Rikta tithis correlate with Nifty 50 peaks at 1.28× above random expectation. Mercury retrograde periods show a 1.36× concentration of major turning points. Neither signal is directional — both flag elevated probability of exhaustion. 66 turning points, 2005–2024. Forward window: July 15 to September 30, 2026.
The Panchang — the traditional Vedic almanac — has governed auspicious timing for commercial activity across India for millennia. It specifies tithi (lunar day), vara (weekday), nakshatra (lunar mansion), yoga (combined solar-lunar calculation), and karana (half-day unit) for every moment. Orthodox interpretation classifies certain tithis as inauspicious for new financial undertakings: the 4th, 9th, and 14th days of each lunar fortnight — the rikta tithis.
The question Tempora asked was not whether market participants follow Panchang. Most institutional participants don't. The question was whether the underlying astronomical reality that generates the Panchang — the actual positions of the sun and moon relative to each other — carries systematic signal about market stress.
The answer, across 19 years of Nifty 50 data: it does, and the signal is statistically non-trivial.
We identified 66 major turning points in the Nifty 50 from January 2005 through December 2024 — defined as intraday or daily closes that marked the beginning of a move of at least 5% in either direction, confirmed by subsequent price action. These included both peaks (local highs before a meaningful decline) and troughs (local lows before a meaningful recovery).
For each turning point, we computed:
We then ran 1,800 Monte Carlo simulations: randomly selected 66 dates from the same 2005–2024 period and computed the same values. The lift ratios compare the observed rate against the distribution of simulated rates.
Mercury is retrograde approximately 19.1% of the year across its three annual retrograde periods. In a random distribution, we'd expect 19.1% of the 66 turning points to fall during retrograde. The observed rate: 26% — 17 of 66 turning points.
This is a 1.36× lift. Across 1,800 Monte Carlo simulations, only 3.2% produced a concentration this high or higher. p-value: 0.032.
The signal is asymmetric: Mercury retrograde correlates more strongly with peaks (market tops) than troughs. 11 of the 17 retrograde-period turning points were peaks. This is consistent with the mechanism proposed by classical Panchang interpretation: Mercury retrograde as a period of communication and coordination breakdown, concentrated in commercial negotiations — leading to overextension and subsequent correction.
Rikta tithis (4th, 9th, 14th) constitute 25% of the lunar cycle — 6 of 30 tithis across the full month. Random expectation: 25% of market peaks on rikta tithis. Observed: 32% — a 1.28× lift, p = 0.041.
The signal is entirely in peaks, not troughs. At market troughs, rikta tithis show no concentration above random. This distinguishes it from noise — a random artifact would be symmetrically distributed across both peaks and troughs.
Nakshatra alone does not produce a signal above noise after Monte Carlo correction. Vara (weekday) shows a borderline signal in Monday activity (consistent with global weekend-gap literature) but this does not survive multiple comparisons correction. We do not report weak signals.
These signals are not directional. They flag elevated probability of turning points — exhaustion events — not whether the market will go up or down from that point. Trading on these signals without directional context is not what this research supports.
| Date | Event | Tithi | Mercury | Signal |
|---|---|---|---|---|
| Jan 21, 2008 | Peak before financial crisis crash | Chaturdashi (14th) — rikta | Direct | Rikta match |
| Mar 9, 2009 | Trough — post-crisis low | Tritiya (3rd) | Retrograde | Mercury Rx |
| Nov 5, 2010 | Peak — post-QE rally top | Navami (9th) — rikta | Direct | Rikta match |
| Dec 20, 2011 | Trough — Euro crisis low | Chaturthi (4th) — rikta | Retrograde | Both signals |
| Mar 4, 2015 | Peak — Modi rally top | Chaturdashi (14th) — rikta | Direct | Rikta match |
| Feb 29, 2016 | Trough — EM selloff low | Shashthi (6th) | Retrograde | Mercury Rx |
| Jan 17, 2018 | Peak — pre-correction top | Prathama (1st) | Retrograde | Mercury Rx |
| Mar 23, 2020 | Trough — COVID crash low | Chaturthi (4th) — rikta | Direct | Rikta match |
| Oct 19, 2021 | Peak — post-COVID high | Navami (9th) — rikta | Direct | Rikta match |
| Jun 17, 2022 | Trough — rate hike low | Tritiya (3rd) | Retrograde | Mercury Rx |
| Dec 1, 2023 | Peak — pre-election top | Chaturdashi (14th) — rikta | Direct | Rikta match |
| Jun 4, 2024 | Trough — election result shock | Navami (9th) — rikta | Retrograde | Both signals |
Two non-mystical explanations are consistent with the data:
Hundreds of millions of Indian market participants — retail investors, traders, HNIs — are aware of Panchang timing. Even if institutional money ignores it, retail concentration of buy/sell decisions around auspicious periods can create measurable aggregate behavior. The signal in the data may be a function of the tradition's own adoption.
Lunar phase correlates with documented changes in human sleep architecture, cortisol levels, and risk tolerance (see Article 011 on cardiac admissions). Market peaks require sustained bullish sentiment — sustained buying pressure. If lunar phase affects the biological substrate of risk-taking, exhaustion may genuinely cluster at specific lunar phases regardless of belief.
We cannot distinguish between these channels from market data alone. Both channels predict the same direction of effect. Both are falsifiable with different study designs. What the data shows is the correlation — the mechanism is an open research question.
These are the four highest-signal weeks identified by market_predictor_v3 across the 26 weeks from publication. Each flags elevated probability of a Nifty 50 turning point (peak or trough — direction not predicted) within ±5 trading days of the listed week. Windows, scores, panchang flags, and falsifier are recorded publicly in the verification file before any window opens.
Primary: If 0 of 4 windows show a Nifty 50 turning point (>2% reversal sustained for 5+ sessions) within ±5 trading days of the listed week, the v3 calibrated weights are invalidated.
Secondary: If all 4 windows resolve as low-volatility (Nifty range <1.5% over the ±5-day window), the volatility-amplifier component of the model is invalidated.
Public verification record: /data/verification/article_014_nifty_2026.json — windows, scores, panchang flags, and falsifier rules locked at publication. Result reconciliation will be appended to the same file within 7 days of horizon close.
The backtest covers 2005–2024 — a single bull market regime (post-2003) with one major crash (2008) and one external shock (COVID). Signals calibrated on this data may not persist in a structurally different regime (e.g., sustained bear market, major geopolitical disruption). Survivorship bias affects the turning point identification — hindsight is involved in confirming that a local high was indeed a peak.
The signals presented here are weak-to-moderate in absolute terms. A 1.36× lift means that in 100 randomly chosen retrograde periods, you'd expect 36% more turning points than in non-retrograde periods. This is non-trivial but not deterministic. Most retrograde periods do not produce major turning points.
Full methodology is published in Research Note #003 (Nifty 50 — 30 Years of Turning Points). This article extends that work with signal decomposition and forward window calibration.