Methodology: how Flash Finance numbers are made
A score you cannot check is only a claim. These pages publish the rules behind every proprietary number on the site (weights, thresholds, data windows and known limits) so that anyone can verify what a figure means before relying on it.
Technical Score MethodologyA transparent 0-100 score: five components of 20 points each, every rule published.Support and Resistance MethodologyClustered swing highs and lows from the last year, ranked by the number of touches.Relative Strength MethodologyA 1-99 percentile rank of weighted 3, 6, 9 and 12-month performance.Market Health Score MethodologySix components, each scored against its own history, weighted into one 0-100 number.Delivery Intelligence MethodologyEach stock's delivery % compared with its own recent normal, as a z-score.F&O Positioning MethodologyPrice × open-interest matrix for futures; strike-wise OI for option walls.FII/DII Flow MethodologyNSE's provisional daily cash-market figures, summed into windows and streaks.Fundamental Ratios MethodologyTrailing-twelve-month results, the latest balance sheet and the latest close.Sector Rotation MethodologyJdK-style RS-Ratio and RS-Momentum on weekly bars, versus the Nifty 50.Multibagger AI MethodologyAn AI research brief for order wins and expansions, graded into verdict tiers, tracked against the market.Market Today MethodologyA daily summary stitched from the site's own exchange-data builds.How Flash Finance Uses AIWhich parts of the site are computed, which are written by AI, and how errors are handled.
Principles
- Describe, don't predict. Data pages describe what the market did. Labels such as "accumulation" or "long buildup" name a pattern in the numbers; they are not forecasts.
- Compare like with like. Stock-level signals are measured against each stock's own history (delivery, volume) or a stated universe (RS rating), not against a single market-wide rule.
- Withhold rather than mislead. When too few inputs are available, such as the market health score with fewer than four of its six components, the number is withheld and the page says why.
- Publish the limits. Where a dataset is short (FII/DII history starts 18 Sep 2026) or approximate (sector breadth from an industry mapping), the page says so.
