Daily Overview — Friday, October 23, 2026
A ~110-second narrated recap of the session generated automatically from TRAI's own datasets, with plain-English definitions of every quant term. Screening/research, not investment advice.
Written summary
Market regime
The VIX, the market's 30-day implied volatility, rose 0.4 points to 19.4, a transitional reading. SPY's realized volatility, how much it actually moved over 20 days, is only 9.8%; that 9.7-point gap is the volatility risk premium. The term structure, short-dated versus 3-month VIX, is in mild contango, which is normal. TRAI's regime score is +17.5: neutral and choppy.
What moved — and why
CRWD rose 8.2% with no earnings in the data, so the cause is unknown here.
Long momentum
Long momentum means a market-beating rise on heavy volume, closing above VWAP, the day's volume-weighted average price. SLB led with a residual z-score of 3.5, a market-adjusted move about 4 times its normal noise.
Short momentum & mean reversion
Short momentum had 1 names, led by T. Mean reversion, big moves on ordinary volume that may fade, flagged 10, led by AXP.
Scorecard — has it worked?
Over 126 replayed sessions, short momentum's expectancy, the average net profit per trade, is -0.46%, with a t-stat of -3.7: beyond ±2, so likely not luck. No bucket has a significant positive edge.
Correlation clusters
Correlation clusters are stocks that move together: the largest, 24 names led by semis, has an average correlation of 0.59, so today's 1 shorts are about 3 independent bets.
Earnings & gap risk
Gap risk is a jump past your stop overnight, often on earnings. 44 names report within 5 sessions, including picks AXP and SLB on October 23.
Sector rotation
Over 20 days, staples, utilities and energy lead SPY by 8 to 11 points; semis and tech trail by 12 and 6.
Concepts in today's overview
Each term used in the recap, in plain English.
Realized vs implied volatility
Realized volatility is how much a stock or index actually moved — here, SPY over the last 20 sessions, annualized. Implied volatility is what options prices say the market expects next; the VIX is the 30-day implied volatility of the S&P 500. Comparing the two tells you whether options are pricing more or less movement than the market has recently delivered.
Glossary →VIX term structure
Compares implied volatility across horizons: 9-day (VIX9D), 30-day (VIX) and 3-month (VIX3M). Normally short-dated vol is lower (contango). When near-term vol rises above longer-term vol (backwardation), traders are paying up for immediate protection — a stress signal.
Methods: regime model →VWAP
The volume-weighted average price is the average price paid during the session, weighting each 5-minute bar by its volume. Closing above VWAP means buyers were in control on average that day; below means sellers. TRAI requires it as confirmation for momentum picks.
Glossary →Residual (market-adjusted) returns
A residual return strips out the part of a move explained by the market: return − beta × SPY return, with beta estimated over the prior 60 sessions. What is left is the stock-specific piece — news, flows or earnings — rather than levered market exposure.
Glossary →Z-scores
A z-score measures a move in units of its normal noise: the move divided by its usual standard deviation. A z of 2 is roughly a 1-in-20 day; 6 is extremely rare. TRAI uses a residual z for momentum and a 1-day return z for mean reversion.
Glossary →Mean reversion
The tendency of an unusually large move to partly reverse. TRAI flags names with a 1-day |z| above 2 on ordinary volume (0.7–1.5× normal), because a big move without heavy participation is more likely to be noise that fades. The bucket shows the fade direction, not a forecast.
Methods: daily screen →Expectancy & t-statistic
Expectancy is the average profit or loss per trade after estimated costs — positive means the rule made money on average in the replay. The t-statistic asks whether that average could be luck: roughly, |t| above 2 suggests the result is unlikely to be chance alone, while smaller values are noise. Past replays do not guarantee future results.
Methods: scorecard →Correlation clusters
Groups of stocks whose daily returns move together (here, average pairwise correlation over recent sessions). Five picks from one cluster are closer to one bet than five. "Effective N" estimates how many independent bets a bucket really contains.
Methods: correlations →Gap risk
The risk that a price jumps past your stop between sessions — usually on earnings, news or macro data — so the loss is bigger than planned. Positions held into an earnings report carry it; being flat by the close avoids overnight gaps.
Methods: earnings & gap risk →Sector rotation
Which sectors are outperforming or lagging the market. TRAI compares 12 sector ETFs with SPY; "20-day relative" is the ETF’s 20-session return minus SPY’s. Leadership shifting toward defensives (staples, utilities) often signals a more cautious market.
Methods: macro & sectors →Learn more
Archive
How this is made
A script on the research box pulls TRAI's public JSON (/api/latest, regime, scorecard, events, correlations, macro, options), writes the narration only from numbers in that data, renders slides with matplotlib, voices it with a neural text-to-speech voice, and encodes an H.264/AAC MP4 with ffmpeg. Moves are only explained when the data contain a cause (e.g. an earnings report); otherwise they are labeled unexplained. See Methods. Screening/research, not investment advice.