Data source
GridBotLab uses public Binance USD-M futures market data collected by the site scanner. The pages do not connect to user exchange accounts and do not use private order or position data.
Research transparency
GridBotLab Research Methodology
How GridBotLab records futures observations, defines events, calculates outcomes and prevents hindsight claims.
Methodology summary
Historical market observations do not predict future performance. GridBotLab publishes factual statistics and clearly separates collecting data, early data and published research states.
Symbol universe
The monitored universe is Binance USD-M perpetual futures that pass eligibility checks for trading status, quote asset and enough available market history. Symbols can enter or leave the universe as Binance listings change.
Snapshot frequency
The collector stores approximately 15-minute snapshots. A missing operational interval is reported by the history health script rather than filled with invented prices.
Quote volume
Quote volume is the 24h futures quote volume reported in the cached market data. It is used for liquidity segmentation and is not market capitalization.
Relative volume
Relative volume compares current or projected activity with the symbol's own recent baseline, so a smaller market can be studied against its normal behavior.
ATR and ATR expansion
ATR14 measures recent absolute movement. ATR expansion compares current ATR with a recent baseline to identify volatility expansion.
MA7, MA25 and MA99
Moving averages describe short, medium and longer market structure. They are used for context, not as proof of future direction.
Expansion Score
Expansion Score combines relative volume, breakout distance, ATR expansion, moving-average context and prior behavior into a research score. It is not a probability or trading signal.
Parabolic and reversal scores
Strict parabolic and capitulation/reversal observations are stored as scanner observations. They are not real trades and are not exchange execution records.
Forward returns
For an event at time T, forward returns use actual later stored snapshots at 15m, 1h, 3h, 6h, 12h and 24h. If the later observation is missing or not mature, the outcome remains missing.
MFE and MAE
Maximum favorable excursion and maximum adverse excursion use observed prices after the event and inside the target time window. Missing future snapshots are not replaced by current price.
Event deduplication
Threshold studies use crossing logic. For example, RV crossing from below 5x to at least 5x counts once, then requires a reset threshold or cooldown before another event.
Liquidity buckets
Research is segmented into <$10M, $10-25M, $25-100M and $100M+ 24h quote-volume buckets because small and large futures markets behave differently.
BTC regime
Where enough data exists, broad regime is labeled by BTC 24h direction and BTC ATR expansion. This keeps the regime model intentionally simple.
Publication gates
Sample sizes below 30 are insufficient. 30-99 is early data, 100-499 is moderate sample and 500+ is large sample. Full 24h studies require matured 24h outcomes.
Limitations
Binance listing changes, short collection windows, delistings, outliers, missing intervals and correlation-versus-causation limits are part of the research record.
Related research tools
Cite this research
Citation: GridBotLab, Research Methodology, accessed 2026-08-12. https://gridbotlab.com/research/methodology