Finding stocks that hold up when the index falls
"This stock fell less than the index, so it is strong" is one of the most common mistakes in intraday trading. This explains why, and what to measure instead.
- Why falling less is not strength
- Down-capture: the stock's own baseline
- What the z-score adds
- Reading the scanner
- What it does not tell you
Why falling less is not strength
Suppose NIFTY drops 1% and two stocks are down 0.4%. The first is a low-beta FMCG name that always falls about 0.4% when the index falls 1%. The second is a high-beta metal stock that normally falls 1.4% in the same conditions.
Both look identical on a screener sorted by percentage change. Only the second is telling you something: somebody is buying it hard enough to absorb a fall it would normally take. The first is simply being itself.
Down-capture: the stock's own baseline
The fix is to compare each stock with its own history rather than with the index. From a year of 15-minute bars we measure:
- Down-capture — the share of the index's falling moves the stock normally takes. A stock with down-capture of 1.4 typically falls 1.4% when the index falls 1%.
- Up-capture — the same for rising moves.
Then, for the window you choose, we compare what the stock actually did with what its history predicted. The gap is the signal. A stock that should have fallen 1.4% but fell 0.3% has outperformed its own behaviour by 1.1 percentage points, which is far more meaningful than "fell less than the index".
Low-beta stocks have low down-capture baselines, so falling a little is exactly what is expected of them and produces no signal. The scanner only flags stocks doing something unusual for themselves.
What the z-score adds
A gap of 1.1 points means different things for different stocks. Some track their baseline closely; others scatter around it. The z-score divides the gap by the stock's own normal deviation, so a naturally erratic stock cannot top the list by chance.
- z below 1 — noise
- z of 2 — a clear deviation, roughly a 1-in-20 event if nothing unusual is happening
- z of 3 or more — rare, and worth looking at the stock's news
Stocks that barely track the index at all are excluded by a correlation filter, since "outperformed its baseline" is meaningless when there is no reliable relationship to begin with.
Reading the scanner
The Holding Up Best page has two views. Strength lists stocks that fell far less than expected during the index's down moves — accumulation. Weakness lists stocks that failed to join the index's up moves — distribution, and often the better short candidates.
The columns show the usual capture, the recent capture, what was expected, what happened, the gap, and its significance. The window selector matters: since open for the day's story, 30 or 15 minutes for what is happening now.
If the index barely moved, there were no falls to resist, and the numbers are noise. The header shows how much NIFTY actually fell across its down bars in the window — if that is small, ignore the list.
What it does not tell you
Relative strength is a statement about behaviour relative to a baseline, not about value or direction. A stock can hold up beautifully all morning and still fall in the afternoon. What it does tell you is where buyers are willing to absorb selling, which is more useful as a filter on ideas you already have than as a source of them.
It also says nothing about why. A stock resisting a fall may have results due, a block deal, or an index inclusion. The scanner finds the behaviour; you still have to look for the cause.