRUE SIM above the member table, set the risk each bot should take, and Algory replays every member together on a single simulated account with your Symbol Lock and aggregate-risk guardrails switched on - so the entries a guardrail would have refused never happen, and the curve is the account you would really have had rather than each bot's history added up on paper. The equity chart gains a fourth view for it, showing the simulated return, how far the account fell below its own peak with open trades counted as they moved, the most it ever had at risk at one moment and when that was, and how many entries the guardrails blocked. Choose fixed sizing to read the result in the same units as the rest of the panel, or compounding to see the account snowball. It takes under a minute for a twenty-bot portfolio, runs in the background while you carry on, and the panel says so plainly if the members or settings have changed since the last run. Pro.
Strategy incubation. Every bot you deploy now starts live at a quarter of the risk you set and climbs to the full amount by doing well on real trades: roughly two clean weeks or ten healthy trades lifts it to half, six weeks or thirty trades with a profitable record lifts it to full, and a bot that comes out on a genuine win streak skips the wait. A bot that is losing, flagged as decaying, or barely trading stays small, so a bad one is found while it is only risking a quarter of normal. Good bots step up on their own and the running bot picks up its new size within minutes, with no MetaTrader restart; prefer to approve each step yourself and one switch makes every step wait for your click. It is on by default with one tick to opt out per deploy, your open positions are never touched, and both interfaces have it. Bots deployed before this update keep trading exactly as they are and join the ladder at their next deploy.
[TOP: The new INCUBATION LADDER shows every deployed bot's current risk, its progress, and who is ready to step up.] The new INCUBATION LADDER on the Portfolios page shows your whole deployed book on one line: how many bots are incubating, proving or at full risk, and who is ready to move up. Click it open for each bot's rung, its current risk against its target, its live trades and its health, with a GRADUATE button when one has earned its step. A WHAT TO EXPECT button - here and on the deploy screen - plays a short animated picture of the whole journey before you commit to anything. And if the Algory Pilot ever comes off its chart, a banner at the top of the page says your bots are trading unwatched and puts them and the Pilot back with one press.
One press now health-checks a whole portfolio. RUN HEALTH CHECK on the Portfolios page re-checks every member on the market data since it finished training, one at a time in the background, and puts the answer on a single line: how many are holding, how many have decayed, and how many are still too early to judge. Verdicts based on less than 30 days of unseen data are counted separately as low confidence rather than being quietly added to the holding total, members that could not be checked are listed with the reason, and the line always says how many of the portfolio those counts actually cover. It reads from the MetaTrader you already have open and never closes or restarts it, members checked in the last 30 days are read from their last result instead of being run again, and if you close Algory part way through it picks up where it stopped instead of starting over. Pro.
The Portfolios page now says how many bots trade at once, and what that means for your risk setting. It reads every trade in the portfolio and reports the most bots that were ever in the market together, the number they sit at 99% of the time while anything is open, and the worst month they had, then works out the per-bot risk that keeps the whole account inside a 9% target at that load. Next to it is what the portfolio puts at risk if every bot is in at once, so the theoretical worst case and the worst that was actually measured sit side by side. The drawdown figures now also come with a floating version that counts open trades as they move instead of only when they close, so a month spent underwater on positions that had not closed yet is no longer drawn as a flat line, and the page says whether that made the drawdown deeper or shallower rather than assuming. Trades with a missing or out-of-order closing time are counted and named on the page instead of being quietly left out.
The shared-account portfolio simulation can now size every trade from the starting balance instead of letting winnings compound, so its return and drawdown read in the same flat-risk units as the Portfolios page, and it can be run at the risk you actually deploy with instead of each strategy's trained risk. Pro.
The portfolio diversity grade is scored on what the strategies did, not on what they are labelled. It grades five things now: how many pairs of members genuinely move together, how much of your capital sits on one market once the same market under two broker names is counted once, how many members have decayed since they were built, how varied the strategies are, and how many you hold. Two members that turn out to be the same bet cap the grade at D and the panel names them, and anything that cannot be measured yet reads as unknown with its counts rather than quietly scoring well.

BUG FIXES

FIXES IN THIS RELEASE

[TOP: The SL CEILING you pick on the launch page now sticks, and is what your campaign trains with.] The SL CEILING choice on the launch page stays selected when you come back to it and is the ceiling your campaign trains with, so Tight Stops and Scalper do what they say.
The incubation ladder now measures each bot's live edge itself, so the EDGE column reads HOLDING, WEAKENING or REPLACE instead of staying blank. It scores your own closed live trades against what the bot did in training, and it holds a bot back from a bigger size while its losing run or its drawdown is worse than its training ever was.
The message shown while a REFRESH runs now says what happens: it takes about a minute, the result opens on its own when the run finishes, and REFRESH on that strategy opens it again any time.
REASSESS now opens its finished verdict when you press the button again, with CHECK AGAIN for a fresh check, instead of starting the check over.
Every message, tip and popup across Algory now reads in plain English. Sentences say what happens instead of trailing off, and the progress and error notes tell you the next thing to do.
The REFRESH button now shows a finished result instead of starting the run again. If you closed the running message and came back, pressing REFRESH replayed it from the start; it now opens the waiting result, with RUN AGAIN there if you want a fresh measurement.
The background-run card at the bottom right now follows the light theme. It was a black bar with unreadable text on the white theme.
Every popup, tip, card and menu now follows the light theme, and the greens, ambers and blues on the white theme are dark enough to read.
A timeframe with no slower chart to check against now trains on its own instead of quietly disappearing from the campaign, and the run names the timeframe it happened to.
When a change to your timeframes makes your higher-timeframe choice impossible, Algory now tells you it has gone back to AUTO instead of changing it behind your back. The launch page and automations both say it.
Under HIGHER TF on AUTO, H2 now trains without a daily context layer, which measured better on every test; you can still pick H4 or D1 for it by hand.
The daily timeframe is no longer offered as a training timeframe. It has never produced a single strategy, because the breeding engine has not been able to run it for months, so picking it only cost you an empty campaign. It is still there as a higher timeframe for your strategies to check.
The fast-timeframe warning now says what was actually measured: on a broad basket of markets the 15 and 30 minute charts produce nothing at all on 40 to 64 percent of them, so keep the fast timeframes to a few liquid markets.
One building block, the inside-bar breakout signal, is retired from new breeding: across fourteen thousand strategies it produced a deployable result half as often as the rest of the pool, and other users' deployed strategies showed the same. Strategies you already have that use it are untouched and still compile exactly as trained.
Index strategies no longer train on price bars your broker never shows. Two of the built-in histories carried flat, made-up bars inside the daily closed hour, so a strategy could learn a pattern that only exists in the backtest; those bars are now stripped on load, and a new check compares the training history against your broker bar-for-bar.
Brokers with shallow minute-level history no longer lose every market in a campaign. Algory's built-in history now tops up the missing years the way it was always meant to; before, it wrongly decided your broker had enough, then skipped each market for not having enough - and told you to fix it by scrolling charts.
When your broker sends no price history for a market and it runs on Algory's built-in data instead, the Max-bars popup now appears with the exact MT5 steps to pull the history down - before, the switch happened silently.
The "not enough history" message now names what is actually limiting you: if your broker's 5-minute history is the bottleneck, it says so with its start date, instead of a bare years count that contradicts the chart on your screen.
The STOP button now stops the campaign every time. Three separate gaps let a run keep going after a green stop - a campaign started from the classic dashboard could not be stopped from the new interface, a stop could report success without killing anything when the process check misfired, and the classic dashboard's stop could silently do nothing after a restart. All three now find the running engine and confirm it is actually gone before reporting stopped.
Markets no longer switch to built-in history while your broker actually has the data. A slow or freshly-restarted MetaTrader could lose a download race and Algory concluded the broker had nothing; it now checks whether the broker really serves the market and waits out the download when it does - and when a switch does happen, the message says the terminal had not finished downloading, not that your broker sent nothing.
A risk number you never typed can no longer reach a live account. Older deploys could carry an invented 0.5% ladder target; the ladder now refuses to assert it and asks you to redeploy with the risk you actually want. Deploying without stating a risk is refused instead of silently defaulting, and the classic dashboard now shows you the exact live risk it is about to set and asks you to confirm it - a number restored from old settings no longer counts as your choice.
On Linux, where Algory runs through Wine, it now starts the background service and opens Algory in your default browser instead of opening the app window, which could only ever show as a black screen there.
Settings save themselves. Both save buttons are gone from the settings page - every change, including the MetaTrader folders, is written as you make it, with a SAVED tick where the button used to be. A folder that does not exist yet is not saved until it does, and a MetaTrader change made while a campaign is running is held until the run finishes so a live campaign is never redirected.
Algory now tells you when a MetaTrader folder is a copy rather than a second installation. Copying a MetaTrader folder does not create a second terminal, and the setups it produces are hard to diagnose; when you point Algory at a folder that is an exact copy of another MetaTrader on the same computer, it says so and gives you the one thing that works - run the MetaTrader installer again and choose a new folder.
Choosing which MetaTrader Algory manages now actually moves your deploys. The terminal list names each one by its broker account and shows its data folder, so two terminals from the same broker are no longer told apart by a random folder name, and picking one moves the Experts folder and MetaEditor with it.
Incubation now says that it runs itself. There is no INCUBATE button because a full deploy starts it for you, and nothing said so: deploy, look for the button, find nothing, and it looks like nothing happened. The deploy confirmation now tells you the bot has started on the ladder, that there is nothing to press, and where to watch it; an empty ladder says why it is empty instead of just sitting there; and the incubation tab carries a GUIDE link to the written walkthrough.
The incubation ladder now tells you when a step up will not change a bot's size. On a small risk target the tier share falls below the smallest size Algory will deploy, so it is held at that minimum - two rungs then look identical and graduating moves nothing. The risk column now says AT MINIMUM when that is what you are looking at, and the graduate button says SAME SIZE. Bots the ladder is deliberately holding because it never got a risk you typed are now listed on the page too, each with what to do about it, instead of only appearing when you pressed graduate.
Algory now starts even when another program on your PC is already using the network port Algory normally uses. That used to open an empty window with a stray Not Found message on it, and reinstalling never helped; Algory now moves to a free port automatically, and when every port is taken, names what to close instead of opening a blank window.
[TOP: Retraining is gone. Check a bot with REASSESS, and replace it when the edge has faded.] Retraining has been removed. Rebreeding a strategy that has stopped working rebuilds it around an edge that has already gone, so the honest move is to replace it. Run REASSESS on the portfolios page to check a bot on data it has never seen; when the edge has faded, take that bot off its chart and breed a replacement. Strategies you bred with retrain in the past are untouched - they stay in your vault, and you can still inspect and deploy them.
Automations can no longer be set to retrain your strategies on a schedule. Retraining rebreeds a strategy from scratch, so one scheduled run could hand you a whole portfolio of results to judge at once. Check whether an edge is still there with REASSESS on the portfolios page, then retrain the ones it flags. An automation you already saved with a retrain step shows it as retired, and asks you to remove the step.
The launch page now tells you how big a campaign you have set up before you start it. A SCOPE line under the targets counts your markets against your timeframes, notes the higher-timeframe layer when it is on, and warns when the shape is large enough to run for days without producing anything - with the fix, which is to split it into two campaigns, one on the hourly timeframes and one on the faster ones. It never blocks a launch.
The portfolios page no longer falls apart when a lot of finished results are waiting for your decision. With a full book's worth waiting, the panels above them were squeezed to nothing and their buttons ended up painted over the results; the page now keeps the portfolio, member and inspector panels at full size and scrolls to the results underneath.
The ONLINE / OFFLINE pill at the top of the page no longer flickers to OFFLINE for a moment while a campaign keeps Algory busy. It now waits for three missed checks in a row before it says OFFLINE, so a single slow reply on a laptop reads as what it is: nothing.
Deploying and parity testing now compile on every MetaTrader install, including the ones under Program Files. When the MetaTrader folder had a space in its path the compiler quietly did nothing, so DEPLOY reported "compile not confirmed" and PARITY TEST reported "no .ex5 produced" for every strategy. The compile now works there, waits properly for the file, and when it still fails the message says whether the compiler ran at all.
A parity test that stalls now tells you why. The test used to be cut off after twenty minutes with only "bench timed out", hiding the real reason; it now reports the cause it found, such as no tester report appearing, and it refuses to start at all when MetaTrader is not logged in, instead of burning twenty minutes per strategy against a terminal that cannot answer.
Parity ERR and PARKED marks now say what went wrong. The reason appears in the portfolio banner and opens in full when you click the mark, so you no longer have to hover over each one.
The Algory Pilot chart now opens on a symbol your broker actually prices. It used to open a plain EURUSD chart, which stays blank on brokers that name it differently, so the Pilot never started and REDEPLOY PILOT only added another blank chart each press. It now uses one of your deployed bots' symbols, rewrites the same chart instead of adding one, and only says VERIFIED once the Pilot has confirmed in MetaTrader.
REFRESH now re-runs a strategy bred on built-in data against the broker you are connected to now, with the exact parameters it was bred with. If your broker's history does not reach as far back as the original, the result says the date it measured from. A refresh that holds up on your broker makes that strategy deployable there; one that does not stays blocked and says why.
The ladder page now shows every risk change from the last 14 days, calls out raises, repairs a bot whose live risk drifted from its ladder position, and flags bots that need a redeploy to reach their tier's risk.
Pairs you add on the launch targets now stay. Added symbols used to vanish whenever the campaign selection reset, so every session started with re-typing them; they now persist as selectable chips until you remove them.
Smart Mode now heals its own learning file. A single damaged byte in that file used to print "Fitness v2 write error" on every strategy for the rest of the campaign - and every campaign after - while Smart Mode quietly stopped learning. It now quarantines the damaged file, restores the newest intact backup, and carries on; the save path is also hardened so a crash mid-save can no longer damage it. Custom stop-loss profiles additionally keep their learning file in the right folder now, so resetting or inspecting Smart Mode data covers them too.
Algory no longer puts its icon on windows that are not Algory. The window branding introduced for remote desktops matched by window title, so another program with "algory" in its title bar - a code editor with an Algory file open, a browser tab on the website - could suddenly wear the Algory logo and become impossible to find. Branding now checks which program owns the window first.
The incubation ladder no longer changes a bot's live risk without telling you. The ladder page now shows every risk change from the last 14 days - raises called out - and repairs itself: a bot whose live risk did not match its ladder position (a graduated bot still trading small, for example) is corrected on the next visit to the page. Bots that cannot reach their tier's risk because they were deployed with a lower ceiling are flagged with the one fix that works: redeploy. And a bot copied to a MetaTrader that Algory does not manage now says so plainly in its Experts log instead of silently staying at its deploy risk.
Restarting MetaTrader (or reloading a bot) can no longer produce duplicate pending orders. A bot's resting order lives at your broker, but the bot's memory of it did not survive a restart - so after a restart it could place the same order again, and both could fill together. Bots now recognise their surviving order on startup and carry on managing it, remove any duplicate they find resting, and cancel an order that has outlived its entry window instead of acting on a stale signal.
If Windows security blocks part of Algory from opening, the message now explains it properly. Reputation filters sometimes flag freshly released files, and the raw Windows wording was shown on screen - which read as an accusation. It now says what happened, that Algory is signed by VAGAFX LTD, and exactly how to allow it.
On a VPS or remote desktop, Algory's window now opens with the Algory icon and name, and opens immediately. The window carries Algory's own taskbar identity too - its taskbar button shows the Algory logo, never the browser's. Some remote machines cannot show Algory's native window, so a browser-based window is used instead - but it carried the browser's own branding, and every launch first spent several seconds trying the native window before giving up. The window is now branded as Algory, and once your machine has shown it needs the browser window, launches open it straight away.
If you have more than one MetaTrader, picking which one Algory manages now takes full effect. Choosing a terminal moves the Experts folder and MetaEditor with it - previously the choice was recorded but deploys kept going to the other terminal - and Algory tells you what it moved. Anything you set by hand yourself is never overwritten.
Algory now recovers on its own when its window is open but its background service has stopped, which could happen after an update. The settings page showed SERVER UNREACHABLE and nothing you changed would save; relaunching Algory looked like it worked but fixed nothing. Launching Algory now detects the stopped service, restarts it, and opens a fresh window.
Pinning a MetaTrader now moves the Experts folder with it. If you have more than one MetaTrader, choosing which one Algory uses saved the terminal but left the Experts folder pointing at the other one, and nothing said so - so a full deploy could put your bots in the account you had just switched away from. Algory now moves the Experts folder to match the terminal you picked and tells you it has done it. An Experts folder you set yourself is never overwritten.
The strategy inspector no longer shows a training split that no longer exists. Strategies without a recorded split were being drawn and labelled as though they had a middle "SEL" section and an 80% boundary, neither of which Algory has used since June. They now read as training and held-out at 70%, and genuinely old strategies still show the split they were really built with.
REFRESH works on strategies bred on Algory's built-in data. It was re-running them against your broker's price history instead of the history they were actually built on, and since your broker does not carry that far back the re-run could never cover the strategy's own training period. It now re-measures each strategy on the data it was bred on, which is what REFRESH was always meant to do.
If Algory cannot show its own window it now opens the dashboard in a browser that can actually draw it. On machines without Microsoft Edge, the dashboard could land in a browser too old to display it and came up as a wall of plain text. Algory now finds Chrome or Brave first, and if nothing suitable exists it shows plain instructions telling you what to install - your strategies and the engine keep running the whole time.
Opening Algory no longer piles up duplicate dashboard windows. If a window was already open, launching again brought a fresh one alongside it; now the existing window is closed first and the new one opens focused, in front of your other windows.
REFRESH now shows its result. The popup rendered its title and a CLOSE button with nothing between them, both while running and when finished, so a refresh looked like it started and then did nothing. That also hid the reason a refresh could not run, including the common one: your broker no longer carries price history far enough back to re-run that strategy's original training period.
The risk you type into DEPLOY ALL is now what the incubation ladder aims at. If a bot had been deployed before, its old target won instead, so the ladder sized live bots off a figure you never entered - deploy 21 bots at 0.4% and the incubation window could show targets of 0.25% and 0.5%. Bots already partway up the ladder still keep their level; only the target changes.
A bot climbing the incubation ladder no longer ends up at half the risk you asked for. Each promotion recalculated the target from the reduced risk it was currently trading, so a 0.5% bot promoted to 0.25%, then treated 0.25% as its target and finished at 0.25% instead of 0.5%. The target is now recorded once and every promotion and step-down reads the same number.
Deleting strategies now warns you first if any of them are in a portfolio, names the portfolios, and asks once for the whole selection - in both the Command Deck and the classic interface. Purging has always protected portfolio members; deleting did not check at all, so selecting everything in the All Strategies tab and deleting could empty your portfolios. The strategies themselves go to the recycle bin as before, but portfolio membership cannot be restored from it.
The health check now always says whose trades it is counting. When it reports a low trade count it states, before the number, that this is what its own replay found on data since training and not the trades on your account, and it says when Algory cannot see your bot's live trades at all rather than showing a live count of zero it has no basis for. Bots deployed before August could never see that explanation, which is why the count read as a broken feature.
If MetaTrader is running in portable mode and Algory's Experts folder belongs to the normal-mode data folder (or the other way round), a full deploy now stops before it touches anything and tells you which folder to set. It used to close your MetaTrader, reopen it in the other mode - a different data folder, usually one that has never been logged in - and then time out several minutes later naming a path that meant nothing.
The deploy progress window now waits in proportion to the size of the portfolio. A large portfolio could exceed the old fixed fifteen-minute limit while MetaTrader was still loading the charts, and it was reported as a failure when it was not one.
If you run more than one MetaTrader and point Algory at the second one, full deploys now go to the one you chose. Before this, Algory quietly sent them to whichever MetaTrader already had the Algory Pilot running, which was always the first one - so a newly chosen MetaTrader could never receive its first deploy, and the compile then timed out looking for the file in the folder you had picked. Algory still redirects when it detected your paths itself and the Pilot turns out to be somewhere else; it now leaves your own choice alone.
OPEN MT5 opens a portable MetaTrader in portable mode. Opened without it, MetaTrader loads a different data folder: no broker account, no charts, and every deployed bot appearing to have vanished.
The PORTABLE setting no longer unticks itself. If you set it yourself, automatic detection stops overriding it; and when what Algory can detect disagrees with what you set, the settings page now says it could not tell instead of silently changing your answer. Only actually watching your MetaTrader run - which happens whenever it is open with a broker connected - can change the setting on its own, in either direction.
If you point Algory at the Experts folder inside a MetaTrader's installation directory, Algory now tells you when that MetaTrader actually loads its robots from a different folder, and names the right one. Deploys into the wrong folder used to compile and report success while the robot never appeared in MetaTrader.
When you have chosen a MetaTrader in settings, Algory's broker check no longer opens a different MetaTrader in the background and reports its account as your live connection. It reads the MetaTrader you chose, or one that is already running, and says it is not connected when neither answers.
If your MetaTrader is set to keep a limited number of bars per chart, Algory now says so before the campaign starts, in a popup with a picture of the exact MetaTrader setting to change. Before this, a capped MetaTrader quietly starved the campaign of price history and every strategy came out marked as built-in-data and undeployable, with nothing telling you why. The engine already coped; you were just never told at the moment you could fix it.
Selecting several strategies at once works again in the Command Deck vault. Hold Shift and click, or hold Shift and press the arrow keys, to select a range; Ctrl and click adds one at a time; Ctrl and A selects everything your filters are showing; and right-clicking inside the selection offers DELETE for all of them at once. The count above the table says how many you have selected.
The health check in the classic interface now reads your price history the same way the Command Deck does. It could report a strategy as having taken no trades on fresh data when your MetaTrader had simply stopped receiving prices, and it printed a score beside a trade count of zero; it now tells you the feed has stopped and the date it stopped, and shows no score where there is nothing to score.
If Algory stops responding while a campaign is running, the campaign panel now says so plainly instead of carrying on as if the run were live: the buttons grey out rather than failing when pressed, and no button gets stuck reading STOPPING. It all comes back on its own the moment Algory answers again, and your campaign keeps running throughout.
The Portfolios page no longer freezes for several seconds when you press STRATEGIES after spending time on HEALTH or INCUBATION. The longer you had been on the other tab, the longer it used to hang.
The small question-mark help buttons on the portfolio maintenance cards, the simulation tiles and the campaign chips are no longer overlapped by the text beside them.
If Algory ever fails to open, the launch now leaves a step-by-step record beside the app of how far it got, so support can see exactly where a machine stopped instead of guessing.
[TOP: Algory no longer opens by itself after you close it.] Algory no longer opens by itself after you close it. An hourly background task that reads your closed trades from MetaTrader was accidentally launching the app instead of running invisibly, whenever Algory was closed and no campaign was running. It now runs silently as intended, which also means live trades from bots attached by hand now actually reach the health system.
The incubation ladder holds its place while the Portfolios page loads, and opening it scrolls inside its own panel instead of squeezing the panels around it.
The engine log on the settings page now runs to the bottom of the page instead of stopping at a fixed height with its own scrollbar beside empty space.
The campaign panel no longer shows PREPARING NEXT ASSET when nothing is running; that message now only appears during a live campaign's move to its next market.
An automation whose campaign could not start no longer gives up on its remaining steps: the step is marked failed in the activity log and the portfolio and maintenance steps still run, so a scheduled portfolio build is never lost to an earlier stall.
An automation that is due but cannot start yet now says so in the activity log, naming the reason - a campaign already running, or work ahead of it in the queue - instead of waiting in silence.
Improved the reliability of saving the campaign queue while it is being read at the same moment, which could interrupt an automation step.
The combined equity figures now follow the view you picked: choosing HELD OUT shows the held-out return and drawdown alone instead of leaving the full-history totals beside a held-out curve.
The portfolio grade now reads your Symbol Lock deploy setting rather than the setting your strategies were trained with. The panel names the setting it scored under, and FULL DEPLOY and the RUN SIM options are where you change it.
Your deploy guardrail choices are remembered as your defaults for next time. The ticks you set in FULL DEPLOY are saved once the deploy succeeds, so the next deploy opens on what you actually use and the portfolio panel's lock label matches it. A cancelled or failed deploy changes nothing.
Switching themes now repaints every chart immediately - no more panels keeping the old background.
Health checks are faster and no longer report a strategy as untradeable when it is trading fine. The check used to replay three to six years before running the part that matters, and a single rough patch anywhere in those years stopped the replay dead, so the strategy came back with no trades and no verdict. It now starts just before training ended, which removes that failure and makes a check take seconds rather than a minute or more. A strategy whose file does not record when its training finished is told to retrain instead of being checked against a window that cannot be fair.
A parity test that is waiting now tells you what it is waiting for. Queued tests that cannot start because your bots are live on charts, or because MetaTrader is busy, used to read only "queued"; the reason is now on the status line and in the corner notice, in full.
[TOP: Stuck parity tests now tell you why they are waiting, and you can run them anyway with one press.] Parity tests no longer get stuck behind bots you already removed. Algory's record of what is on your charts only updates when MetaTrader saves them, so taking bots off charts by hand while it is open left parity refusing to run forever. When that is what it is waiting for, you now get a MY CHARTS ARE EMPTY - RUN ANYWAY button; it asks you to confirm that nothing is trading, then runs the tests you have queued. It applies to that queue only.
Parity testing now tells you up front that it needs a Pro plan. Starting one on the free plan used to queue the test and then leave it stuck reading "test could not finish"; it now says what it needs before anything is queued.
Campaigns on 15 and 30 minute charts now start correctly regardless of where the price history begins, and higher-timeframe checks work on every such campaign.
Crypto campaigns now reach as far back as every other market. The training window was a fixed number of candles rather than a length of time, so on a market that trades all week it covered about 2.4 years where forex and gold got 3.4, and Algory refused to breed on it and told you your broker was short of history. The window is measured in years now, so it reaches the same distance back whatever you trade, and no market ends up with a shorter window than it had before.
The health check in the classic interface now explains its trade count, the same way the new one does. When it says too early it states when the strategy was trained and how much new market data there has been since, says plainly that it is counting the trades its own re-run found rather than the trades on your account, and where your bots were deployed through Algory it shows both numbers side by side so a low count next to a busy account makes sense.
Campaigns now make full use of your broker's price history: Algory asks your terminal to download missing minute data before starting, keeps your broker's costs even when deep history has to be supplemented, and tells you before a campaign would train on built-in data instead.
Deploying a strategy onto a different broker than it was trained on now asks you first, the same way the classic interface always has. A strategy's edge is proven against one broker's spreads, commission and swaps, so the safe path is retraining on the broker you are connected to; you can still deploy anyway.
The combined equity chart on the Portfolios page no longer smears when you move the mouse over it. Hovering was repainting the chart on top of itself, so the shaded held-out region darkened with every movement until it covered the curve, and the gridlines doubled up.
The live parity line on the Portfolios page works again. It stopped updating in the last update, which also meant STOP PARITY TESTS never appeared while tests were running.
The REFRESH card on the Portfolios page does something when you press it. It had never been connected.
Deploy and MetaTrader path messages now name the folder involved and what to do about it. A deploy that runs out of time says which file in which MetaTrader it was waiting for instead of just telling you to check MetaTrader, and if a compiled file arrives from another installation it says so. If the Experts folder you chose is no longer on your computer, Algory says that and asks you to pick it again, rather than telling you to let it detect one. Choosing a folder that is not part of any MetaTrader Algory can find now warns you at the time you choose it, and still lets you use it.
Algory opens in its own window on a VPS or remote desktop. It was giving up before a slower machine could finish opening it, so you got the interface in a browser window instead, or nothing on screen at all while Algory sat running in the background, and clicking the icon again while it was starting sent it back to the beginning.
Pausing a campaign twice can no longer trap it in a freeze that RESUME could not undo. A second pause on an already-paused campaign now changes nothing, any resume fully wakes the campaign however it got stuck, and a deploy that briefly pauses a running campaign while it restarts MetaTrader no longer cancels a pause you chose yourself.
[TOP: Algory opens straight into its own window on machines that need software rendering, instead of pausing or opening in a browser.] Algory now opens straight into its own window on machines that need software rendering, instead of pausing or opening in a browser.
Running two MetaTrader installations no longer triggers a repeated install-mismatch prompt at startup. Algory now checks that your MetaEditor and Experts folder belong to the same MetaTrader installation and remembers your choice, and picking either path in Settings sets the other to match so the pair cannot be left half changed.
Deploying to MetaTrader now finds the compiled file even when a second MetaTrader installation is present, instead of timing out. If the file is written to the other MetaTrader, Algory moves it where it belongs and the deploy carries on.

UI CHANGES

INTERFACE CHANGES

Every popup in both interfaces has been rewritten to be shorter and clearer: the thing you need to know or do comes first, long paragraphs are split, step-by-step instructions are bullet lists, and the words that matter are bold, with red kept for the things you must not miss. 124 of 343 popups changed; the rest were already fine.
The settings page's action buttons - SAVE CHANGES, RELOAD FROM DISK, RESTORE DEFAULTS, RESTART TUTORIAL and OPEN DATA FOLDER - and the account controls now sit at the top of the page instead of below every panel, and the page is organised under clear section headings: BREEDING, BROKER AND COSTS, VAULT MAINTENANCE, SYSTEM, SMART MODE and SUPPORT.
The vault purge criteria gain a BUILT-IN DATA BRED switch that clears out strategies bred on Algory's built-in price history, the ones that cannot deploy until retrained. Strategies whose history was merely topped up while your broker stayed connected keep your broker's costs and are not touched.
Automated parity testing is gone from the Automation page, along with its weekly vault sweep. Parity testing is something you do to a portfolio: open the portfolio and press PARITY TEST there, or add a parity step to an automation and point it at the portfolio you want graded. Nothing sweeps your vault in the background any more.
SMART and its ON OFF switch stay on the same line as each other in the Command Deck campaign row, however narrow the window is.
The engine log in settings has a COPY button.
DEPLOY ALL in the classic interface now offers the same two choices the Command Deck does: compile the portfolio into MetaTrader, or a full deploy that sets the risk and guardrails for every member, starts them on the incubation ladder if you want, puts them on their charts and confirms in MetaTrader that they are running. Before this it could only ever compile them.
Deploying while a campaign is running no longer leaves your bots sitting on their charts unstarted. The campaign is paused for the few seconds MetaTrader takes to restart and picks up on its own, the same as it always has in the classic interface.
[TOP: The equity chart has new visuals, glow, sweep and a moving shine. Turn effects off in settings and you get the plain fast chart.] The combined equity chart has new visuals: the curve glows, sweeps in when it changes, and carries a moving shine along the line. While a portfolio simulation runs, the chart dims and says what is happening under a scanning animation. If you have effects turned off in settings, none of this runs and you keep the plain, fastest version.
The correlation matrix got the same treatment: pairs that move together glow with their own colour and a slow sheen sweeps the grid, while decorrelated pairs stay calm. Effects off keeps the flat grid.
Each portfolio card now names the broker its members were trained for, and says MIXED with the brokers listed when they span more than one.
The REFRESH, REASSESS, PARITY TEST and DEPLOY cards on the Portfolios page are now single-line buttons, freeing vertical space for the tables around them.
Expiry columns read 89D instead of 89d LEFT, so the tables sit tighter.
The Portfolios page now has three sub-tabs: STRATEGIES (your list, the members and the full inspector), HEALTH (the combined equity chart and the diversity panel) and INCUBATION (the ladder, open and full width).
The risk box above the combined equity chart is simply RISK now, starts at 1%, and every view of the panel is read in it; press TRAINED to read the panel at each strategy's own trained risk, and either choice is remembered per portfolio.
RUN SIM sits beside the chart its result lands on, and its settings dialog lines up label-left, control-right. The REASSESS ALL header button is gone; set SCOPE to ALL and press REASSESS instead.
The incubation ladder's decay column now speaks plainly: HOLDING, WEAKENING or REPLACE, with old checks marked in words instead of brackets, and each verdict explains itself on hover.
Clicking a bot on the incubation ladder opens the checks behind its rung: enough live trades, time in tier, making money, no decay warning and trading normally, each marked PASS or HELD with the real numbers, and a closing line saying what it is waiting for.
With AUTO on, the ladder's step-up buttons grey out and name the rung each bot is heading for, so automatic steps and your own never compete. Turning AUTO on asks you first, and sets every bot to the rung its live results have earned.
The incubation ladder is scoped to the portfolio you have selected and says how many other deployed bots the same MetaTrader is running, so a portfolio living on another machine honestly shows none of its own.
Pressing DEPLOY on a single member now offers the same choice the vault does: compile it into MetaTrader, or a full deploy that sets its risk, its guardrails and whether it starts in incubation, puts it on its chart, and confirms in MetaTrader that it is running.
Deploying no longer asks you to set the Algory Pilot up first or to drag anything onto a chart. Algory installs it and attaches it, says plainly when a chart has not confirmed yet, and offers a RETRY ATTACH that only re-attaches: nothing is rebuilt.
The Algory Pilot now names itself in the middle of its chart, so the one chart you need to leave open is obvious.
The AUTOMATION page holds together on a wide screen: the console and the editor sit at a readable width, SAVE AUTOMATION, PREVIEW and CANCEL group to the right of their own row, and automated parity testing folds into a one-line strip you open when you want it.
Opening the Portfolios page selects the top strategy automatically, so the inspector is never blank.
When a simulation's account blows up, the chart now says so where it happened: a red marker names the day the account fell 15% below its peak and the first tile explains that the replay ends there, instead of a curve that quietly stopped years early.
The portfolio simulation no longer ends a run early on a drawdown rule that is not one of yours: a 15% fall from the account's own peak is marked on the chart for prop-firm readers, and the replay runs to the end of the data.
A simulation result the engine itself cannot stand behind is now labelled THIS RESULT IS NOT TRUSTWORTHY, with the reason, instead of being presented as an answer.
The portfolio simulation now replays the whole history it was asked for: a deep drawdown no longer silently ended the run early with positions closed at the wrong prices.
The Discord community button now sits beside the ALGORY title instead of in the top right corner.
TRUE SIM can now replay only the held-out window, so its return and drawdown line up with the held-out figures beside it instead of covering a different period.
Enter the risk you actually deploy at and the whole combined equity panel - including the simulation - speaks in those terms.
Every strategy list now shows its latest health check beside the score, and a decayed strategy's score is dimmed to match.
The Portfolios page is now built around the combined equity view. The chart is much larger, draws a real date axis, and you can hover it for the exact figure on any day. It marks where the held-out region begins and dims the stretch before it, and a switch above it shows the full history, the full history with the held-out run drawn over it, or the held-out run on its own. The lineage panel and the RETRAIN card have left the page: the family tree is in the vault inspector, and retrain is still on any member's right-click menu and in the vault.
The Portfolios members table is managed by right-click now, like the vault. The per-row REMOVE button is gone; right-clicking a member gives you the same list the vault gives you, plus REMOVE FROM THIS PORTFOLIO and an ADD TO PORTFOLIO submenu that shows which portfolios already hold it. You can also choose which columns the table shows from a COLUMNS button or by right-clicking the header, including win rate, broker and the campaign a strategy came from, and the choice is remembered.
The actions that need a Pro plan now say so before you press them. On the free plan REFRESH, REASSESS, REASSESS ALL, PARITY TEST and DEPLOY render greyed with a small PRO tag on the portfolios and vault pages, and clicking one explains in a line what it does and why it needs the plan. The buttons stay where they are and their help still opens.
Each maintenance action explains itself in two sentences. The wall of help text under the action cards is gone; REFRESH, REASSESS, PARITY TEST, DEPLOY and REASSESS ALL each carry a small question mark that opens a short explanation floating above the page, so opening help never moves anything.
One line on the Portfolios page now shows everything Algory is doing to that portfolio: the reassess batch, single refresh and reassess runs, and the parity queue, with how far through each one is and how long is left. It sits in the bottom-right corner with the other notices, so it never covers the members table, and it reads from the server, so work started on another page or before you reopened Algory shows up too.
The Portfolios page reads tidier and answers faster. The maintenance cards are button height rather than panel height, the health line now says why members are too early to judge ("13 too early, they have 5-13 of the 15 trades needed") instead of leaving you to guess, a failed parity test gets a RE-RUN button next to the warning so you no longer have to hunt for the member and start it by hand, the header controls stay lined up on the right when you make the window narrow, and the small print under the combined equity figures has moved behind a question mark on each tile so the numbers stand on their own. The figures the tiles still show beside each stat - the held-out window, covered counts, trading days, median hold - now sit to the right of the number instead of under it, so the tile column is a good deal shorter; on a narrow window they drop back under the number rather than being squeezed. The INSPECTOR beside it uses the space that freed up: it now shows the member's last REASSESS verdict with its date and its PARITY grade, both blank until the member has been checked, and the empty gap under the stats is gone.
REASSESS and REFRESH now run over a whole portfolio, or over the members you pick. Setting the scope to ALL and pressing either one works instead of telling you to pick a single member, and ctrl-clicking several rows runs exactly those. Both go through the same background queue as RUN HEALTH CHECK, one strategy at a time, so a campaign or a parity test still waits its turn, closing Algory part way through picks up where it stopped, and the corner line names which of the two is running and how long is left. A refresh sweep skips members already refreshed in the last 30 days unless the engine has moved on since, and one CANCEL stops the lot and says how many of each it dropped. Pro.
An automation's portfolio step can now feed a portfolio you already have. The PORTFOLIO field is a dropdown of your portfolios plus NEW PORTFOLIO, and when you pick one of yours the step only ever adds to it - it never removes or replaces anyone already in there - while NEW PORTFOLIO still builds a fresh one under a name you type. The activity log says which it did and how many.
Short help notes are available across the app from small question-mark buttons. The vault, settings, campaign, command, automation, insights and map pages now explain what a control does and what a number means in two sentences, floating above the page so nothing moves, and the longer blocks of help text they replace are gone.
The Portfolios page is sub-tabbed instead of one crowded screen, it remembers which sub-tab you were on, and pressing RUN SIM takes you to HEALTH by itself when the result lands, so you never have to hunt for the answer you asked for.
Clicking a member now shows its full passport, the same one the vault shows, in a permanent column beside the table. Score gauge, DNA strip, equity curve, promotion verdict, every stat, the execution settings, the active genes and the actions all sit in one place, so a strategy reads the same whether you open it in the vault or inside a portfolio. The small summary panel and the slide-over panel it replaces are gone, and the last action is REMOVE FROM THIS PORTFOLIO rather than a delete that could be mistaken for it.
A running campaign now shows both timeframes when it is breeding multi timeframe. Where the active task used to read EURUSD [M15] it reads EURUSD [M15 + H1], on the command deck, the campaign monitor, the control page and the classic interface, so you can see which slower chart the run is using - including when you left the choice on AUTO and Algory picked it for you.
[TOP: Every Algory guide is now linked from the GUIDE tab.] Every Algory guide is now linked from the GUIDE tab.
The portfolio panel now shows held-out results, measured only across the stretch where every member is past its own training data, side by side with full history and with the dates of that window. Correlation cells with too little shared trading history now say so plainly instead of guessing, the drawdown tile states that it counts realised profit only, and the worst-correlated pair now names the two exact strategies involved.

GET THE UPDATE

UPDATE FROM INSIDE ALGORY, OR DOWNLOAD FRESH

GET ALGORY