New Highs and New Lows

New highs and new lows measure how many issues are at their own extremes. A market making index highs with a large number of new highs is being confirmed by leadership. A market making index highs with new highs drying up, and new lows quietly rising, is narrowing. The difference, highs minus lows, is a simple daily breadth number. The cumulative version behaves a little like an advance-decline line built only from extremes, so it reacts less to ordinary up and down days and more to genuine pushes to new territory. A spike in new lows is a selling extreme. A spike in new highs is a thrust. Both can be climactic or the start of a trend. The next readings decide, not the spike alone.

Use the 52-week figures your data vendor publishes, and know whether they are updated intraday or after the close. Intraday new highs expand as the session runs, so a morning reading is not comparable with a published closing reading. Watch the difference for the regime and the raw new lows for stress. A day with many new lows and many new highs is a split market, not a healthy one, even if the difference is near zero. The two sides are both active. After a crash, new lows will stay elevated because the window still contains the old prices. That persistence is information about the damage, and it is also a window effect. Do not expect the count to normalise the week after the low.

Use rising new highs, with new lows quiet, as confirmation of a long bias in individual breakouts. Use a surge in new lows as a reason to stop fading weakness and to respect breakdowns, until the surge exhausts and the market actually bases. A common reversal pattern is a spike in new lows followed by a higher low in the index while new lows contract. That is participation improving. It is the same logic as a breadth divergence, and it still needs the index to stop making lower lows before it is a long. Use the cumulative line as a slow confirmation, not as an oscillator you cross for entries.

Treating a handful of new highs in a bull market as a problem without looking at the history of the count. Some markets run with modest new highs for a long time. Also comparing a 52-week count with a 20-day count and calling them one indicator. And using new highs as a sell because "everyone is at a high." New highs are what uptrends produce. They become a warning when they diverge from the index, not when they exist.

The count is universe-dependent and window-dependent. A new listing can be a "new high" with almost no history, depending on the vendor's rule, or it can be excluded. Corporate actions scramble the 52-week range. The measure ignores how much of the index the new-high stocks represent. Fifty tiny new highs and one giant new low can be a down day for the index. Pair the count with the index and, if you can, with a version restricted to the index members. Otherwise you are measuring a market you may not even trade.