Drawdown Analysis
Natural Gas
NG · Cyclic market
A living read of where this market sits in its own drawdown history, using per-market quintile cuts because there is no industry convention for cyclic markets.
Where this market sits today
As of market close on 11/10/2026.
Native phase
Crisis low
Drawdown from same-week seasonal high
-85.66%
Normalized phase
Historic
Own-history severity percentile
99.7
NG · Drawdown trajectory
Weekly closes of the primary drawdown metric (dd_seasonal) across full market history. Horizontal bands show the phase cuts; the gold marker is today’s reading (11/10/2026).
Scroll or drag the brush below the chart to zoom. All data from the Market Lab Analytics drawdown atlas.
Time in each phase
Share of this market’s full history spent in each own-history quintile. Because these buckets are defined by the market’s own quintile cuts, each phase holds roughly 20% of history by construction. The depth ranges shown are what matters: they reveal what a “Normal pullback” or a “Crisis low” actually means in percent terms for THIS market.
Based on 9,523 daily observations (38 years of history).
The numbers
Every drawdown lens we compute for this market, all at once. The primary drawdown metric (bolded by convention in the card above) is what drives the native phase label; the rest are context.
What the labels mean here
Cyclic markets like this one do not have an industry convention the way equities do. There is no equivalent of “correction equals -10%” for corn or coffee. Pretending otherwise is dishonest.
So the native phase uses this market’s OWN drawdown history, divided into five quintiles. Shallowest 20% of this market’s history gets “At cycle highs,” deepest 20% gets “Crisis low.” The labels mean the same thing across cyclic markets, but the depth thresholds are per-market because the markets themselves have different pain distributions.
- At cycle highs. Shallowest 20% of own history
- Normal pullback. 20 to 40th percentile
- Cyclical low. 40 to 60th percentile
- Stressed low. 60 to 80th percentile
- Crisis low. Deepest 20% of own history
The normalized phase uses the market’s full all-time drawdown as the severity denominator. The current severity of 99.7% means today’s drawdown is deeper than 99.7% of this market’s all-time drawdown history.
Today: Crisis low in this market’s own cycle terms, Historic on the full-history severity scale.
What historically happens from here
When NG has historically been at today's severity decile, here is what happened next. Means are over daily forward windows; confidence intervals use a 20-day moving-block bootstrap to respect the overlap in overlapping-window returns. Base-rate numbers, not forecasts — the past regime is one sample.
The full 10-decile table below shows the pattern across the market's entire history, so you can see whether forward returns move monotonic-ally with severity or follow a different shape.
Today · Decile 10 · Severity 100%ile
Forward 20-day mean return
-0.27%
95% CI [-0.80, +0.23] · 46% win rate
Forward 60-day mean return
-0.70%
95% CI [-1.58, +0.40] · 42% win rate
Based on 5,266 prior days where NG was in this severity decile. Green = 95% CI excludes zero on the positive side; red = negative side; grey = indistinguishable from zero.
| Decile (sev) | N | Fwd 20d mean (95% CI) | Win rate 20d | Fwd 60d mean (95% CI) | Win rate 60d |
|---|---|---|---|---|---|
| D1 · 0-10%ile | 462 | +0.05% [-0.46, +0.80] | 49% | -0.69% [-1.48, +0.67] | 44% |
| D2 · 10-20%ile | 222 | -0.27% [-0.43, +0.13] | 38% | -0.62% [-1.27, +0.37] | 32% |
| D3 · 20-30%ile | 87 | +0.20% [-0.53, +0.35] | 51% | +0.02% [-0.69, +1.25] | 51% |
| D4 · 30-40%ile | 246 | +0.12% [-0.50, +0.46] | 50% | +0.03% [-0.82, +1.18] | 52% |
| D5 · 40-50%ile | 383 | +0.14% [-0.24, +0.67] | 54% | -0.22% [-0.99, +1.39] | 52% |
| D6 · 50-60%ile | 478 | -0.12% [-0.33, +0.43] | 50% | -0.39% [-1.34, +1.09] | 52% |
| D7 · 60-70%ile | 422 | -0.53% [-2.28, +2.21] | 51% | -2.30% [-5.74, +2.27] | 46% |
| D8 · 70-80%ile | 430 | -0.96% [-2.43, +0.96] | 45% | -4.22% [-7.11, -0.79] | 36% |
| D9 · 80-90%ile | 830 | -0.27% [-1.19, +0.94] | 40% | -0.16% [-1.89, +2.17] | 42% |
| YOU ARE HERE › D10 · 90-100%ile | 5,266 | -0.27% [-0.80, +0.23] | 46% | -0.70% [-1.58, +0.40] | 42% |
Cross-market drivers
When NG enters its deepest-decile drawdown state, which other markets systematically move with it? Below are the markets whose daily log-returns shift most reliably in the same deep-NG-drawdown window, filtered to those with a 95% confidence interval excluding zero.
Shifts are measured in basis points per day (1 bp = 0.01%). Positive = that market strengthens alongside NG’s deep drawdown. Negative = that market weakens alongside it.
Weakens alongside NG deep drawdowns
MW · Wheat (Minneapolis)
Daily log-return weakens by -6.2 bps/day when NG is in its deepest-decile drawdown state.
95% CI [-13.4, -0.1] bps/day
TY · 10-Year Treasury Note
Daily log-return weakens by -4.5 bps/day when NG is in its deepest-decile drawdown state.
95% CI [-7.6, -1.8] bps/day
JY · Japanese Yen
Daily log-return weakens by -3.4 bps/day when NG is in its deepest-decile drawdown state.
95% CI [-6.8, -0.3] bps/day
EBL · Euro Bund
Daily log-return weakens by -3.3 bps/day when NG is in its deepest-decile drawdown state.
95% CI [-5.7, -0.6] bps/day
JGB · Japanese Gov Bond
Daily log-return weakens by -2.9 bps/day when NG is in its deepest-decile drawdown state.
95% CI [-4.9, -1.1] bps/day
Strengthens alongside NG deep drawdowns
KC · Coffee (Arabica)
Daily log-return strengthens by +18.6 bps/day when NG is in its deepest-decile drawdown state.
95% CI [+4.0, +35.5] bps/day
NKD · Nikkei 225
Daily log-return strengthens by +11.9 bps/day when NG is in its deepest-decile drawdown state.
95% CI [+4.7, +19.0] bps/day
NQ · E-mini Nasdaq-100
Daily log-return strengthens by +8.0 bps/day when NG is in its deepest-decile drawdown state.
95% CI [+2.8, +12.6] bps/day
ES · E-mini S&P 500
Daily log-return strengthens by +5.6 bps/day when NG is in its deepest-decile drawdown state.
95% CI [+2.3, +10.0] bps/day
How we think about drawdown
Drawdown math is not a one-size-fits-all formula. The same equation that honestly describes the S&P 500 is actively misleading for corn, and both of those break for the Dollar Index. So Market Lab Analytics computes drawdown three ways and chooses the one that fits each market’s nature.
For cyclic markets the primary drawdown is seasonal: price divided by the same-week-of-year historical maximum. This avoids dishonest reads where a corn price that is trading normally for its current supply-demand cycle still shows as a massive drawdown from an unrelated weather spike years earlier.
Two complementary pain-over-time metrics appear in the Numbers grid above: the ulcer index (Peter Martin, 1987; root-mean-square of drawdown over a trailing window, emphasizes deep drops) and the Underwater Volume Index (Dstendahl; linear integral of |dd| over the current episode, weighs long shallow bleeds on equal footing with short deep ones). A -5% drawdown that drags on for years can produce a larger UVI than a -20% drawdown that resolves in weeks; ulcer ranks the second more painful. Both lenses are honest.
Every reading on this page is anchored in real market data. The phase labels are computed from the market’s own full history. The cross-market drivers are measured with block-bootstrap confidence intervals so only statistically distinguishable shifts are surfaced. If a number is on this page, you can trace it back to the data that produced it.