The VCP study named the constraint: G4. Only about one candidate in five has a pivot low within 5% of the close, and that one gate — not momentum, not strength — is what keeps the book at a handful of admissions a month against twelve slots. Two options were left. One works and one does not, and the numbers say which.
The scanner already finds a contraction — three to five bars whose closes sit within 5% of each other — because WEEKLY hangs its stop there. Its low sits a median 5.5% below the price against the pivot low's 9.3%. The proposal is one line: take whichever of the two is nearer the price. Risk per trade does not widen — the 5% cap still binds — but far more setups fit inside it.
| the shipped shortlist, under each stop rule | n | mean | median | win | worst |
|---|---|---|---|---|---|
| pivot low only (today) | 27 | +9.28% | +6.72% | 59% | -7.2% |
| tighter of the two | 85 | +8.88% | +1.99% | 52% | -10.1% |
Per-trade return is unchanged — +8.88% against +9.28%, a gap of -0.40 points at p=0.907. The extra admissions are not worse trades; they are the same trades with a stop drawn somewhere the scanner was already looking.
The split holds. First half +9.04% (55% win), second half +8.73% (49% win) — both positive and close, on 85 trades rather than 27.
The book gets safer, not riskier. Worst drawdown ₹-79k against ₹-92k, and the largest single trade falls from 26% of the net to 17%.
If admissions cannot be raised, twelve slots is the wrong shape: the capital sits idle. Holding ₹60L and the risk at a full book constant, the slot count was varied from twelve down to three — fewer, larger positions, each risking proportionately more.
| shape | trades | net | win | worst DD | risk/trade | trades | net | worst DD |
|---|---|---|---|---|---|---|---|---|
| 12 × ₹5L | 36 | ₹26.49L | 64% | ₹-79k | ₹25k | 25 | ₹11.79L | ₹-92k |
| 8 × ₹8L | 25 | ₹28.56L | 60% | ₹-120k | ₹36k | 21 | ₹16.84L | ₹-138k |
| 6 × ₹10L | 21 | ₹29.73L | 52% | ₹-160k | ₹49k | 18 | ₹18.34L | ₹-184k |
| 4 × ₹15L | 13 | ₹30.05L | 62% | ₹-149k | ₹71k | 12 | ₹14.61L | ₹-131k |
| 3 × ₹20L | 11 | ₹36.65L | 55% | ₹-199k | ₹94k | 10 | ₹6.49L | ₹-175k |
Columns 2–6 use the new stop rule; 7–9 use today's pivot-only rule.
With the new admission rate, twelve slots is already the best shape: it makes ₹26.49L with the smallest drawdown (₹-79k). Concentrating to six slots buys ₹+3.25L more and doubles the drawdown to ₹-160k while the win rate falls to 52%. Three slots is worse on every measure.
Without option 1 the picture reverses — at today's admission rate, six slots at ₹10L makes ₹18.34L against ₹11.79L for twelve, because twelve slots simply cannot be filled. That is the choice: fix the admission rate and keep the shape, or keep the rate and concentrate. Doing neither leaves ₹60L funding a book that is rarely half deployed.
Return the higher of the last pivot low and the current
contraction low, each times 0.98. Both detectors already exist —
indicators.pivot_low and indicators.contraction_bounds — and WEEKLY uses
the second one today.
The 5% cap, the close-based exit, the disaster line, breakeven +8%, twelve slots at ₹5L. This changes where the stop is drawn, nothing about how much is risked.
The worst trade widens from -7.2% to -10.1%: a contraction low is closer, so it is hit more often, and a gap through it lands further from the entry in percentage terms. The disaster line is what bounds that.
"Best of the best" uses today's RS and EPS, and the twelve months those measure contain the six under test. Every level on this page is therefore optimistic — the same trap the VCP study names. What survives it is the comparison: both stop rules are filtered identically, so the 3.1× in admissions and the p=0.907 on return are unaffected by it. Read the ratios, not the rupees.
Other limits, unchanged from the earlier studies: the population is reconstructed against today's universe, so names delisted since March are absent; fills are modelled at the next open where the 0.2% trigger is met; exits carry 0.5% slippage and a sixty-session horizon.