Methodology & base rates

What the measurements mean, and how setups like these have actually resolved.

Breakouts are wrong more often than they are right. Roughly half fail. The approach works — when it works — because the winners are much larger than the many small losses, and that asymmetry only shows up over many trades and through real drawdowns.

The universe

Every NSE stock in the EQ series with median 60-day turnover of ₹5 crore or more, above that floor on at least 80% of the last 60 sessions, priced ₹20 or above. The consistency test is what removes stocks whose liquidity arrives in one-day bursts. Membership is stored per date and never recomputed from today's list, so historical measurements are not contaminated by knowing which stocks survived.

How a base is detected

A base is a consolidation after an advance. The detector requires a prior move of 25% or more, then a range of at least six weeks holding beneath its ceiling, with depth between 8% and 40%. Prices are adjusted for splits and bonuses; days locked at circuit are counted and disqualify a base beyond a small tolerance.

The measurements

Tightening — second-half range divided by first-half range. Below 1 means the base is quietening.
Volume dry-up — second-half average volume over first-half. Supply withdrawing is the mechanism the pattern is trying to detect.
Delivery trend — the share of volume actually taken to demat, later half versus earlier. Above 1 means a rising proportion of buyers are keeping the shares rather than trading them intraday.
Contractions — the sequence of pullbacks inside the base, measured at the scale that reveals its structure rather than its noise.

Entry states

A breakout and a buyable breakout are different things. Within 5% of the pivot the stop is tight and the risk-reward is intact. Beyond that the same setup carries more risk for the same reward, which is why extended names are labelled rather than hidden.

Forward returns by year

Every breakout the detector has found since 2018, measured 63 sessions forward. No filtering, no hindsight selection.

Loading…

Which measurements actually help

Edge is the average 63-session return minus the equal-weight universe return over the identical window.

Loading…

Measurements only, not investment advice. Not SEBI registered. Past behaviour of similar setups is not a prediction. Backtest-style figures are hypothetical and exclude costs, slippage and taxes.