20261008 US stocks maintain high levels, but interest rates and small-cap stocks are shaking the foundation of the rise

Market commentary as of 21:15 JST on October 8, 2026. Final prices are based on the October 7 ET close, and information from October 8 ET is distinguished as pre-market news. The conclusion of this report is that the continuation of the long-term upward trend and the weakness in the conditions for chasing gains from current prices coexist.

News: A resurgence in crude oil and interest rates cooled the day after reaching record highs.

In the US market on October 7, the S&P 500 fell about 0.2% to 7,801.77, and the NASDAQ Composite fell about 0.2% to 27,538.69. The Dow fell about 0.7% to 51,179.87, and the Russell 2000 fell about 1.3% to 2,793.20. While large-cap indices saw minor adjustments, the decline in small-cap stocks was notable. The S&P 500 had hit a record high the previous day. AP, October 7 closing data

Reuters reported that stocks retreated from record highs against a backdrop of concerns over crude oil supply and rising long-term Treasury yields. The 30-year Treasury yield reached its highest level in about 24 years, and the decline in stock prices narrowed during the session due to a pullback in crude oil prices. Sector differences emerged, with industrial stocks being weak and healthcare being relatively strong. This is not a phase where funds are spreading across all cyclical stocks. Reuters, October 7 market commentary

The pressure continued before the market opened on the 8th. According to Reuters’ 06:14 ET observation, S&P 500 futures fell 0.40%, NASDAQ-100 futures fell 0.57%, and the 10-year Treasury yield was reported at 5.34%. In addition to the surge in crude oil, concerns were raised that massive fundraising by AI-related companies is intensifying the competition for capital. Strong earnings forecasts for semiconductor companies did not lead to a broad rally in related stocks. These are pre-market snapshots and not the closing prices for the 8th. Reuters, October 8 pre-market, updated at 06:53 ET

What can be inferred from this combination is that stock prices can no longer be explained solely by expectations for corporate earnings. Even if earnings increase, if the interest rates used to discount those earnings to present value or the burden of securing investment funds rise simultaneously, the reaction of stock prices can be sluggish. This is an interpretation based on the news and not the result of estimating how much the rise in interest rates pushed down stock prices. The continuation of the AI growth story and the question of whether current prices have sufficient margin must be confirmed separately.

4-Report Analysis: Reading the environment, capital status, price position, and horizon-specific outlooks separately

The input consists of four daily reports: market-regime-daily-radar (MRDR), capital-flow-pulse-daily (CFPD), o-chan-technical-analysis (O-chan), and multi-horizon-market-forecast (MHMF), with a common price reference date of October 7 ET. MRDR covers the market environment, CFPD covers balances, relative prices, and participant positions, O-chan covers trends, support/resistance, and timing, and MHMF covers 10-day and 20-day model forecasts. In this report, the analysis target of MHMF is limited to 10 and 20 days.

MRDR: Under quiet price movements, internal confirmation of indices has weakened

The Market Regime Daily Radar composite score is 54/100, classified as Neutral-, down 2 points from the previous day. While the index technical axis fell 9.27 points from the previous day, the VIX is low at 15.08. Even if there is no major fear, the breadth of participation and momentum supporting the rise are not necessarily sufficient.

The S&P 500 is above its 50-day average of 7,680.46 and 200-day average of 7,241.31. Meanwhile, the Russell 2000 is 4.87% below its 50-day average and only 0.25% above its 200-day average. The ratio of MRDR’s top 200 large-cap stocks above their 200-day average is 50.25%, but the actual target is 199 stocks, not the entire market’s advance-decline line. The total market A/D is missing, so it cannot be certified that ‘market-wide participation breadth has recovered.’ Read the MRDR alternative acquisition path and moderate evidence confidence in conjunction with the composite score.

Do not summarize macro indicators into the single word ‘liquidity’

Regarding interest rates, the original source for MRDR is the daily par yield curve published by the US Treasury at the 2-year and 10-year points, with 2-year at 4.77%, 10-year at 5.28%, and a spread of 51bp. The publishing entity is the US Treasury, and the spread is calculated by MRDR. The weight of the interest rate axis is 20%, which does not mean that equal weight is given to each maturity. The target is two points on the Treasury curve, not the policy rate itself. Alternative/comparison effects are limited to verifying yfinance values and do not replace the original source. For noise removal, use not only daily changes but also multi-day changes and the spread, and the publication lag is generally considered to be from the current day to the next business day. Since the 5.34% pre-market figure on the 8th differs in both timing and observation path, do not mix it as the same final value.

The calculation and publication entity for M2 is the Federal Reserve Board, the distribution series is FRED’s M2SL, and the target is the US monthly money stock. The MRDR year-on-year figure for August is +5.66%. The weight is 15% of the entire liquidity axis, not 15% for M2 alone. Do not use other series as substitutes for M2, and distinguish between levels and growth rates. Use the difference between the year-on-year figure and the previous month’s year-on-year figure to suppress seasonal and level noise, accompanied by a publication lag of several weeks from the target month and revisions. It does not measure the daily inflow of funds for stock purchases.

The Fed balance sheet is the total assets of the consolidated Federal Reserve Banks based on the Federal Reserve Board’s H.4.1, using FRED’s WALCL. The observation on September 30 was $6.743 trillion, with a weekly change of a decrease of about $4.67 billion. This is also a component of the 15% liquidity axis, but it is not a substitute for M2, but a different observation of central bank assets. Compare not only the single-week level but also the weekly change with the previous change, and maintain the weekly lag of the usual Thursday publication. Do not certify a stop to QT or QE as policy based on small changes in the balance.

HY OAS is the option-adjusted spread for US high-yield corporate bonds calculated by ICE BofA and distributed by FRED, with an observation of 303bp on October 6. It is a credit stress confirmation element of the 15% liquidity axis and cannot be replaced by Treasury yields or HYG prices. Map the spread level to the credit stress band to avoid single-day price noise, accompanied by a publication lag of usually about the next business day. The MRDR text does not contain details of the stock-specific weights within the OAS series, and this has not been confirmed in the reference report. The exact allocation within the M2, Fed assets, and OAS axes has also not been confirmed from this text.

While MRDR includes values converted from CFNAI to a PMI scale, it is based on the Chicago Fed’s 85-series National Economic Activity Index, not the PMI itself, which is based on ISM corporate responses. In this report, we do not use this proxy value as a decisive factor for economic judgment. Even though the indicators are around 50, they originate from different sources, so we want to avoid assuming they are interchangeable.

CFPD: An increase in reserve liquidity is not the same as an increase in stock buyers.

In the Capital Flow Pulse Daily, the reserve liquidity balance proxy, calculated by subtracting TGA and RRP changes from Fed asset changes, increased by approximately $53.7 billion in the weekly comparison of source history. The main contribution is the decrease in TGA. On the other hand, in the difference from the previous day’s snapshot, both TGA and RRP have increased. Weekly increases and same-day absorption directions coexist, and one cannot say that ‘inflows are accelerating’ by connecting values with different comparison criteria.

The calculation entity for this proxy is CFPD, and it targets balances related to reserve liquidity, with coefficients of +1 for the Fed and -1 each for TGA and RRP. This is a different formula from the liquidity axis weight in MRDR. It does not serve as a substitute for M2 or net inflows into stocks. Handling noise involves separating the differences between source history and saved snapshots while maintaining the publication frequency. The Fed is weekly, TGA and RRP are daily, and the strict publication lags for each balance, the details of the TGA/RRP calculation entities, and the presence or absence of seasonal adjustments have not been confirmed in the reference report.

The CFPD relative price preference composite is +31.12, which is a non-linear transformation of the average of the standardized values of SPY/TLT, HYG/LQD, and the sign-inverted DXY. The calculation entity is CFPD, and the target is relative price preferences between assets. At the averaging stage, the three series are treated equally, and all series are required. Although scale differences are adjusted through standardization, it does not act as a substitute observation for the decomposition of net subscriptions/redemptions, order flow, or investment gains/losses. The strict reflection lag of prices and the noise removal effect other than standardization have not been confirmed in the reference report.

Therefore, even if relative prices are stock-oriented, it cannot be concluded that ‘new funds have entered the entire market.’ In CFPD, large-cap and growth stocks continue to dominate, and the small-cap/large-cap ratio is low at the 2.8th percentile within the history. While the direction is consistent with MRDR’s small-cap underperformance, both share prices or price ratios, so they are not counted as two independent votes. The CFPD publication judgment is for continued observation, and there is insufficient evidence to certify the structure or forecast.

Positioning: Check month-over-month changes, congestion, and exits in order.

The FINRA margin debt balance is a monthly value for the end of August, and the source history difference corresponding to the month-over-month comparison is +$36,607 million, or approximately $36.6 billion. The level is at the 91.7th percentile of the same-frequency history, and the standardized value for about one year is +1.79. Regarding the long-term range, all that can be said is that it is high within this reference history; the multi-year minimum/maximum values and the complete definition of the period have not been confirmed in the reference report. A high borrowing balance indicates leverage burden, but it does not prove buyer sentiment or the timing of a reversal. It is also not information that new borrowing increased on that day.

The CFTC target is limited to the Asset Manager net position of E-mini S&P 500. The September 29 observation is 904,003, the weekly publication difference is -30,452, and the approximate month-over-month comparison using four publication differences is -30,177. Although it is low at the 13.5th percentile of the same-frequency history, it does not represent the physical holdings of all institutional investors or other index futures. It has a time lag where the state on Tuesday is published on Friday at 15:30 ET, and the strict upper and lower limits of the long-term range have not been confirmed in the reference report.

Since the NAAIM observation is from August 19 and is stale, it is excluded from the direction judgment. The approximate month-over-month difference of +10.47 and the historical rank of 75.0th percentile in the report are diagnostic values of an old observation and do not support the current bullishness of operators. The latest month-over-month difference, long-term range, and current congestion level have not been confirmed in the reference report.

As an auxiliary observation of congestion, the total CBOE option open interest on October 6 was approximately 658.6 million, with an approximate month-over-month difference of +71.46 million, at the 87.7th percentile of history. Even if the high open interest can be observed, the number of people betting on the same side and the dealer’s hedging direction have not been confirmed. FINRA’s cash surplus is also not a measure of exits. Direct exit liquidity, such as order book depth, market impact, and redemption restrictions, has not been confirmed in the reference report, so we do not conclude that ‘exits will be clogged because open interest is high.’

Realized return calibration is verified by post-hoc checking 20 days after CFPD. In 62 observations, the average change in the SPY/TLT ratio was +2.3%, with a minimum of -2.1% and a maximum of +8.1%; the XLK level averaged +0.4%, with a minimum of -10.1% and a maximum of +9.8%. The change in SPY/TLT is not the return of SPY alone, and these are not expected returns narrowed down to the current FINRA/CFTC conditions. The return calibration by current congestion level and the loss distribution at the time of position liquidation have not been confirmed in the reference report. One cannot leap from observation to future superiority.

O-chan: Separate the upward trend from the timing here and now.

O-chan Technical Analysis confirms the price position independently of the comprehensive judgment of the other three reports. The 200-day average of SPY is 718.67 and is trending upward, but the closing price of 777.22 is 8.15% above it. QQQ is also at 757.73 against a 200-day average of 669.54, a divergence of 13.17%, indicating both a rising long-term line and chasing risk.

SPY is within the judgment zone of 773.38–777.44, and the time series that turned to support after breaking through resistance has not been confirmed. The 752.87–760.15 below is a candidate support zone where support was confirmed after a past resistance breakthrough. For the S&P 500 index, 7,771.48–7,816.70 is the judgment band, and 7,577.92–7,638.17 is the candidate support band. Do not convert index points and ETF prices; read them as separate levels. Do not reinterpret the current value of O-chan as being outside the judgment band based on the ‘no corresponding resistance zone’ display in MRDR.

The near candidate support band for QQQ is 736.85–748.35, but support confirmation after the breakthrough has not been confirmed. The candidate support band for the NASDAQ-100 is 30,328.79–30,770.63, and the Fibonacci defense line candidate of 30,976.28 remains auxiliary structural information. There is no newly confirmed RSI divergence this time, and additional layers of volume and supply/demand have not been output. The authenticity of the break and the support by volume have not been confirmed in the reference report.

O-chan’s timing classification is ‘wait in judgment band’ for SPY/S&P 500 and ‘wait’ for QQQ/NASDAQ-100. This is not a declaration of a shift to a long-term bearish stance, nor is it a classification derived from MRDR scores. For small-cap stocks, the price is close to the Russell 2000 200-day average of 2,786.31, with 2,795.48–2,832.37 above as a candidate resistance band and 2,706.36–2,735.10 below as a candidate support band. The ‘stocking candidate (before visual confirmation)’ in the report is not a final buy decision. The support band in the text does not overlap with the MA200, and consistency confirmation remains for the overlap expression in the summary. While small-cap relative underperformance continues, one cannot be certain of a rebound based solely on mechanically extracted candidates.

MHMF: Read the 10-day and 20-day figures along with the reservation of predictive power.

The 10-day and 20-day rise probabilities for the Multi-Horizon Market Forecast are 35.1% and 40.9% for QQQ, and 42.1% and 46.6% for SPY. The expected return proxies are -0.71% and -0.63% for QQQ, and -0.29% and -0.19% for SPY. The model’s outlook is weak, but it is clearly stated in each column that there is ‘no predictive power,’ so they cannot be used as figures that accurately represent the actual rise probability.

The displayed loss proxies are -6.82% and -9.89% for QQQ at 10 and 20 days, and -4.78% and -6.77% for SPY. Since the rise probability this time is less than 50%, these correspond to directional loss proxies with the 95th percentile of reference returns sign-inverted. Do not depict them as future decline widths of ETFs or lower-limit forecasts with probabilities. Expected returns are also not conditional expected values from independent regression, but proxies created from probabilities and past return scales.

All four targets have their direction lamps Off, execution judgment is No Trade, and the verification seal is not_ready. The completeness of price data does not imply the predictive power of the model. While the view is that MRDR is weaker than neutral, CFPD is an uncertified concurrent observation, and O-chan is awaiting timing, MHMF adds weaker numerical values accompanied by insufficient verification. Rather than building confidence based on the agreement of the four reports, it is necessary to leave the overlap of price information and the unconfirmed parts of each as they are.

What to watch in the next US market is the breadth of the rise rather than the record highs.

First, in the regular trading session on October 8 ET, we will confirm whether the pre-market interest rate and crude oil pressure continue. Regarding prices, I would like to focus on the judgment band of SPY between 773.38 and 777.44, and the resistance band near and just above the 200-day average of the Russell 2000. Whether the relative underperformance of small-cap stocks, not just large-cap stocks, eases, and whether the price breakout is accompanied by confirmation of support, will be the material for evaluating the foundation of the rise.

The direction for the next day is not determined solely by the CFPD balance proxy or monthly high leverage. Until O-chan’s volume confirmation and total market A/D are obtained, we will not extend the strength of prices to the health of the entire market. This article is a market commentary and does not convert support bands, classifications, or model values into trading instructions.

Sources

• AP: How major US stock indexes fared Wednesday 10/7/2026. October 7 ET closing summary, published 20:17 UTC (October 8 05:17 JST). Confirmed in the retrieved text.

• Reuters / MarketScreener: Wall Street ends lower, off record highs, as Treasury yields climb. Published October 7 16:01 ET, updated 19:13 ET (October 8 08:13 JST). Background on sector differences, crude oil, and interest rates.

• Reuters / Mint: Wall Street ends lower, off record highs, as Treasury yields climb. Published October 8 01:31 IST (05:01 JST). Background cross-checked. Since the prices in this article are provisional, the confirmed prices in the article use the matching values from AP and the daily report.

• Reuters / MarketScreener: Wall St futures slide as rising oil, yields dampen mood. Published October 8 06:51 ET, updated 06:53 ET (19:53 JST). Confirmed in the retrieved text. Futures price timestamp is 06:14 ET.

• market-regime-daily-radar / Market Regime Daily Radar — 2026-10-08 (MRDR). Text time 06:58 JST, US market standard October 7 ET. Market regime and macro decomposition.

• capital-flow-pulse-daily / Capital Flow Pulse Daily — 2026-10-08 (CFPD). Snapshot 06:40:58 JST, US market standard October 7 ET. Balance, relative price, entity position, and post-verification.

• o-chan-technical-analysis / O-chan Technical Analysis Index Daily — 2026-10-07 (O-chan). Created October 8 06:41:52 JST, latest bar October 7 ET. Trend, support/resistance, and timing judgment.

• multi-horizon-market-forecast / Multi-Horizon Market Forecast — 2026-10-07 (MHMF). Created October 8 07:03:09 JST, confirmed price standard October 7 ET. Only the 10-day and 20-day analysis columns are used.

The absolute path of the input, SHA-256, observation criteria, and news confirmation range are recorded in the input manifest in the same folder.