Phase Space Research · Quantfund

Signals

Conclusions from the quantfund research programme · updated 2026-08-19 22:11 UTC · build 20c1ed0e

The finding, stated once

This programme set out to select which delivered options alerts to trade. That question is answered: none. The median alerted contract has already repriced +4.26% before the first tradable quote; post-alert continuation is +0.00% in every year 2023–2026; and every route to a liquidity-taking strategy — alert selection, day selection, exits, dips, upstream flow, volatility entry, stock execution, overnight and multi-day carry, the next-day gap — has been measured and closed, most of them twice.

What survives is narrower and genuinely real: this data forecasts short-horizon information intensity — how much a name is about to move, never which way — confirmed by three independent instruments. The programme has pivoted accordingly, from trade alerts to volatility/event forecasting.

An earlier edition of this page claimed tradable post-alert edges. Those claims rested on a measurement defect and were retracted in full; the corrections history at the bottom preserves that record deliberately.

The final question — whether the alerts were informative about a mechanism rather than about future option returns — was settled the way it should have been from the start: seven falsifiable gates and a kill rule, written down and committed before the analysis ran. It failed 6 of 7 on 899,444 snapshots, and the sealed holdout was never spent. The programme's durable output is that procedure, not a strategy: kill the hypothesis, keep the laboratory.

That procedure has now been run twice. A second dataset — cross-company information propagation, a different economic species of claim on a different data-generating process — was pre-registered with a hashed exposure graph, a profitability hurdle instead of a significance test, and eight gates. It failed six of them, and the sealed holdout is still unspent. Both programmes are recorded here in full, including the corrections found while running them.

What is established

3 studies · 165k+ prospective snapshots LIVEThe one thing this data does: forecast information INTENSITY

magnitude predictable three independent ways · direction predictable zero ways

Three instruments, three samples, one answer. (1) Alerted names move 0.105% → 1.378% more than the market in absolute terms from one minute to the next close, with signed excess ≈ 0 at every horizon (n=12,325). (2) Extreme option flow of EITHER sign precedes larger stock movement — the flow-decile curve is U-shaped (most-bearish decile +3.08 bps ≈ most-bullish +2.97 vs ~+1.5 in the middle; 106,008 alert-free snapshots). (3) Implied-vol underreaction predicts subsequent IV correction monotonically across all ten deciles (59,445 alert-free snapshots), and delta-hedged straddle P&L turns gross-positive in exactly the predicted state at exactly the predicted horizons.

So what: This is a short-horizon volatility / information-event forecasting capability, and it is where the programme now goes: the expanding tick archive is its training data, the sealed 60-session holdout its one-shot validation.

Caveat: Forecasting skill, not trading profit — the sections below document why no liquidity-taking expression of it survives its own transaction costs.

14,546 trades LIVEUpside is concentrated in a handful of trades

top 5% of trades hold 42.9% of all upside · top 1% hold 17.0%

Top 10% hold 58.8%, top 20% hold 76.7% (n=14,546, corrected excursion set — concentration is MORE extreme than the pre-correction edition reported).

So what: Any rule that caps winners destroys the payoff; this is the arithmetic reason scaling out and trailing stops fail here.

Caveat: Computed on maximum favourable excursion — what the contract OFFERED, not what a position captured.

894 sessions LIVEDays cluster — but do not persist

68.5% of sessions lose money · top 20% of days carry 94.5% of gains · lag-1 rho +0.023

Top 5% of days carry 56.3% (894 sessions, executable marking). But the previously published day-to-day persistence of +0.397 collapses to +0.023 on corrected data — ask/bid +0.023, mid/mid +0.014, and NEGATIVE on opportunity marking. Clustering is variance; persistence would have been predictability, and only predictability was actionable.

So what: Good days are rare and carry everything, and nothing detectable in advance separates them from the rest. There is no day-level state to detect.

Caveat: The +0.397 was largely the horizon defect itself: a return series that partly samples the following session inherits its level, manufacturing dependence.

What is closed, and why

20,018 trades · 893 sessions · 6 studies DEADTrading the delivered alert, in any form

median repricing +4.26% BEFORE the first tradable quote · 60s continuation +0.00% in every year 2023–2026

By the time a subscriber can transact, the move has happened: the median contract reprices +4.26% between the last pre-alert and first tradable post-alert quote, and continuation from that quote is +0.00% every year. Median alert returns −18.4% mid-to-mid; the entire +3.9% mean sits in the top 0.1% — 19 contracts. Against matched non-alerted contracts on the same name at the same instant, median excess is $0–1.50/contract, every interval spanning zero (underpowered: 94 of 425 matched). Buying the post-alert dip: all 30 cells negative, and the drawdown filter is a SPREAD filter (10.5% → 13.3% monotone in depth). The one conditional rebound that cleared the spread (+1.19% in liquid IV-driven dips) is fully reproduced by a random-anchor placebo — microstructure, not information.

So what: The alert is a lagging notification of completed price discovery. Every conditioning refinement either found nothing or found a generic property of options masquerading as alert information.

Caveat: All measurements are UPPER bounds — ts is message time; real delivery is later.

31 re-derived analyses DEADDay selection and per-trade selection engineering

day models AUROC 0.470–0.552 · all 80 gate-grid cells negative · Kelly f* = 0.00 for all 22 exit rules

With no day-level persistence there is nothing to detect, and nothing does: two-book day models are coin flips in and out of holdout; the three-state policy loses less only by trading less of a negative book (−5.5%/day vs −10.8%); 19 market-open indicators all span zero; the day's first alerts do not predict its rest. Per-trade: IV rank grades the day not the trade; the 0–100 score ranks (AUROC 0.626) without raising expected return; waiting for confirmation buys the move after it happens; dealer gamma points the wrong way; every stop, ladder, scale-out and profit target sizes to zero.

So what: Selection and exit engineering cannot rescue a book whose per-trade expectation is negative and whose upside is 19 contracts.

Caveat: Figures from the corrected re-derivation (2026-08-06); full per-test cards in the repository.

5 studies · 122k prospective snapshots · 372 ticker-days DEADUpstream of the alert: the mechanism is real and the venue is fatal

pre-alert ramp alert-specific · final 2 minutes are pure IV expansion · best edge = 1% of its own spread

The pre-alert repricing is a minutes-long ramp, finished before the alert fires — and it IS alert-specific: against non-alert threshold crossings matched on size, spread, realised vol and quote activity, alerted moves are back-loaded (paired median −0.42 at −12m, n=4,807). Greek decomposition (n=7,356): the stock leads for ~20 minutes while IV falls; in the final two minutes the stock stops and vega takes over (share 0.68). The tradable implication — buy delta-neutral volatility during the underreaction — is real and gross-positive in the predicted state (+$0.22 at 2m)… against a $17.50 straddle round-trip: the edge is 1% of its own transaction cost. Executing in the stock instead: signed flow carries nothing (U-shaped deciles, D10−D1 spans zero on 106k snapshots), and gating equity breakouts on unsigned flow magnitude adds nothing on 122,560 snapshots (every difference interval spans zero; directionally harmful at 1m).

So what: The forecasting edge exists AT MID, and mid belongs to the liquidity provider. A taker cannot reach it at any level of model quality. The liquidity-taking search on this archive is closed on arithmetic, not on model failure.

Caveat: Scanner reconstruction and predict-the-alert are closed by the same arithmetic: discovery finishes 2–10 seconds before the alert, so a PERFECT reconstruction buys seconds in which the price has already moved.

12,486 holds · 11,666 matched pairs · 2 studies DEADPast the bell: overnight holds, multi-day holds, and the next-day gap

overnight median −22.7% · median −40% by session 3 · next-open differential fails its matched placebo

Carrying a triggered contract from the day-1 closing bid into the next session loses 22.7% on median (n=12,486, win 35.8%), and the bleed compounds: the median surviving position is down ~40% by the third session. Winsorised mean ≈ 0 (−1.3%) — the fat right tail pays for the bleed almost exactly, so there is no edge in either direction: harvesting the decay means shorting the tail. Single-name contracts are destroyed overnight (median −54%); SPX holds are a median coin flip. On the STOCK side, the alert-day next-open gap (+0.166%) was the one open quantity in the record; against controls matched on realised vol and volume it collapses to +0.066%, CI [−0.028, +0.165] — non-alert days that look like alert days gap the same way. Closed at both the option and the equity layer.

So what: There is nothing downstream of the session anywhere: intraday, overnight, multi-day, option or underlying. The 2026-07-31 edition's claim that overnight holds are WORST after good days is retracted — the −55.9% asymmetry was a correction artefact; re-derived it is −3.4% with an interval spanning zero.

Caveat: Overnight is close-to-next-close (EOD windows), so the next day's intraday move is inside the leg; a zero next-session bid counts as −100%, not missing data — dropping those marks alone flips the mean positive. Holdout unspent throughout.

899,444 snapshots · 319 tickers · 2,735 ticker-days DEADThe dealer-hedging mechanism — pre-registered, then killed by its own rule

6 of 7 gates fail · 899,444 prospective snapshots · holdout unspent

The last surviving hypothesis was a mechanism, not a strategy: customer option flow forces dealers to hedge, and that hedging pushes the underlying. It is compatible with every trading rejection on this page — an alert can be informative about a contemporaneous mechanism without forecasting future option returns. Seven falsifiable gates were frozen in writing before the analysis ran (sign, dose-response, volatility/volume controls, breadth, mechanism location, stability, materiality), with a stated kill rule: anything less than 7/7 ends the programme and leaves the sealed holdout unspent.

Result on 899,444 snapshots across 319 tickers: signed pressure moves the beta-adjusted underlying by −0.85 bps at 30 minutes, CI [−2.82, +1.05] — wrong sign, spanning zero, and an order of magnitude below the 6 bps the gate required. Every horizon from 5 minutes to the close is negative and spans zero. Only 49.5% of tickers show a positive contrast, and the sign flips between chronological halves.

The dose-response ran significantly backwards — all nine decile steps the wrong way — and that traced to how the pressure variable was read rather than to the market: dividing option flow by underlying share volume makes the measure inversely proxy the stock's own liquidity and volatility. The high-pressure decile is 22% less volatile than the low. A quantity named for hedging demand was substantially measuring how quiet the underlying is. The resulting standing check is narrow and cheap: when a denominator is economically related to the outcome, test the ratio against its own components before believing it.

So what: The mechanism is dead on this dataset by a rule written before the data was seen, and the holdout was never touched. That is the point: the programme's output is not a strategy, it is a process that reliably prevents us from believing a false one.

Caveat: One gate (mechanism location) passed and is NOT claimed as partial support. It is a conditional test — it asks whether the effect concentrates where dealer hedging should bite hardest, which only discriminates a mechanism from a volatility story if an effect exists. The contrast it compares is itself ≈ 0, so 'stronger' there means 'less negative', and heterogeneity in a null is not evidence of the mechanism that would have produced it. Excluding index ETFs, whose abnormal return against SPY is definitionally ≈ 0, reproduces the identical 6-of-7 failure.

399 events · 705 legs · 49 edges · 25 reporters DEADDataset #2: cross-company information propagation — killed by its own rule

6 of 8 gates fail · −46.87 bps, wrong sign · pre-event drift is 3× the effect · holdout unspent

A different dataset and a different economic species of claim: when company A reports, does its already-public disclosure predict a subsequent move in an economically linked company B? The friction would be attention and mapping, not latency — and unlike the options work, the security generating the information is not the security being traded. A 49-edge exposure graph (supplier, capex, cloud-consumption, peer, retail-brand links) was hand-built, given per-edge vintage dates so no relationship is used before it was publicly establishable, and hashed before any return was joined. Entry is 15 minutes after the open, conceding B's own gap. The test was profitability against a pre-declared 10 bps hurdle, not significance.

Result on 399 earnings events: −46.87 bps, CI [−90.14, −4.89] — the wrong sign, and the whole interval below the hurdle. The relationship-strength dose runs backwards (strongest links most negative), and the interaction term is negative where the mechanism requires positive.

The decisive gate was the one that measures nothing about the event. In the five sessions before A reports, the linked firms already move −151 bps against the predicted direction — three times the size of the post-event result. Sectors that sell off into a print tend to clear a lowered bar and gap up, so the shock's sign selects events whose linked names were already falling. What looked like propagation is the continuation of a trend that predates the cause.

So what: The failure is more informative than Dataset #1's. This did not measure nothing — it measured something real, sizeable and pre-existing that the design would have mistaken for propagation. The pre-event drift control existed to catch exactly that, and it did. Any future event study here carries it from the start.

Caveat: One correction is disclosed rather than buried: the sign placebo was first implemented as a single permuted draw with a bootstrapped interval, not a permutation null. Fixed to 500 permutations — found after results were visible, and it flipped that gate from fail to pass. The verdict is unchanged; the profitability gate is dispositive alone. The matched placebo graph behaved correctly throughout (γ ≈ 0, spread CI spanning zero), so the linked graph does carry real structure — pointing the wrong way.

Standing method

Measurement. Entry from the first TRADABLE quote, never a pre-alert quote. Fixed-dollar P&L, never percentage (percentage optimisation mechanically favours near-zero option prices). Medians beside means — the mean here describes 19 contracts. Bootstraps clustered by session or ticker-day; the statistic of interest gets its own interval, never inferred from its components' intervals.

Placebo before belief. Any rule touching the alert only through its anchor time is re-run from a random anchor; any backward-looking ramp is tested against endpoint-matched non-alert controls, because first-passage geometry manufactures ramps. Placebos killed the two most promising results this programme produced.

The liquidity efficiency ratio. LER = |gross edge| / round-trip cost, computed only once the edge's own interval excludes zero. Below 0.1: kill. 0.1–0.5: interesting, economically dead. 0.5–1: needs exceptional execution. Above 1: candidate. The vega result scored 0.01; a 40-ticker-day subsample once scored 1.8 on noise, which is why the interval comes first.

Extreme conditioning selects wide spreads. Three independent conditioning variables (drawdown depth, IV-normalisation label, IV residual) each selected the widest-spread contracts. Every conditioning variable must now report its subset's entry spread beside its effect; an effect smaller than the excess spread it selects is dead regardless of significance.

Timing. Three defects shared one signature — a bucketed quantity treated as instantaneous (date-range payloads, pre-alert anchoring, minute-binned flow against bar-start stamps). Every time-indexed feature must state which interval its bucket spans and on which side of the signal instant it falls.

Pre-registration. The final hypothesis was tested under a specification committed to version control before the analysis ran: outcome variable, primary horizon, seven numeric gates and an explicit kill rule, plus a register of adverse priors. Amendments are permitted only for facts about the data that make a gate untestable rather than failing — one was needed, it is disclosed in the card, and it moved against the hypothesis' interest. Every peek at partial results, including a 1%-sample smoke test, is disclosed rather than hidden.

Holdout. The original 4,016-trade seal was spent 2026-08-04. A fresh 60-session seal (2026-05-04 → 2026-07-29, sha256 29fadd63…) remains unopened through everything on this page — including the pre-registered mechanism test, which failed its gates and therefore never earned its one look. It is spent once, on a frozen specification, or not at all.

Corrections history

Kept prominently and permanently — the record of finding and fixing these is the page's strongest credential.

2026-08-05, the horizon defect. Quote paths fetched as date ranges let ms-of-day filters walk across session boundaries, corrupting 72.4% of alerts. Separately, entry was anchored on the last PRE-alert quote, booking the alert's own +4.26% repricing as profit. All 31 published analyses were re-derived on corrected data; the apparent +4.4% post-alert edge became +0.00%.

2026-08-06. The published day-persistence figure (+0.397) collapsed to +0.023 under correction — it was largely the defect itself. Every claim resting on it retracted; the page's card prose was stripped of all pre-correction numbers.

2026-08-08. The first event-time analysis compared two arms normalised on different baselines and lacked a placebo; both corrected, reversing its liquidity conclusion. A reported paired-difference statistic mixed medians and means, overstating the effect ~60%; restated on the paired median.

2026-08-10. A results table shipped with a doubled percent sign; and sortable headers were added to this page's tables after repeated omissions — every column now sorts, numerically where numeric.

2026-08-10, overnight re-derivation. The 2026-07-31 overnight-hold card predated the horizon correction, was stripped in the retraction, and had never been re-derived. Re-run on corrected sources: the headline (overnight holding is negative on median) stands, −21.4% → −22.7%; the asymmetry claim ("worst after a good day", −55.9%) is retracted as a correction artefact — re-derived, −3.4% with an interval spanning zero. A second defect fixed in the re-run: zero-bid next-session marks are total losses, and dropping them had flipped the winsorised mean from −1.3% to +10.7%. The equity next-open differential — the record's one open quantity — was then closed by its own pre-registered test: it fails the vol/volume-matched placebo. The sealed holdout was not touched by either result.

Earnings Move Premium

Is the volatility priced into an options straddle before an earnings announcement set above, or below, the move that actually arrives? 7,746 events, measured on the front expiry with implied and realised covering the identical horizon.

Events7,746front expiry, 2020–2026
Implied move8.14%median 7.22%
Realised move7.97%median 5.74%
Mean edge-0.172%95% CI -0.378 to +0.015
Median edge-1.056%41.6% of events favoured the buyer
Round-trip cost0.682%LER 0.25 — below 1.0

Where the research stands

Seven gates were frozen before the differenced statistic existed. A result must clear all of them; failing on cost alone returns ECONOMICALLY DEAD rather than inconclusive.

G1
Sign
Implied priced above realised, as the friction predicts.
Directionally consistent
G2
Differenced, not raw
Earnings minus control. A raw edge that vanishes on differencing is a property of the names, not of announcements.
Blocked — controls harvesting
G3
Cost
|edge| ÷ round-trip cost above 1.0, priced at ask on entry and bid on exit.
LER 0.25 — likely killer
G4
Clustered inference
95% CI excludes zero, bootstrapped over calendar months, not rows.
CI spans zero
G5
Temporal stability
Sign holds in every era.
Mean flips in 2023–2024
G6
Breadth
Holds across subsectors, not carried by the top few names by event count.
Not yet evaluated
G7
Method falsification
Jump variance on control windows must be ≈ 0. There is no event there. This tests the method, not the hypothesis.
Blocked — needs back expiry

Implied against realised, one point per announcement

The diagonal is break-even: a straddle bought at the last close before the report and held to expiry. Above the line the buyer won. The grey band is the round-trip bid–ask cost, so anything inside it is unreachable in either direction — which is most of the distribution.

7,746 events plotted
Seller won — realised below implied Buyer won — realised above implied Inside round-trip cost

Axes clipped at 30%. 196 points fall outside and are drawn at the edge. The full per-name figures are in the table below and do not depend on this plot.

What the crush looks like, and what it looks like when nothing happens

NVDA, at-the-money implied volatility, re-selected to the nearest strike every session — tracking one strike through a gap puts you deep in-the-money where implied vol is barely identified. Left: the February 2024 announcement, where implied volatility falls from 0.851 to 0.492 overnight. Right: a matched control window with no scheduled event at its centre, which stays inside 0.21–0.42 throughout.

Earnings window — 2024-02-21 Control window — 2023-04-18

Stability across eras

Gate G5 requires the sign to hold in every era. On the mean it does not: 2023–2024 flips positive. On the median it holds throughout — which is the distinction the whole page turns on, because a strategy earns the mean.

EraEvents Mean edge %Median edge %
2020-20222783-0.394-1.186
2023-202429000.164-0.780
2025-20262063-0.343-1.267

The cross-section

Per ticker, names with at least four measured announcements. Any real edge would live here, in the spread between names, rather than in the pooled average.

473 names

By sector

Sector Events Mean edge % Mean |edge| % Buyer won %
Technology17400.2186.11542.5
Communication Services3780.0425.44846.3
Consumer Cyclical1200-0.3275.30939.4
Industrials961-0.3334.92340.7
Healthcare799-0.3314.65443.6
Basic Materials4210.3434.57143.5
Real Estate231-0.2084.20540.7
Energy322-0.5574.02137.6
Consumer Defensive543-0.3593.88740.7
Financial Services961-0.4633.78140.8
Utilities1820.3253.37645.1

A sector needs at least three names in the period to appear. Mean edge is the direction claim; mean |edge| is the dispersion claim, and on inspection the second is the one that moves.

By name

Ticker Sector Events Implied % Realised % Mean edge % Median edge % Cost % LER
AALIndustrials246.384.81-1.577-2.1610.3214.91
AAPLTechnology244.213.31-0.896-0.9760.1386.51
AABasic Materials247.066.89-0.175-1.2010.5520.32
ADBETechnology246.817.010.200-0.3390.4030.50
AKAMTechnology248.8410.201.3641.3640.8561.59
ALKIndustrials249.179.350.182-0.0240.8550.21
ALLYFinancial Services246.835.29-1.543-2.5940.5702.70
AMDTechnology248.218.00-0.212-2.7070.1551.37
AXPFinancial Services243.974.040.063-0.5640.3420.19
BACFinancial Services243.943.34-0.604-1.2360.1543.92
BAIndustrials244.914.77-0.141-0.5880.3190.44
BEIndustrials2415.6813.20-2.4820.0761.4791.68
BKNGConsumer Cyclical245.464.56-0.897-1.2690.7381.22
CATIndustrials244.794.28-0.511-0.4250.4091.25
CLFBasic Materials248.6810.321.6440.6360.3994.12
CLXConsumer Defensive245.635.770.143-0.0180.5460.26
CMGConsumer Cyclical247.077.090.021-1.5990.6300.03
COFFinancial Services245.025.340.3240.0970.4310.75
COSTConsumer Defensive243.222.60-0.619-0.8680.2003.09
CROXConsumer Cyclical2412.1315.313.1773.2620.9023.52
CRWDTechnology249.727.84-1.877-1.9060.4124.55
CSGPReal Estate249.6910.000.3070.8030.7760.40
CVNAConsumer Cyclical2414.6416.712.063-0.1381.0501.96
CFinancial Services243.643.760.121-0.6610.1880.64
DALIndustrials245.694.48-1.213-1.4560.3133.87
DDOGTechnology2410.9610.84-0.117-4.0920.7360.16
DISCommunication Services248.886.01-2.872-1.6570.27210.55
DKNGConsumer Cyclical2410.169.69-0.471-2.1570.5620.84
DOCUTechnology2411.5812.771.195-2.5560.5602.13
FConsumer Cyclical246.416.17-0.238-1.3250.2900.82
GEIndustrials245.794.84-0.951-1.5500.3452.76
GOOGCommunication Services245.706.200.4960.5670.1932.58
GSFinancial Services243.873.85-0.022-0.5300.2790.08
IBMTechnology245.486.280.8011.3720.2682.99
INTCTechnology247.768.791.033-0.0310.2274.55
JPMFinancial Services243.312.72-0.596-1.5030.1863.20
KOConsumer Defensive242.772.26-0.507-0.9110.1553.27
LRCXTechnology246.386.11-0.270-1.9200.5040.54
MCDConsumer Cyclical243.302.71-0.587-0.9860.2562.30
MSFTTechnology244.914.82-0.085-1.0510.1570.54
MUTechnology248.156.70-1.458-2.5160.1987.37
NEMBasic Materials244.534.750.216-0.5300.2470.87
NFLXCommunication Services248.4410.231.7890.8360.2188.19
NKEConsumer Cyclical247.168.050.889-0.2160.2593.43
NOWTechnology247.285.76-1.519-3.2530.4323.52
ORCLTechnology247.528.821.3032.8700.3054.27
PEPConsumer Defensive242.963.530.5710.8010.2622.18
PFEHealthcare244.453.34-1.111-1.4770.2654.20
PLTRTechnology2413.1517.804.6531.2550.26917.27
PYPLFinancial Services248.548.800.258-0.4830.3030.85
QCOMTechnology246.747.110.3700.0070.3371.10
SBUXConsumer Cyclical245.724.77-0.946-2.0680.2433.89
SHOPTechnology2410.5714.824.2542.8610.5547.68
SPOTCommunication Services249.609.48-0.122-0.2200.7160.17
TSMTechnology243.622.95-0.675-1.3280.1903.55
TXNTechnology245.385.570.1910.0510.3990.48
UBERTechnology248.657.51-1.145-3.0560.3333.44
UNHHealthcare243.996.062.0680.7820.3945.25
VZCommunication Services243.653.750.101-0.4260.2800.36
VFinancial Services243.683.13-0.549-1.2910.2102.61
WFCFinancial Services244.284.17-0.1030.2970.2060.50
ABNBConsumer Cyclical238.566.84-1.717-3.9480.2846.05
ALBBasic Materials237.926.33-1.591-1.9130.6342.51
ALGNHealthcare2310.5712.381.808-1.8981.5121.20
ALLFinancial Services235.495.39-0.1010.0700.5080.20
AMATTechnology237.764.10-3.660-1.5300.32311.32
AMGNHealthcare234.074.650.580-0.9220.4001.45
AMZNConsumer Cyclical236.466.16-0.302-1.0710.1521.99
AONFinancial Services235.785.65-0.129-0.5510.7840.16
BABAConsumer Cyclical236.115.39-0.716-2.4260.3202.24
BBYConsumer Cyclical237.467.45-0.016-1.8510.3880.04
BILICommunication Services2311.1012.090.992-0.8930.7501.32
BILLTechnology2315.4515.840.390-1.0971.2280.32
BLKFinancial Services233.473.17-0.306-0.8700.4880.63
BMYHealthcare233.723.830.110-0.2180.4310.26
CAGConsumer Defensive234.593.25-1.334-1.9870.5862.28
CBFinancial Services235.265.370.112-1.5080.3730.30
CCIReal Estate235.633.55-2.078-3.2260.5353.88
CCLConsumer Cyclical237.807.54-0.254-2.7270.3690.69
CRMTechnology237.347.910.569-2.3170.2312.47
CSCOTechnology235.525.880.354-0.5440.2101.68
CVSHealthcare235.265.270.0060.2800.3620.02
DASHConsumer Cyclical2310.678.95-1.720-4.2260.6292.73
DELLTechnology238.8710.801.922-2.5570.5633.41
DGConsumer Defensive236.977.440.465-2.3120.5950.78
DKSConsumer Cyclical239.228.42-0.801-1.8360.9640.83
DVNEnergy235.786.260.485-0.7940.3381.44
GOOGLCommunication Services235.705.990.2850.7770.1931.48
HDConsumer Cyclical234.193.81-0.384-0.5520.2841.35
ISRGHealthcare235.846.620.783-1.0890.4731.65
JNJHealthcare232.742.01-0.728-1.0300.2233.26
MAFinancial Services233.392.98-0.407-0.4700.3241.26
MRKHealthcare233.694.210.5190.3430.3331.56
MRVLTechnology239.1810.591.412-0.1120.4513.13
MSFinancial Services233.803.890.088-0.0250.2560.34
PANWTechnology238.188.400.217-1.9360.5470.40
RTXIndustrials234.084.790.710-0.5190.3352.12
TMUSCommunication Services234.554.630.0800.1070.5170.15
TSLAConsumer Cyclical237.6810.082.4051.9260.12519.31
TCommunication Services234.274.27-0.004-0.8490.2790.02
WMTConsumer Defensive234.434.39-0.045-0.9830.1810.25
XOMEnergy232.732.45-0.278-0.8140.2451.14
AAPConsumer Cyclical2210.4911.961.464-1.2870.9191.59
ABBVHealthcare223.614.130.5210.3950.3661.42
ADMConsumer Defensive224.774.960.1890.2660.5630.34
ADSKTechnology226.295.97-0.323-0.8120.4670.69
AEMBasic Materials227.266.91-0.345-1.2030.5080.68
AGNCReal Estate223.954.340.391-0.6250.4540.86
AMKRTechnology2212.0712.360.294-1.7821.1930.25
AMTReal Estate225.556.170.625-0.8710.4401.42
ANETTechnology229.578.21-1.363-3.5530.5882.32
APAEnergy226.215.37-0.839-1.7190.4641.81
ASMLTechnology225.616.851.2440.1780.6981.78
AVGOTechnology225.816.420.603-2.0760.3611.67
AXONIndustrials2211.6713.041.370-0.6491.0701.28
AHealthcare226.826.74-0.078-0.7230.5680.14
BIDUCommunication Services226.645.22-1.414-1.7430.3813.71
BKREnergy228.248.440.194-1.4520.7650.25
BOXTechnology2210.058.91-1.138-1.2020.8681.31
BPEnergy224.364.870.5030.1310.3031.66
BXPReal Estate227.235.55-1.683-2.9110.7792.16
BXFinancial Services224.125.201.0810.3110.4512.40
CDNSTechnology227.956.89-1.067-1.1210.7031.52
CELHConsumer Defensive2214.5414.44-0.102-1.1691.1300.09
CHRWIndustrials227.827.72-0.101-0.7570.6660.15
CHWYConsumer Cyclical2211.9010.31-1.582-1.5960.7102.23
CVXEnergy222.472.780.3100.2540.2971.04
DEIndustrials224.714.43-0.285-0.6020.4900.58
DHIConsumer Cyclical225.515.05-0.452-1.0050.6860.66
DLTRConsumer Defensive228.678.57-0.098-0.1930.6080.16
DPZConsumer Cyclical226.685.97-0.710-1.6230.8760.81
DVAHealthcare228.5111.743.2291.4200.7234.46
GILDHealthcare224.164.160.005-0.3110.2820.02
KLACTechnology226.336.05-0.277-0.5780.5200.53
LINBasic Materials224.864.54-0.319-0.3020.5810.55
LLYHealthcare225.306.741.441-0.4260.5632.56
MELIConsumer Cyclical228.168.990.8260.3561.0140.81
MSTRTechnology228.3810.882.493-1.2290.8402.97
NEEUtilities224.936.641.7081.6810.4174.09
NVDATechnology227.616.52-1.090-1.9960.1328.27
PDDConsumer Cyclical2210.3614.774.4094.9530.7845.62
PGConsumer Defensive222.972.01-0.964-1.2860.2933.29
PMConsumer Defensive223.863.990.129-0.8870.5480.24
SNOWTechnology2210.8013.833.0330.3400.3329.12
TJXConsumer Cyclical224.643.80-0.839-1.5190.4182.01
ACNTechnology215.065.300.2400.4310.4540.53
AEPUtilities214.504.540.039-0.4200.4420.09
AREnergy217.957.31-0.641-1.7870.9410.68
BJConsumer Defensive218.847.51-1.328-1.3470.8301.60
CALMConsumer Defensive218.246.05-2.186-4.0610.9402.33
CARRIndustrials217.507.36-0.137-1.8180.7110.19
CARIndustrials2112.959.56-3.398-5.4741.1842.87
CBREReal Estate217.328.080.767-0.7860.8330.92
CBRLConsumer Cyclical219.547.73-1.804-4.7441.0591.70
CCJEnergy216.365.72-0.639-1.4840.4891.31
COINFinancial Services2110.417.50-2.908-5.2950.4995.83
COPEnergy213.483.24-0.234-1.0470.3830.61
CPBConsumer Defensive214.854.890.040-1.3470.5880.07
CPNGConsumer Cyclical2110.038.37-1.659-2.0410.6392.60
CSIQTechnology218.737.41-1.327-1.2360.9541.39
DECKConsumer Cyclical2110.0111.791.7851.0011.3001.37
DLRReal Estate215.315.890.584-0.0750.5930.98
DUtilities214.365.260.8990.4130.4851.85
INTUTechnology215.525.760.240-0.6520.3940.61
SCHWFinancial Services214.975.090.116-0.7890.4820.24
TMOHealthcare214.885.120.246-0.3680.5100.48
UNPIndustrials213.263.610.3550.2210.4000.89
ABTHealthcare203.814.550.739-0.5200.4091.81
AFRMFinancial Services2015.7813.31-2.473-5.3160.5714.33
AITechnology2016.3613.28-3.079-3.1030.7903.90
AMPFinancial Services206.746.07-0.671-1.0940.6631.01
ANFConsumer Cyclical2012.9314.471.536-0.2681.2001.28
ASOConsumer Cyclical2010.477.44-3.025-3.0801.0332.93
AVAVIndustrials2011.3912.240.8450.2041.0410.81
BGConsumer Defensive206.964.77-2.194-3.0740.7143.07
BWAConsumer Cyclical207.617.43-0.175-1.8130.9010.19
BYDConsumer Cyclical208.426.51-1.907-3.5290.6942.75
CFBasic Materials205.494.59-0.901-1.1030.6111.48
CINFFinancial Services206.415.32-1.090-1.7840.8211.33
CMFinancial Services204.646.571.9340.9960.5253.68
CRUSTechnology209.308.72-0.580-2.3560.8440.69
CTSHTechnology207.537.08-0.446-1.4560.8080.55
CTVABasic Materials206.696.800.117-0.8460.7820.15
CWHConsumer Cyclical2011.6010.64-0.960-1.5761.1480.84
CZRConsumer Cyclical207.706.53-1.176-1.4980.9161.28
DUKUtilities203.492.88-0.615-0.5670.4501.37
HONIndustrials203.394.060.6710.9200.3661.83
HOODFinancial Services2011.428.93-2.491-3.5700.4146.02
SOFIFinancial Services2013.828.82-4.998-4.6510.33614.89
VSTUtilities207.086.06-1.023-2.1990.9371.09
ADPTechnology194.163.99-0.169-0.4530.6080.28
AEOConsumer Cyclical1910.4110.21-0.200-1.5440.8960.22
AIGFinancial Services194.603.55-1.058-1.7390.4982.13
AJGFinancial Services195.375.27-0.101-1.8280.8530.12
AMBATechnology1912.5513.621.0722.3461.4150.76
APOFinancial Services195.165.180.017-1.3880.7210.02
APTVConsumer Cyclical198.519.531.020-2.4491.0460.98
ATIIndustrials1910.4112.842.433-1.7261.0132.40
AZOConsumer Cyclical195.214.51-0.702-2.2230.6841.03
BLDRIndustrials1910.6610.58-0.079-1.1261.0880.07
CAHHealthcare195.215.780.5690.7920.8000.71
CAKEConsumer Cyclical199.328.45-0.866-1.0520.7541.15
CGFinancial Services197.796.85-0.940-1.9131.0370.91
CMCIndustrials197.285.90-1.375-2.1370.8541.61
CMIIndustrials195.855.16-0.691-1.3430.5391.28
CNIIndustrials195.705.16-0.537-0.5610.4691.15
CRSIndustrials1911.0014.873.8712.6741.1283.43
DBXTechnology197.475.69-1.780-3.9620.8462.10
DDBasic Materials195.015.250.2370.2720.4860.49
DOCSHealthcare1917.0121.204.1951.9581.2143.46
DRIConsumer Cyclical197.065.27-1.794-2.4320.8652.07
ETNIndustrials195.214.19-1.016-2.1180.6431.58
ADITechnology184.734.910.180-0.1310.5050.36
ANConsumer Cyclical188.945.64-3.303-5.1170.9873.35
APHTechnology187.698.210.528-0.5340.7710.68
APPCommunication Services1815.1320.935.7942.5800.9036.42
ASANTechnology1816.4217.501.077-4.4821.3080.82
BMOFinancial Services184.695.981.2950.1340.4372.96
BROSConsumer Cyclical1813.4315.071.633-0.1481.1081.47
BURLConsumer Cyclical189.018.86-0.149-3.9671.2010.12
CCKConsumer Cyclical186.926.960.041-1.9760.7150.06
CHKPTechnology187.156.41-0.743-1.9560.6911.07
CIHealthcare184.673.72-0.947-1.0480.5111.85
CMCSACommunication Services184.924.63-0.291-1.2850.5620.52
CNQEnergy186.265.61-0.644-2.4580.6381.01
CPRTIndustrials186.314.48-1.833-3.6200.6282.92
CSXIndustrials183.863.12-0.731-1.3510.5551.32
DARConsumer Defensive189.088.70-0.3840.0570.8750.44
DEOConsumer Defensive184.885.130.2510.1170.5650.44
DGXHealthcare186.015.68-0.329-0.7180.7400.45
DOWBasic Materials184.294.410.118-0.9150.2980.39
DUOLTechnology1815.2016.601.398-0.6001.3371.05
SNPSTechnology187.317.17-0.144-1.7630.8130.18
ACLSTechnology1712.069.57-2.490-3.1591.3611.83
AFLFinancial Services173.444.100.6650.3300.4651.43
ALNYHealthcare1710.0712.232.157-1.4631.0802.00
AMEIndustrials175.336.851.5191.1450.6442.36
APDBasic Materials175.638.072.4442.2050.4964.92
AUBasic Materials178.1910.141.9523.0440.9092.15
AVTRHealthcare179.126.86-2.266-3.6801.1292.01
AYIIndustrials178.8510.031.179-1.0901.3220.89
BBWIConsumer Cyclical179.869.53-0.3310.2621.0770.31
BCOIndustrials177.896.92-0.973-3.8440.9021.08
BIIBHealthcare175.074.53-0.539-0.2970.9560.56
BNTXHealthcare179.235.85-3.376-3.3041.0793.13
BTUEnergy1711.219.38-1.823-3.2780.7542.42
CARGConsumer Cyclical1712.5612.810.256-3.1471.3570.19
CENXBasic Materials1713.9314.550.625-0.2781.7580.36
CEBasic Materials178.7310.671.9470.4240.9612.03
CIENTechnology1711.0912.831.738-0.3931.2661.37
DHRHealthcare175.166.721.5611.6550.5123.05
DOCNTechnology1716.6717.530.866-6.1231.5280.57
METACommunication Services178.5010.572.0620.6640.15713.13
AGBasic Materials166.627.090.466-0.0400.7500.62
ALCHealthcare1611.566.95-4.6080.3050.8905.18
ALGTIndustrials1610.4011.290.891-0.5281.4110.63
AMSCIndustrials1618.5118.620.108-4.1442.1060.05
APPNTechnology1612.7413.350.607-3.4271.4910.41
AVBReal Estate165.145.260.120-1.7180.5660.21
AXFinancial Services169.7310.350.6220.0951.1180.56
BSXHealthcare164.954.30-0.651-0.5700.4871.34
BUDConsumer Defensive164.575.941.3731.6840.4553.02
BYNDConsumer Defensive1616.1312.90-3.227-3.9671.0802.99
CACCFinancial Services1610.787.82-2.958-5.0651.9401.53
CCBasic Materials168.217.40-0.812-2.1291.0370.78
CEGUtilities167.469.011.5520.7790.7762.00
CMEFinancial Services163.442.48-0.964-1.5630.3752.57
CNXEnergy169.877.12-2.748-3.6151.3152.09
COURConsumer Defensive1616.3515.36-0.991-1.4762.3120.43
CTASIndustrials166.005.08-0.916-1.1980.6771.35
DLOTechnology1617.0321.484.445-1.5412.1002.12
DOVIndustrials166.216.590.3740.0760.8100.46
DTTechnology1610.989.57-1.413-0.7881.2281.15
SMCITechnology1613.8618.304.4407.0920.36512.15
VRTIndustrials1611.4811.910.434-1.4851.1200.39
ACMRTechnology1514.4516.151.7033.7501.6061.06
APPSTechnology1518.6923.524.8277.4590.9754.95
BBWConsumer Cyclical1516.8815.17-1.717-5.9091.8060.95
BEKEReal Estate1511.0511.820.7680.3100.8320.92
BENFinancial Services156.966.40-0.557-0.8450.9530.58
BMBLCommunication Services1514.9415.640.700-5.0611.1790.59
BTIConsumer Defensive154.835.480.6501.0200.5591.16
CLConsumer Defensive153.402.44-0.965-0.9300.4841.99
CNCHealthcare156.326.480.159-0.8770.7570.21
CNXCTechnology1510.809.57-1.229-1.2350.8301.48
COTYConsumer Defensive1511.659.47-2.179-3.2120.8702.51
CPIndustrials155.494.49-1.002-1.8610.5171.94
CRLHealthcare158.998.40-0.582-1.6751.0630.55
CVEEnergy158.286.19-2.092-3.8750.9112.30
CVLTTechnology1510.1912.292.1081.4101.3871.52
IONQTechnology1514.2615.020.757-4.8781.2700.60
SPGIFinancial Services153.904.590.6980.9410.6981.00
ACIConsumer Defensive147.269.842.5770.7351.0312.50
AERIndustrials147.087.550.467-1.3110.7660.61
AESUtilities147.318.931.624-0.2711.0261.58
ARCBIndustrials1410.699.97-0.719-0.6231.0620.68
ARMKIndustrials145.965.87-0.090-1.4140.9670.09
AWKUtilities144.554.720.167-0.6300.4110.41
AXTABasic Materials147.177.410.248-1.4490.9300.27
CLSKFinancial Services1413.3115.462.153-1.8961.2041.79
COOHealthcare147.047.150.1100.2120.9610.11
CRKEnergy1411.5812.080.503-0.9461.6530.30
CRSPHealthcare148.046.61-1.432-2.9730.9551.50
ADNTConsumer Cyclical1311.316.78-4.533-5.8421.4043.23
AMCCommunication Services1314.4811.24-3.238-3.9800.3918.29
AMNHealthcare1310.3613.242.881-0.2871.1852.43
AMRBasic Materials1311.9111.990.0860.2501.3050.07
ARGXHealthcare138.927.20-1.725-3.6581.4321.20
ASTSTechnology1316.9914.05-2.941-5.3791.2542.34
AVTTechnology137.046.21-0.823-0.4510.9400.88
BALLConsumer Cyclical137.227.380.1640.1740.6940.24
BKEConsumer Cyclical138.405.73-2.672-3.5031.1892.25
BNSFinancial Services134.657.983.3284.0320.5286.30
BOOTConsumer Cyclical1312.278.78-3.488-3.3921.9651.77
CACITechnology137.6610.673.0131.4671.3122.30
CAVAConsumer Cyclical1312.2512.05-0.199-0.3011.0690.19
CFGFinancial Services136.594.74-1.849-2.5760.7132.59
CGNXTechnology1310.0910.990.892-1.3621.2580.71
CHDConsumer Defensive135.293.06-2.233-2.6780.6123.65
CNKCommunication Services1310.937.79-3.140-4.3201.2202.57
COHRTechnology1311.8412.000.155-3.0891.4400.11
WDCTechnology137.834.18-3.654-3.5720.5266.94
AAOITechnology1220.7920.22-0.574-7.3882.3490.24
ACMIndustrials125.554.86-0.689-1.9500.8110.85
AEHRTechnology1221.0030.379.3704.9971.5865.91
AGIBasic Materials127.899.511.6275.1320.9351.74
APPFTechnology1212.2913.861.5692.8401.9830.79
ARMTechnology1210.5211.000.478-5.4520.4201.14
BAHIndustrials128.5111.432.9253.4131.2392.36
BLTechnology1210.4012.111.7090.6801.3721.25
CALXTechnology1215.6515.47-0.178-0.1821.4250.12
CNPUtilities124.994.52-0.478-1.6660.8850.54
CPRIConsumer Cyclical129.8412.943.102-1.5871.1252.76
CXWIndustrials1210.369.87-0.487-3.1831.6810.29
ALGMTechnology1111.6314.933.3022.2731.0943.02
ANSSTechnology117.707.64-0.056-1.0170.7910.07
AREReal Estate117.1810.303.1161.0050.8423.70
BCConsumer Cyclical118.196.29-1.901-2.3610.9242.06
BMRNHealthcare117.385.77-1.602-1.9510.7132.25
CASYConsumer Cyclical117.147.310.177-0.2870.9690.18
CLHIndustrials118.018.970.959-1.2490.7581.27
DQTechnology1114.7110.02-4.693-9.5361.2863.65
AGCOIndustrials107.735.96-1.770-2.5651.1571.53
AIRIndustrials1010.339.84-0.486-1.5751.2830.38
ALABTechnology1016.7013.18-3.520-7.8091.7861.97
APAMFinancial Services107.255.32-1.925-1.0710.7032.74
APLDTechnology1017.5918.651.0594.0811.5470.68
ARCCFinancial Services105.153.81-1.348-1.9700.7911.70
ARRYTechnology1018.2321.222.9911.7961.8031.66
AWIIndustrials107.2711.854.5805.9841.3023.52
BAXHealthcare105.624.51-1.104-1.6610.8541.29
BDXHealthcare105.044.28-0.758-1.8490.7770.98
BHFFinancial Services109.0410.451.4161.2191.0221.38
BHPBasic Materials105.786.120.340-2.0860.4630.73
BRZETechnology1015.5915.35-0.242-0.9631.2610.19
BXMTReal Estate107.074.56-2.516-2.8661.1292.23
CCSReal Estate1011.558.39-3.164-3.6181.4352.20
CDWTechnology106.638.131.4951.0551.0071.49
CHTRCommunication Services106.867.460.6060.0500.8670.70
CLMTBasic Materials1013.4014.661.265-0.1692.0910.60
DANConsumer Cyclical1010.759.79-0.958-3.8351.6930.57
GEVIndustrials107.608.641.042-0.5510.9101.15
RDDTCommunication Services1014.4013.90-0.503-5.1761.0540.48
RKLBIndustrials1013.579.01-4.567-6.6341.6552.76
AGOFinancial Services96.877.951.0750.9100.8701.23
ALKSHealthcare99.758.98-0.771-1.8521.3110.59
AUPHHealthcare918.7617.72-1.0410.1802.5210.41
AXSMHealthcare915.9413.71-2.224-3.8782.1551.03
BBTechnology912.2010.81-1.387-3.1620.7471.86
BCRXHealthcare914.4910.05-4.439-5.1571.4733.01
BCSFinancial Services98.293.78-4.510-5.6981.0504.29
BLKBTechnology99.526.01-3.510-3.9431.1233.13
CALConsumer Cyclical912.0112.900.881-0.8191.3630.65
CARTConsumer Cyclical911.409.79-1.6031.2930.9331.72
CLSTechnology913.7115.631.9162.5791.4821.29
CNMIndustrials99.938.44-1.482-2.8081.2641.17
CSTMBasic Materials99.019.400.393-3.1371.6160.24
DINOEnergy98.205.32-2.880-5.7860.9213.13
DIODTechnology99.059.090.046-1.7891.3130.03
DKEnergy911.4213.782.3610.7681.9201.23
ARLOIndustrials817.4514.41-3.042-3.7962.3531.29
AVYConsumer Cyclical86.094.73-1.359-1.4600.8661.57
BBBY—814.874.43-10.437-10.8731.04310.01
BCCBasic Materials89.9210.230.317-0.8891.4240.22
BFHFinancial Services811.818.89-2.928-2.8841.1242.60
BIRKConsumer Cyclical89.799.21-0.578-2.6851.0360.56
CHEFConsumer Defensive812.0110.90-1.110-2.8671.7390.64
CRCEnergy88.028.130.114-1.8191.3540.08
CSLIndustrials88.138.11-0.017-2.6371.4580.01
CVIEnergy811.528.00-3.520-6.3181.8641.89
DBRGFinancial Services810.9011.260.353-2.5981.3350.26
DINConsumer Cyclical810.247.46-2.781-2.0351.7271.61
DVCommunication Services814.3214.820.495-7.4021.2950.38
ACHRIndustrials712.7216.013.284-2.0101.0963.00
AEISIndustrials711.1312.351.2153.2361.6340.74
ALRMTechnology79.405.59-3.803-3.9621.7982.12
AOSIndustrials76.595.73-0.866-0.6000.9320.93
ASConsumer Cyclical714.2217.663.4474.8421.2792.69
ATKRIndustrials710.667.04-3.620-4.4311.4742.46
AZZIndustrials77.057.610.5581.2990.7690.72
BIOHealthcare78.414.91-3.506-3.7061.1623.02
BLMNConsumer Cyclical710.844.37-6.464-7.0711.3164.91
CGCHealthcare712.367.93-4.434-6.4970.8825.03
COCOConsumer Defensive714.3319.805.473-1.2712.1312.57
CORZTechnology711.819.19-2.628-6.4531.5361.71
CORHealthcare75.614.52-1.092-1.1250.3902.80
CPAIndustrials77.404.50-2.903-4.2371.0612.73
CRDOTechnology719.7316.57-3.160-6.6692.0421.55
CRHBasic Materials76.006.160.163-1.1930.8120.20
CRIConsumer Cyclical710.796.79-4.006-4.2551.5982.51
DBFinancial Services76.925.29-1.6270.0780.8481.92
DTEUtilities74.264.500.241-0.2380.5190.46
ABMIndustrials69.516.36-3.149-5.0441.3732.29
ABRReal Estate68.435.89-2.537-3.1041.0282.47
AEEUtilities63.983.00-0.984-1.7500.3992.46
AGXIndustrials616.6015.50-1.093-5.2962.1020.52
AXSFinancial Services66.064.81-1.249-2.3690.8261.51
BAMFinancial Services65.2010.855.6491.1560.6868.23
BANCFinancial Services69.555.81-3.742-3.2781.2722.94
BEAMHealthcare614.398.08-6.305-7.7582.3292.71
BJRIConsumer Cyclical612.4415.683.2443.6891.9491.66
BOKFFinancial Services67.3611.534.1692.9871.2313.39
BRBRConsumer Defensive68.6111.703.086-1.3480.9363.29
BWXTIndustrials68.939.200.2621.4891.7440.15
BZCommunication Services616.939.08-7.850-11.8492.0673.80
CORTHealthcare615.437.30-8.130-9.1622.3173.51
CRBGFinancial Services66.294.67-1.613-1.6590.9561.69
CRWVTechnology615.2423.017.7727.9290.8159.54
CRIndustrials68.589.811.2221.2341.1731.04
DLBIndustrials66.716.25-0.463-2.4440.7380.63
DSGXTechnology66.082.76-3.321-3.4550.6395.20
NBISCommunication Services612.5921.428.83610.5720.82910.66
SNDKTechnology612.4410.53-1.911-0.9351.4261.34
ACGLFinancial Services55.676.140.4711.6790.6420.73
AITIndustrials56.546.24-0.2970.1290.6740.44
ALTHealthcare517.5311.56-5.971-11.3142.3122.58
ALVConsumer Cyclical56.644.44-2.202-3.1320.9432.34
ARESFinancial Services56.804.44-2.359-4.6261.2291.92
ARWRHealthcare512.834.42-8.408-7.9101.8284.60
ASHBasic Materials55.595.880.291-1.0980.6070.48
AVPTTechnology511.5914.723.1341.4331.6801.87
BCECommunication Services54.913.74-1.1700.0450.7521.56
BROFinancial Services55.388.653.2713.9760.7744.23
BTDRTechnology513.9017.403.495-2.1142.2901.53
BBasic Materials56.207.201.007-0.1161.1520.87
CRCLFinancial Services512.1013.080.979-4.9850.7111.38
CWIndustrials57.086.31-0.773-0.9781.3290.58
DAVETechnology519.0215.80-3.222-1.9992.7491.17
DDSConsumer Cyclical515.8915.77-0.1282.1392.9340.04
DJTCommunication Services518.0110.75-7.261-2.1031.6504.40
ABGConsumer Cyclical49.188.18-1.000-5.8931.7440.57
ACHCHealthcare414.6117.522.9116.2182.0051.45
ADTIndustrials413.0811.97-1.115-1.8862.1670.51
AGMFinancial Services46.005.46-0.538-2.0670.8820.61
AIZFinancial Services45.735.69-0.031-0.9580.5660.06
ALSNConsumer Cyclical47.717.55-0.166-0.4830.8170.20
AMGFinancial Services48.622.57-6.057-6.0701.4414.20
AMPLTechnology416.927.00-9.915-10.6561.9645.05
AMRCIndustrials412.569.51-3.046-3.9771.9111.59
AMTMIndustrials412.643.00-9.644-9.7441.8335.26
ANDEConsumer Defensive410.909.65-1.243-6.6181.8710.66
ATENTechnology413.515.99-7.522-8.3232.2283.38
BKUFinancial Services49.236.43-2.799-3.3161.1512.43
BULLTechnology49.747.70-2.038-1.4240.6822.99
CAMTTechnology411.429.21-2.209-4.4992.2570.98
CDEBasic Materials49.1611.772.6061.4211.6451.58
CFRFinancial Services48.069.641.579-2.3520.9921.59
CHDNConsumer Cyclical47.675.93-1.743-1.0530.9451.85
CIFRTechnology415.6714.54-1.124-2.9211.8990.59
CLOVHealthcare414.378.29-6.074-6.4350.9056.71
COLMConsumer Cyclical48.528.530.009-0.4191.4750.01
CYTKHealthcare417.4811.58-5.901-3.0301.8033.27
DOXTechnology44.545.731.1880.0420.6831.74

LER is |mean edge| ÷ round-trip cost. Below 1.0 the effect cannot pay for its own execution. Sorted by event count; columns become sortable, and the period filter appears, once scripting loads. A company reports once a quarter, so choosing a single quarter selects one event per name and the Events column reads 1 — a true point-in-time cross-section, in which the mean and the median of one number are the same number. Choose a year to pool roughly four and let them separate again.

Does edge cluster by sector?

A hypothesis put by the principal: in particular quarters, certain sectors outperform on earnings and are more correlated — the example being NBIS (+31.0%) and PLTR (+28.5%), the two largest positive edges of 2026 Q3, both AI names. The observation is exact. Technology is 25.6% of that quarter’s 195 names, so the top two both landing there runs at about 1 in 15 by chance — uncommon, and not on its own evidence, because we are looking at it precisely because it happened. The question below is therefore not whether the top two are related, but whether sector explains more of the cross-sectional spread than shuffled labels do.

Does buying the straddle pay, sector by sector?

The question that decides money, and not the one η² answers. η² asks whether sectors DIFFER from each other; this asks whether any single sector’s mean signed edge is positive and clears its own round-trip cost. It is also not what |edge| measures — |edge| is symmetric, so a name that missed by −15 points counts exactly as much as one that missed by +15. Only the SIGNED mean says whether realised volatility arrived above the implied that was paid for it. Intervals bootstrap calendar months, because earnings cluster in season.

SectorEvents Mean implied %Mean edge % Median edge %CI low CI highNet of cost %
Basic Materials4217.420.343-0.624-0.1950.873-0.415
Utilities1825.330.325-0.440-0.2800.909-0.224
Technology17409.850.218-1.199-0.2470.678-0.475
Communication Services3788.690.042-0.816-0.7930.810-0.454
Real Estate2317.06-0.208-0.986-1.0360.644-0.954
Consumer Cyclical12008.94-0.327-1.647-0.7620.097-1.123
Healthcare7997.59-0.331-0.877-0.9420.234-1.056
Industrials9618.25-0.333-1.098-0.9360.214-1.148
Consumer Defensive5436.84-0.360-1.033-0.7990.088-0.940
Financial Services9616.41-0.462-0.953-0.892-0.029-0.999
Energy3226.87-0.557-1.229-1.041-0.032-1.198

Not one sector clears its cost. The two intervals that exclude zero are both NEGATIVE — Financial Services and Energy, where selling paid. Technology is positive on the mean (+0.218%) and deeply negative on the median (−1.199%): you lose on most Technology events and the average is carried by the tail. That gap is the NBIS/PLTR phenomenon exactly — real, and a tail rather than a harvestable edge.

Direction — do certain sectors outperform?

Sector labels shuffled within each quarter, 2,000 times. The shuffle holds the quarter's common market factor and the sector size distribution fixed, so only the sector assignment varies.

Pooled η² 3.64%of cross-sectional variation
Shuffled null 3.13%median of 2000 permutations
p, pooled 0.0410clusters beyond chance
Largest quarter Q2 ’25η² 7.05%
p, family-wise 0.7181max η² across 25 quarters
Quarterη² % EventsNull η² % p
Q2 ’257.052923.200.032
Q1 ’256.793482.650.010
Q3 ’266.711954.860.212
Q4 ’204.922803.250.186
Q4 ’214.803492.710.074
Q1 ’264.603043.080.175

Dispersion in percentage points — and why this one is a trap

The same test on |edge|. It looks like the strong result. It is not a result at all: |edge| is measured in points of spot, and a name priced for a 20% move can miss by 10 points where one priced for 4% cannot. Across the 11 sectors, mean implied move and mean |edge| correlate at r = 0.986.

Pooled η² 5.05%of cross-sectional variation
Shuffled null 3.13%median of 2000 permutations
p, pooled 0.0005clusters beyond chance
Largest quarter Q2 ’26η² 9.41%
p, family-wise 0.4128max η² across 25 quarters
Quarterη² % EventsNull η² % p
Q2 ’269.413102.990.002
Q3 ’269.281954.680.048
Q1 ’257.833482.690.006
Q4 ’217.233492.680.005
Q2 ’237.013222.850.014
Q1 ’216.582703.170.074

The same claim, scale-free — |edge| divided by the implied move

Dividing out the implied move removes the units effect and leaves the question actually being asked: is the options market PROPORTIONALLY worse at pricing some sectors? This is the controlled test, and it is the one that decides.

Pooled η² 3.13%of cross-sectional variation
Shuffled null 3.14%median of 2000 permutations
p, pooled 0.5137indistinguishable from shuffled labels
Largest quarter Q3 ’26η² 6.43%
p, family-wise 0.8291max η² across 25 quarters
Quarterη² % EventsNull η² % p
Q3 ’266.431954.740.253
Q2 ’225.612384.040.219
Q1 ’235.583232.830.051
Q1 ’215.572703.450.116
Q2 ’264.063103.030.258
Q1 ’243.953892.390.114

2026 Q3 was a good quarter to point at. It ranks second of 25 on the points version of dispersion and first on the scale-free one, so the instinct about the quarter was sound. Neither ranking clears the family-wise bar: against the null distribution of the largest η² across 25 quarters, the best quarter gives p = 0.41 in points and p = 0.83 scale-free. Note what the per-quarter p-values do on their own — four read below 0.05 in the points version, which is exactly the yield of 25 draws from noise. That gap between the per-quarter column and the family-wise figure is the entire reason the family-wise figure is computed.

This page carried the uncontrolled version for about an hour. The points result was published as “dispersion holds” before the scale control was run. It was caught by asking what dispersion actually meant, which is the question that exposes a units effect. The correction is recorded here rather than quietly swapped, because the sequence — strong result, obvious-in-hindsight denominator, result gone — is the more useful artefact.

Standard sectors, and “AI” is not one. NBIS, PLTR, APLD, CIFR and DOCN all land in Technology alongside 120-odd unrelated names, so a genuine theme effect is diluted here rather than isolated. A hand-drawn AI list would isolate it — and would be drawn by someone who has already seen which names did well, which is the failure this project pre-registers against. Testing it properly needs a theme list frozen on evidence that predates the returns.

What would change this

Control windows (G2). 8,232 matched non-event windows are queued. Until they land, none of the figures above are attributable to the announcement rather than to the names.

The back expiry (G7). With two expiries that both contain the event, total variance separates into a diffusive part that scales with time and a jump part that does not, which identifies the implied move directly. With one expiry it is not identified. The same decomposition run on control windows must return a jump variance of approximately zero — there is no event there. That check tests the method, not the hypothesis, and it is the one worth watching.

Two silent defects were already found in this estimator. Files written before 2026-08-16 carry no column header and parsed to nothing — 387 of the first 400. And daily bars are stored split-adjusted while option strikes are not, so comparing across the two turned NVDA’s 13% February 2024 move into an 89% one. That second bug moved the mean edge from −0.17% to +13.29% — a sign flip, and a spectacular-looking finding. Nothing errored. The only tell was the mean and the median disagreeing in sign, which is why this page leads with a distribution rather than a number.

Phase Space Research · measurement only, no position implied measured 2026-08-18

Naked Options

Ranked point estimates for buying calls and puts outright — on the desk's own alert triggers, and around earnings. A different question from the rest of this site, and deliberately a different bar.

The other tabs ask: is it real?

Pre-registered gates, permutation nulls, family-wise correction, a cost hurdle. The objective is to avoid believing something false, and the correct answer is usually “no”.

This tab asks: which is least bad?

No significance filter anywhere. Every cell ranked on its point estimate, because when a position is going to be taken regardless, demanding significance to choose between alternatives imports the wrong decision rule.
Alert positions19,99020230103 to 20260729
Best alert cell-3.6%of 18 exit policies tested
Earnings cells408 with a positive mean
Sweep-max inflation+4.1%median best cell of a SHUFFLED grid
Drift-adjusted-10.8%the surviving cell, market drift removed

Four things are carried over from the stricter method, because they survive the change of objective: entry at the ask (a mid-based number is not reachable), the whole grid shown rather than the winner alone, the sweep maximum measured against a shuffled grid, and win rate, median and worst-decile beside every mean — a naked option is a lottery payoff and the mean is the single worst summary of it.

Alert triggers — the desk's own calls

Every option and SPX alert with a resolvable contract, 19,990 positions from 20230103 to 20260729. Returns are on the option premium, not the underlying.

Hold to close

ChannelDTE nWin % Close % Max offered % Median offered % Ever +25%Ever +100%
spx8+402742.0%-1.9%+14.0%+7.9%16.9%0.7%
options3-7434532.8%-6.3%+47.8%+21.7%47.3%14.8%
spx3-774335.7%-6.8%+23.3%+12.0%31.5%3.0%
spx1-255730.2%-12.0%+55.5%+25.3%50.6%14.9%
options1-2484526.8%-13.4%+68.3%+26.7%51.5%22.6%
options8+26720.2%-17.7%+17.0%+1.9%29.6%8.2%
options0DTE504114.4%-21.9%+145.9%+35.8%55.7%33.2%

Max offered is the best the position ever showed before it closed — the maximum favourable excursion. The gap between it and the close column is the part of the outcome decided by exit rather than by entry, and on the options channel that gap is 102.5 percentage points: mean max offered +88.2% against a mean close of −14.3%. A third of 0DTE alerts touched +100% at some point.

The mirror is just as large. Those same positions showed a mean WORST of −63.9%; 67.9% touched −50% and 29.3% touched −90% before closing. Most positions visit both a large peak and a deep trough, which is why the exit grid below reports two bounds instead of one number.

Exit policy

A limit take-profit, with and without a stop. The path extremes do not record their order, so a position that touched both the target and the stop is genuinely ambiguous; both bounds are shown rather than one number that hides a choice.

ChannelTarget Stopn Win %Optimistic % Pessimistic %
optionshold to closenone1449824.2%-14.3%-14.3%
options+25%none1449852.9%-17.8%-17.8%
options+25%-50%1449852.6%-8.6%-29.7%
options+50%none1449842.3%-17.4%-17.4%
options+50%-50%1449841.4%-6.9%-26.6%
options+100%none1449833.6%-15.8%-15.8%
options+100%-50%1449831.9%-4.5%-22.8%
options+200%none1449827.9%-15.2%-15.2%
options+200%-50%1449824.8%-4.1%-19.3%
spxhold to closenone541739.5%-4.2%-4.2%
spx+25%none541745.3%-3.7%-3.7%
spx+25%-50%541745.2%-3.0%-5.1%
spx+50%none541741.6%-3.6%-3.6%
spx+50%-50%541741.5%-2.8%-4.5%
spx+100%none541740.2%-4.0%-4.0%
spx+100%-50%541740.0%-3.2%-4.3%
spx+200%none541739.6%-4.1%-4.1%
spx+200%-50%541739.3%-3.3%-3.9%

The finding worth the most here is the one that looks like good news. A +25% take-profit lifts the win rate from 24% to 53% and makes the expectation worse — it caps the few winners that carry the whole distribution. The exit policy that makes an equity curve feel survivable does not make it profitable, and it is the policy most often adopted for exactly that reason.

Earnings — the full moneyness grid

Bought at the ask on the last close before the report, held to expiry, split by side, moneyness and the name's trailing 20-session realised volatility. Every cell with at least 150 contracts is shown.

SideMoneyness OTM % rv20n Win %Held to expiry % Median % Max offered % Median offered % Ever +100%
put+25 to +50mid1585.7%+125.3%-100.0%+125.3%-100.0%5.7%
call+8 to +15mid233112.4%+18.6%-100.0%+96.0%-100.0%17.2%
call+3 to +8mid295921.5%+10.4%-100.0%+92.8%-100.0%30.0%
call+3 to +8low294918.0%+6.2%-100.0%+102.5%-100.0%26.6%
call-3 to +3mid482834.4%+3.4%-100.0%+73.9%+24.4%34.3%
call+15 to +25mid11927.0%+3.1%-100.0%+71.7%-100.0%10.2%
call+3 to +8high270723.8%+1.6%-100.0%+71.6%-66.9%29.4%
call-3 to +3low527735.5%+1.5%-93.3%+69.9%+22.2%34.9%
call-10 to -3mid589542.1%-1.5%-20.2%+51.0%+32.3%23.2%
call-25 to -10high1015542.8%-2.3%-9.7%+38.0%+23.1%11.6%
call-25 to -10mid1147343.6%-3.2%-6.5%+26.5%+17.3%6.2%
call-10 to -3high470840.2%-3.5%-29.5%+58.4%+36.4%24.8%
call-3 to +3high384733.4%-3.8%-97.9%+67.4%+23.5%32.2%
call-25 to -10low1160444.7%-4.1%-4.4%+17.7%+11.8%2.1%
call-10 to -3low641144.3%-4.2%-11.1%+38.8%+26.6%16.0%
call+15 to +25high22999.2%-4.8%-100.0%+92.2%-100.0%15.3%
call+8 to +15high284015.8%-5.2%-100.0%+72.4%-100.0%22.7%
put-25 to -10mid940944.4%-6.9%-6.3%+11.6%+6.4%2.9%
put-3 to +3low510033.5%-7.0%-100.0%+16.5%-40.2%22.7%
put-25 to -10low927240.8%-7.0%-7.5%+6.4%+2.4%1.3%
put-3 to +3mid473234.4%-7.1%-93.0%+17.7%-28.0%23.7%
put+8 to +15mid267710.1%-7.4%-100.0%+10.4%-100.0%9.9%
put-10 to -3mid530742.7%-7.8%-18.2%+20.0%+10.2%13.1%
put-25 to -10high820742.4%-9.5%-9.8%+15.6%+6.6%4.6%
put-10 to -3low556741.2%-10.4%-18.2%+12.6%+4.4%9.4%
put+15 to +25low3073.9%-10.8%-100.0%-6.2%-100.0%3.9%
call+15 to +25low2822.8%-12.0%-100.0%+104.2%-100.0%5.0%
put+15 to +25mid11374.9%-12.5%-100.0%+10.3%-100.0%4.8%
put+3 to +8mid324819.3%-13.8%-100.0%+5.2%-100.0%17.8%
put-10 to -3high436638.3%-15.4%-32.4%+15.3%+0.0%14.4%
call+25 to +50high14734.8%-17.7%-100.0%+110.2%-100.0%7.9%
put-3 to +3high380830.3%-20.6%-100.0%+5.6%-40.6%20.5%
put+3 to +8low339614.6%-21.0%-100.0%-1.4%-100.0%14.2%
put+3 to +8high281821.7%-28.7%-100.0%-4.5%-100.0%18.6%
put+8 to +15low18216.4%-38.6%-100.0%-26.6%-100.0%7.2%
put+8 to +15high319811.5%-41.8%-100.0%-21.3%-100.0%12.4%
call+8 to +15low13975.7%-43.2%-100.0%+44.1%-100.0%11.9%
put+15 to +25high23866.3%-51.2%-100.0%-44.4%-100.0%6.8%
call+25 to +50mid3911.3%-61.4%-100.0%+20.0%-100.0%4.9%
put+25 to +50high7311.0%-95.5%-100.0%-94.6%-100.0%0.8%

The earnings max-offered figures are a LOWER BOUND. The chain archive stops at the report date, so the option's own price after the announcement is not yet on disk — that is the opt_post harvest stage. What can be measured is the peak intrinsic value along the underlying's path, which ignores whatever time value remained at the peak and can only see daily closes. The true figure is higher than every number in those two columns.

The top cell of any grid this size is mostly selection. Shuffling the outcomes across cells and taking the best of the shuffled grid gives +4.1% as a median and +25.8% at the 95th percentile — so treat anything below that as indistinguishable from luck.

What survives the controls

One earnings cell is stable across the sample: calls, +3 to +8% out of the money, mid trailing volatility — +10.4% overall on 2959 contracts, and positive in every era (+13.3%, +10.9%, +6.3%). It is the only cell in the top ten that does not flip sign across the sample.

And most of it is the market, not the option. The same cell measures +11.9% as observed and -10.8% once the market's +1.00% mean drift over the holding window is removed — so 190% of it is the bull market, not option mispricing. The underlying rose +1.00% on average over these holding windows and was up 52.2% of the time. That lifts every call cell and depresses every put cell for reasons that have nothing to do with whether options were mispriced — which is also why calls occupy almost every top row of the grid above.

Position size, not edge, is what decides the outcome for this payoff. Every historically positive cell sits far out of the money, which is exactly where the win rate collapses — several are below 10%. A genuinely positive expectation at a 10% hit rate still ruins an account sized as though it were a coin flip, and the worst-decile column reads −100% almost everywhere: the ordinary outcome for a long option is that it expires worthless.

What would resolve the open half. The peak-versus-trough ordering needs intraday option paths for these positions — the archive holds daily extremes only. That is a different harvest from the one currently running, and until it exists the honest statement is that hold-to-close is a measured loser and peak-capture is unmeasured. No result on this page settles the second.

This is measurement, not advice, and nothing here is a recommendation to trade.