What the workspace does
SuperQuant is a crypto research and practice workspace. Its documented browser research edition uses a deterministic backtesting engine and paper-trading history. It does not place exchange orders or produce a verified real-money trading record. Check the current product and model boundaries before applying a workflow.
Market history and observations
Included example market histories are synthetic. Optional public venue observations and user-imported histories have different provenance from one another. An article or chart identifies its data source, time interval and coverage; example illustrations are visibly labeled hypothetical.
Backtest conventions
A result depends on timestamp availability, signal timing, order assumptions, sizing, fees, adverse slippage, funding, stops, liquidation modeling, incomplete bars and the tested period. SuperQuant's research model documents a simplified flat maintenance and liquidation proxy. It does not reproduce every venue's tiered collateral logic, impact, margin interactions, insurance or exact order fills.
Data the engine accepts
Backtests use contiguous one-hour candles only. In the hosted workspace, BTC, ETH, SOL and HYPE load 90 days (2,160 hourly bars) of public Hyperliquid candles and funding, with a clearly labelled Binance USD-M fallback when Hyperliquid is unavailable; a user import is never overwritten. The histories bundled for offline use and the published walkthroughs are seeded synthetic data, 2,160 hourly bars each, not real market history. Imports need at least 60 ascending, unique, contiguous one-hour UTC candles; gaps and duplicates are rejected, not filled. Public observations are read-only market data from Hyperliquid, Binance and Bybit.
How a bar is processed
Signals are computed from the last closed bar, and entries fill at the next bar's open. Entries and exits are filled at the reference price moved against the trader by a set slippage in basis points (default 2 bps). One fee rate (default 0.05% per fill) is charged on each entry and exit; maker and taker tiers are not modeled. Stops, targets and liquidation are checked against each bar's high and low, and when one bar crosses both a stop and a target the stop is assumed to fill first. An open position is closed at the final close.
Funding and margin
Funding is settled hourly at each candle open from supplied rates; a positive rate means longs pay shorts. Missing rates are never treated as zero: a funding-inclusive backtest is blocked until the gaps are fixed or funding is excluded. In paper trading, an hour without a funding rate is skipped without a payment. Margin is a simplified isolated-margin proxy for one position with a flat maintenance rate (default 0.5%) and leverage from 1x to 20x. There is no cross margin, tiering, auto-deleveraging or insurance fund, and a price gap can lose more than the posted margin.
Diagnostics
The walk-forward check splits the later half of a history into four chronological test windows and runs the same fixed rules in each, with no re-optimization. Trade resampling draws closed-trade net dollar outcomes with replacement 1,000 times (at least 10 trades are needed) and reports return and drawdown percentiles and the share of resamples ending below zero. These describe the simulation, not a verified live return.
Strategies and tools
The research engine supports four rule families: moving-average crossover, RSI reversion, range breakout and funding contrarian, with an optional funding filter, percent or ATR stops and trailing stops. It has no general VWAP or Bollinger rule builder, no grid order manager and no portfolio-wide covariance estimate.
Paper trading and accounts
Paper trades are simulated on historical candles that you step through: manual orders fill at the bar close plus slippage, deployed strategies at the next bar's open, and nothing is sent to an exchange. An optional hosted paper observer runs a strategy on public hourly candles and fills at the signal bar's close. Workspace data is stored in your browser. SuperQuant does not hold exchange API keys or connect to exchange balances. A public Hyperliquid address can be read for its fill history, which is labeled unverified, and hosted access can link a Solana wallet by signed message to read $STONKB or staked JUP holdings.
Diary, news and calculators
The Diary keeps your notes and describes patterns in them; it does not diagnose, predict or recommend trades. The news view collects public RSS and Atom headlines on a schedule, every five minutes by default; it is not a real-time feed, and a quiet feed is not proof that nothing happened. Resource calculators apply their formulas only to the values entered in the browser.
Interpret results carefully
Historical simulation, a paper fill, a public quote, forward observation and an actual exchange fill are separate evidence. Costs can be wrong; markets and rules change. A favorable backtest does not demonstrate a persistent edge, future result or suitability for any reader. Preserve rejected tests, data limitations and the chosen benchmark alongside a result.
Changing mechanics
Venue fee rates, margin tiers, funding formulae, order rules and asset eligibility change. Venue-specific pages link to primary documentation and state when it was checked. The venue documentation remains the authority for a live account.
Technical reads, Radar and the Telegram bot
How a technical read, the Radar lenses and the Telegram bot work is documented on the technical reads methodology page.