Quick answers to the questions we hear most often. If something is missing, tell us through the feedback button in the app.
Those platforms backtest on their own centralised data feeds, which may not match what your actual broker fills you at. Spreads, tick timing, and even bar construction differ between vendors and instruments.
With BYOD you test on the data your broker actually quotes. If your provider's platform exports its tick history, you can test your strategy on that provider's spreads, session cuts and data gaps: DataConvert keeps the bid and ask of every tick from MetaTrader 5 and NinjaTrader tick exports, Dukascopy tick files and any tick CSV with a time, a bid and an ask column, and the spread column of MetaTrader bar exports. An export without quotes (TradingView, crypto exchange bars, plain OHLCV files) has no spread to keep, and a backtest on it charges a configured spread. That narrows one of the most common sources of difference between backtest and live results. It does not close it: slippage, the time an order is executed, partial fills, commissions and the live price feed still differ (Terms, section 8a).
From your own broker or a data provider you subscribe to. The right source is whichever platform you actually trade on: export the historical data from your broker's platform, MT4/MT5 history center, or a licensed data feed you pay for.
Not meaningful at all for real analysis, and that is intentional. The demo data is synthetically generated and does not represent any real market. It exists purely so you can explore the interface and run a backtest before uploading your own data.
Any strategy that looks profitable on demo data tells you nothing about what it would do on a real instrument. Treat it as a sandbox, not a signal.
partiqon works with BAR6 binary files, which you can create from almost any source using the free BAR6 DataConvert tool. It accepts:
MT4/MT5 history exports · NinjaTrader export format · TradeStation format · generic OHLCV CSV · Binance klines · Bybit, Kraken and Coinbase OHLCV exports with Unix-second timestamps
If your broker exports tick data with bid and ask, or bar data, in a layout DataConvert reads (the formats are listed in the DataConvert docs), the converter turns it into BAR6; you do not need to write any code. For a bar CSV in another layout, the converter's window lets you say which column holds the date, open, high, low and close.
Bar files work fine and are much smaller. partiqon works with bars of 1, 5, 10, 15, 30 or 60 minutes, or with tick data. The trade-off is precision: a bar gives one high and one low, not the order they came in, so when a bar opens between a position's stop and target and reaches both, partiqon books the stop, the less favourable of the two. Tick data shows which level was reached first. For most intraday strategies the difference is small, but tight strategies with very close stop-losses will be more accurately modelled with tick data.
For getting started, M5 bars are a practical balance of accuracy and file size. Storage limits are tier-dependent; see pricing.
Storage limits depend on your plan. The pricing page has the current numbers. Free covers enough data for meaningful backtests on your own instruments; Pro and Premium allow larger files, more of them and more storage in total.
Only if you have the right to use it. That means data you personally recorded, data from a provider whose licence explicitly permits the use you intend, or data your broker makes available to you as a client.
partiqon does not provide, distribute, or resell any market data, and does not advise on which external data sources are legally appropriate for your situation. If in doubt, use data from your own broker.
No. partiqon is a strategy research and backtesting tool. It has no broker licence, does not connect to any brokerage on your behalf, and never touches your money or your trading account.
A live strategy runner is on the roadmap but will run locally on your machine (not through partiqon's servers), so you remain in full control of execution and your broker relationship.
No. Backtesting shows how a strategy would have performed on historical data. Past results do not guarantee future results. This is true of all backtesting tools, and partiqon is no exception.
Walk-forward validation (Pro and Premium) and out-of-sample testing measure a strategy on data its parameters were not chosen on. They do not eliminate market risk. Trading always carries risk of financial loss.
No. partiqon does not give investment advice of any kind. Text2Strategy is a translation tool: it converts a strategy description written in plain language into a runnable graph of indicator blocks. The AI does not predict markets, select entry points, or recommend anything. It builds what you describe.
The resulting graph is yours to inspect, modify, and backtest. You decide whether to use it.
Before a Structure Finder run (Premium) starts its search, partiqon tests the in-sample part of each walk-forward fold for 36 fixed conditions: every order of rising and falling closes over the last one to four bars (30 conditions), and a close above the highest high or below the lowest low of the previous 5, 10 or 20 bars. For each condition it compares the move of the bar after the condition with the move of the bar after all other bars (Mann-Whitney U test), with each move divided by the mean absolute move of the 8 bars that end at the bar the condition is judged on, so that volatile and calm stretches are on one scale. A condition is tested only when it occurs at least 30 times, and it counts as found only after a Benjamini-Hochberg correction at 5% for testing many conditions at once.
The correction runs within each window, so the stated 5% is per window: on data without structure about 5% of windows list a condition. A run mines three folds by default (1 to 12 can be set), so on data without structure about one run in ten lists a condition in at least one fold (between 8.6% and 10.3% of runs at three folds, measured on simulated series of 20,000 bars without structure); more folds raise that share. Each condition shows the number of folds that found it; a condition found in every fold is more evidence than one found in a single fold, although the folds’ windows overlap.
Each finding is shown with its figures and then measured on the fold's out-of-sample part, without being tested again; the reserved holdout is never read. Findings do not steer the search, and they describe past data, not a trading signal. When no condition passes the test, the result says so. On tick data the ticks are first built into 1, 5, 15 and 30 minute bars, and all four are corrected as one family. The documentation describes the method in full.
Any modern browser on desktop or laptop: Chrome, Firefox, Safari, or Edge. No installation required. Backtests run server-side so the load on your machine is minimal. Runs continue on the server when you close the browser; you need a connection to start them and to see their results.
Mobile browsers work for reading results but the graph editor is designed for a larger screen.
Yes, uploaded data files are stored on partiqon's EU-based servers so that the backtest engine can read them. Other users have no access to your files, strategies or results, unless you share a result through a link or submit a strategy as a template. The people who run partiqon can open an account's strategies and run history, and run a strategy on the account's uploaded data, to help with a support case.
Data is encrypted in transit (HTTPS). We do not sell your data. Service providers process parts of it on our behalf, for example the hosting provider, the payment provider and the AI provider that Text2Strategy and the Pine import send your text and scripts to, which stores them in the USA; the Privacy Policy names them.
Account deletion is available from your profile settings at any time. When you permanently delete your account, everything stored with it (uploaded files, strategies, backtest results, shared links and published templates) is deleted from our servers; what the law requires us to keep stays, without your e-mail address. There is no recovery after permanent deletion.
If you would rather close your account than delete it, contact us through the feedback button in the app and we will deactivate it. To use a deactivated account again, write to support@partiqon.com and we will reactivate it. The Privacy Policy describes how long data is retained.