Loading
Another calendar belief, and its 19-year record — a difference of 0.02% per day
Published by The Retail Trader · JUL 24, 2026 · Weekly Outlook
After the weekday test, we asked the same question of the month: does the first half of the month behave differently from the second half? SIP flows, the salary cycle and F&O expiry are all reasons given behind this idea.
The Nifty's 4,622 sessions, split by date:
• First half (1st-15th) — average +0.055% per day, 53.3% of days green, n=2,259
• Second half (16th-31st) — average +0.035% per day, 52.7% of days green, n=2,364
The difference is 0.020 percentage points per day, and 0.6 points in the hit rate. Smaller even than the weekday test.
Two different calendar theories were tested — weekday and month-half — and in both the difference turned out smaller than transaction cost. This is not a coincidence. A great many people are watching these patterns; if there were money in them, it would have been taken out long ago.
In contrast, the same dataset shows a clear pattern at the monthly level: January's average of -1.27% and April's of +3.26% — a 4.5 point difference. So calendar effects are not entirely useless; they just are not found at very short time-scales.
Here too the same question of sample size arises. At the monthly level there are only 19 observations for each month, so even that 4.5 point difference is not a rule, it is background. At the day level there are 900+ observations — so the 'found nothing' there is more reliable, because there was enough data and still nothing came out.
This may seem backwards, but it is the real face of statistics: the smaller the effect you want to find, the more data you need. At the day level we had plenty of data and found no effect. At the monthly level an effect showed up, but the data is thin.
Changing entries or size on the basis of the calendar is not justified by this data. The things that are justified — position size according to the volatility regime, and exposure according to drawdown — come not from the calendar but from the market's own condition.
All figures in this article are computed from The Retail Trader's own daily index database, as of 24 July 2026. Past performance is no indication of the future, and this analysis is for educational purposes only — it is not a buy or sell recommendation.