Symmetry in Markets
What physical chemistry's mirror planes and rotation axes can, and can't, tell you about a price chart.
Symmetry in physical chemistry
Chemists sort molecules by their symmetry elements: the moves you can make that leave the molecule looking exactly the same. That classification predicts real behavior: which vibrations show up in an infrared spectrum, whether a molecule is polar, which orbitals can bond. There are only a handful of elements:
- E, identity: do nothing. Every object has it.
- Cn, rotation axis: turn by 360°/n and nothing changes. C2 is a half turn, C3 a third of a turn.
- σ, mirror plane: reflect through a plane. A plane containing the main axis is σv; one perpendicular to it is σh.
- i, inversion center: send every point straight through the center to the opposite side.
The full set of elements a molecule has is its point group.
Bigger, more symmetric molecules have bigger point groups. Counting every move that leaves the molecule unchanged:
Water: C2v 4 operations
A half turn about the dashed axis swaps the hydrogens (C2), plus two mirror planes: the page itself, and the vertical plane through the axis.
Ammonia: C3v 6 operations
Seen from above, a third of a turn moves each hydrogen to the next spot (C3), and there are three mirror planes, one through each hydrogen.
Methane: Td 24 operations
A perfect tetrahedron (wedge = toward you, dashes = away). Four C3 axes, one through each hydrogen, three C2 axes and six mirror planes.
Benzene: D6h 24 operations
A six-fold axis through the center (C6), six two-fold axes in the ring's plane, seven mirror planes (two drawn) and an inversion center.
When symmetry is missing: chirality
Put four different groups around one carbon atom and every mirror plane, inversion center and rotation that could flip it vanishes. The molecule and its mirror image are built from exactly the same atoms, bonded in exactly the same order, yet no rotation lines one up with the other, just as your left hand won't fit a right-handed glove. Chemists call such molecules chiral (Greek cheir, hand); their point group has no reflections at all.
Same carbon, same four groups (colors), reflected. Turn either one any way you like: the blue group can be toward you only when the green one points the wrong way.
Your body is built from chiral molecules too, so it can tell the two hands apart. That's why mirror-image drugs can act completely differently:
| Molecule | One hand | The other hand |
|---|---|---|
| Citalopram (antidepressant) | The S form blocks serotonin reuptake: it does the work. | The R form does very little, and may even blunt the S form. The pure S form is sold on its own as escitalopram. |
| Albuterol (asthma inhaler) | The R form opens the airways. | The S form is essentially inactive at the same receptor. The pure R form is sold as levalbuterol. |
| Carvone | One form smells like spearmint. | Its mirror image smells like caraway (rye bread). Your nose is chiral. |
| Thalidomide (the cautionary tale) | One form was linked to the sedative effect. | The other to severe birth defects. The body converts each into the other within hours, so even a pure single-hand pill would not have been safe. |
Markets are chiral too. Run a price chart backwards and it looks different: prices tend to grind up and drop fast (traders say "stairs up, elevator down"), and volatility jumps when prices fall. That built-in handedness is likely one reason clean mirror symmetries turn up less often in SPY than in a shuffled copy of its own history (see the chance levels below).
Why a chart only gets three
A molecule sits in space where every direction is equivalent, so it can have 3-fold or 6-fold axes. A price chart can't: its two axes are time and price, and you can't turn one into the other. A quarter turn would make time run up the page. So of all the symmetry elements, only these survive:
| Element | On a chart | Familiar pattern |
|---|---|---|
| Mirror plane σv | The path after a date mirrors the path before it. | V-bottoms, head-and-shoulders (mirrored about the head), rounded tops |
| 2-fold rotation C2 | Turn the chart 180° about a point and it lands on itself. | S-shaped turns: a slide, a pivot, then an inverted copy of the slide |
| Glide reflection σh + shift | A later stretch is an upside-down copy of an earlier one. | A rally followed by a matching decline ("measured moves") |
The glide is the horizontal mirror's survivor: flipping price upside down is only interesting if you also slide along in time, the same combination crystallographers call a glide plane. The most symmetric a chart can ever be is C2v, the same point group as a water molecule.
Found in SPY
Our symmetry finder scanned the last ten years of SPY daily closes for each element, at pattern widths from three weeks to about a year. Here are the strongest of each. The orange dashed line is the first half of the pattern mirrored, rotated or glided onto the second half: where it hugs the real price, the symmetry holds.
Match 75% · chance level 83% below chance
Match 88% · chance level 89% about chance
Match 88% · chance level 90% about chance
Read the chance level first
Scan thousands of windows and something will always look symmetric. So next to every match we show the chance level: the best score the same search finds on SPY's own history after its daily moves have been shuffled in 20-day blocks. That keeps SPY's real crashes, calm spells and fat tails but destroys any longer pattern. A match near the chance level is what shuffled noise produces anyway; only one well above it hints at real structure. Markets are close to random, so even a decade's best symmetry often sits right at chance. That's worth knowing before anyone sells you a pattern.
Physics vs. markets: the scorecard
Symmetry is only one idea physical chemistry offers. Here's how the main ones map onto markets, and whether they actually predict anything. Where it says tested here, we ran it ourselves: walking forward through ten years of daily prices for SPY, QQQ and about 100 stocks (two random samples), using only data available on each day, and comparing against the simplest baseline of always guessing "up".
| P-chem idea | Market version | Does it predict? | Status |
|---|---|---|---|
| Symmetry elements (mirror, C2, glide) | Finish a half-formed symmetric pattern | No Called direction right 50% of the time, vs. 59% for always guessing up. A faint extra signal (mostly glides) fades once 2020 is left out. | Tested here |
| Wilder's Adam Theory (a C2 rotation about today) | Repeat the last move: momentum | No Lost to always-up at 10 to 60 days. At 40 days moves tended to reverse instead. | Tested here |
| Lunar cycles (Delta-style timing) | Turning points cluster by moon phase | No 18 tests on SPY, QQQ and four stocks since 2000; none significant. Turns landed near a new or full moon 34–43% of the time vs. 41% by chance. | Tested here |
| Temperature (statistical mechanics) | Volatility: how violently price jiggles | Yes, strongly Calm spells stay calm and storms stay stormy. Recent-volatility forecasts beat the long-run average by about 25% (t ≈ 8), and their 90% expected-range bands held the actual move 90% of the time. Members: Market Temperature | Tested here |
| Langevin dynamics (Ornstein–Uhlenbeck: a spring in a heat bath) | Mean reversion of yield spreads and VIX | Not at 20 days Treasury spreads did worse than "stays where it is": they revert too slowly (half-lives of 6–10 months) and their equilibrium keeps moving. VIX reverted (half-life about 18 days) but only slightly beat no-change. | Tested here |
| Entropy | Orderly vs. random stretches of price | Unproven Do ordered stretches come before big moves? | Planned |
| Phase transitions (systems near a tipping point) | Bubble and crash hazard (log-periodic power laws) | Controversial Has flagged some famous bubbles, with many false alarms. | Planned |
| Diffusion (Fokker–Planck, the heat equation) | Option pricing: Black–Scholes is the heat equation | Risk, not direction Prices the range of likely moves. | Established research |
| Scale invariance (fractals) | Hurst exponent: does a series trend or revert at a given scale? | Weak Measurable, but small. | Planned |
The pattern so far: physics describes the size of market moves well and their direction barely at all. Results as of October 2026; rows update as we run each test.
Members: the Symmetry Finder and Market Temperature
Symmetry Finder
Run the finder on any ticker, daily, weekly or hourly: the strongest mirror planes, rotation points and glides, each scored against that ticker's own chance level, plus any symmetry still unfolding today and where its mirror image would go.
Market Temperature
The one physics idea that passed our test: volatility as temperature. For any ticker, how hot it's running against its own history, and 90% expected ranges for the next 5 and 20 trading days, with a track record showing those ranges really held about 90% of the time.
Today's volatility ranking
501 stocks, each ranked against its own last five years. The percentile is how much of that history was calmer than today.
| Hottest | |
|---|---|
| PTC | 102.0% (100th) |
| TMUS | 60.8% (100th) |
| CTVA | 63.1% (100th) |
| CCI | 59.0% (100th) |
| FICO | 119.5% (99th) |
| Coldest | |
|---|---|
| LYV | 16.8% (0th) |
| KIM | 13.0% (0th) |
| REG | 11.7% (0th) |
| BEN | 19.1% (0th) |
| LVS | 22.0% (0th) |
Symmetry describes the shape of past prices. It is not a forecast and not investment advice. Matching uses log prices, so a 10% move counts the same at any price level; a match score is 1 minus the typical gap between the two halves, as a share of the pattern's size.