dataofmen

Overnight AI Research (2026-07-14)

16 min read

2026-07-14 Overnight AI Research

Summary

07:00 briefing context available to this clipping job did not expose a clean AI-research final briefing body; the retrievable local clipping content for today’s run contained a 07:00 KST research briefing focused on trading microstructure, crypto liquidity, ETF/stablecoin flows, and risk-sizing. I preserved it as raw clipping input rather than inventing an AI-specific summary.

Core theme: liquidity state is the hidden variable. Across crypto futures, order-book herding, stablecoins, ETF flows, MSTR beta, stress scenarios, and split-session risk, the practical lesson is to classify market microstructure and flow mechanism before taking directional exposure or increasing size.

Raw Briefing

Overnight Trading Domain Deep Research - 20260714

[07:00 KST] The Quarter-Hour Effect: Periodic Algorithmic Trading and Return Predictability in Cryptocurrency Futures

  • URL: https://arxiv.org/abs/2607.09426v1
  • Source Type: paper
  • Domain: microstructure, crypto-structure, execution
  • Trading Relevance: 9
  • Time Horizon: intraday
  • Actionability: study
  • Henry Risk Link: execution, liquidity, FOMO
  • Why it matters: Crypto perpetuals may have clock-time liquidity and volatility bursts that are not visible in daily candles. Henry should treat execution time as a risk variable, not a neutral detail. A signal that works around quarter-hour boundaries can also create FOMO if interpreted as directional conviction rather than microstructure timing.
  • Raw excerpt / abstract: The authors report periodic bursts in volatility and volume at one-, five-, and quarter-hour marks in six Binance perpetual contracts. They associate the bursts with algorithmic trading because trade-size roundness declines sharply during the bursts. Their Autocorrelation Map reveals serial dependence in order flow and returns at quarter-hour openings, and opening order imbalance forecasts four-to-twelve-hour returns.
  • Cold read: This is a market clock effect: algos may concentrate activity at predictable timestamps, creating short-lived information and adverse-selection zones.
  • Trading insight: Do not place market orders blindly at neat clock boundaries in crypto perps. For any crypto execution, compare fills and slippage inside versus outside the one-, five-, and fifteen-minute windows.
  • Change sensor: Watch quarter-hour volume spikes, order imbalance persistence after the clock boundary, and whether the effect decays as participants arbitrage it.
  • Follow-up question: Does the quarter-hour effect survive after fees, spread, and realistic order queue position for a non-market-maker?

[07:00 KST] When Does Order Flow Matter? State-Dependent L2 Liquidity-State Transitions in Crypto Futures

  • URL: https://arxiv.org/abs/2607.09230v1
  • Source Type: paper
  • Domain: microstructure, execution, crypto-structure
  • Trading Relevance: 10
  • Time Horizon: intraday
  • Actionability: study
  • Henry Risk Link: liquidity, execution, regime
  • Why it matters: The paper says pre-event L2 liquidity state is the first-order predictor of post-event liquidity regime, while order flow only adds value after the L2 state model. This directly supports Henry’s risk-first frame: before asking where price goes, ask what liquidity state the book is in. For leveraged products, bad liquidity state can turn a correct thesis into a poor trade.
  • Raw excerpt / abstract: The study uses Binance BTCUSDT and ETHUSDT futures from 2023-2026 with top-20 L2 order book data, trade-flow records, and macro-event windows. It finds a coarse pre-event liquidity state baseline strongly predicts post-event liquidity regimes; shallow nonlinear L2 models add robust gains; order flow adds further value only when layered on top of the L2 state model. ETH shows broader order-flow value, while BTC shows isolated passes.
  • Cold read: State beats story. Event labels and directional narratives are secondary if the book is already in a fragile or resilient state.
  • Trading insight: A pre-trade checklist should include L2 depth and spread regime before any event-window trade. If liquidity state is stressed, reduce size or avoid stop orders that can be harvested.
  • Change sensor: Monitor top-20 depth imbalance, spread widening before macro events, and symbol-specific differences between BTC and ETH liquidity transitions.
  • Follow-up question: Can a simple retail-accessible proxy, such as exchange depth and spread snapshots, approximate the paper’s L2 state baseline well enough for sizing decisions?

[07:00 KST] Herding and Liquidity in Order-Book Markets: A Robust Liquidity-Stress Crossover

  • URL: https://arxiv.org/abs/2607.08907v1
  • Source Type: paper
  • Domain: microstructure, reflexivity, risk-sizing
  • Trading Relevance: 9
  • Time Horizon: swing
  • Actionability: study
  • Henry Risk Link: liquidity, narrative, correlation, regime
  • Why it matters: This paper frames liquidity dry-up as an emergent crossover from herding and order-book mechanics, not just a news shock. Henry’s leveraged ETFs and high-beta positions are most vulnerable when reflexive buying becomes one-sided and then suddenly loses depth. The relevant danger is not volatility alone; it is the one-sided book.
  • Raw excerpt / abstract: The model combines fundamental-anchored zero-intelligence liquidity with a mid-anchored chartist herding layer. It finds an emergent liquidity-stress crossover where the fraction of events with a one-sided book rises to about 0.34 at high herder fraction and strength, while scrambled null cells show zero. Price-momentum herding carries a large reflexive component where buying begets buying.
  • Cold read: Trend strength can be liquidity fragility in disguise. The same reflexive mechanism that creates upside acceleration can erase exit capacity.
  • Trading insight: When a move becomes crowded, raise the required liquidity discount before adding. Position size should shrink when the market is rewarding momentum purely through herding rather than fundamental confirmation.
  • Change sensor: Watch one-sided order books, narrowing participation behind a move, and volume spikes without depth replenishment.
  • Follow-up question: What observable proxy best flags a transition from healthy trend to liquidity-stress crossover in US leveraged ETFs?

[07:00 KST] Robustness in Sequential Decision Making under Evolving Uncertainty: Evidence from High-Frequency Market Making

  • URL: https://arxiv.org/abs/2607.08291v1
  • Source Type: paper
  • Domain: execution, microstructure, risk-sizing
  • Trading Relevance: 8
  • Time Horizon: intraday
  • Actionability: study
  • Henry Risk Link: execution, sizing, liquidity
  • Why it matters: The key finding is uncomfortable: more robustness is not always better. Excessive conservatism can reduce profitability in illiquid markets by removing execution opportunities. Henry should separate risk control from paralysis; the goal is calibrated robustness, not zero risk.
  • Raw excerpt / abstract: The authors define two dimensions of robustness: uncertainty tolerance and action robustness. Simulation and empirical evidence show action robustness has a substantially larger impact than uncertainty tolerance. Excessive robustness may reduce profitability in illiquid markets by limiting execution opportunities.
  • Cold read: Risk controls are themselves strategy parameters. Too little robustness creates ruin risk; too much robustness creates missed execution and negative selection.
  • Trading insight: Predefine when to be conservative and when to be opportunistic. In illiquid names, a rigid limit-only rule can miss fills, while market orders can overpay; the execution rule must adapt to spread, depth, and urgency.
  • Change sensor: Track missed-fill rate, adverse selection after fills, and opportunity cost from overly passive orders.
  • Follow-up question: What execution metric should Henry record to know whether he is being disciplined or merely hesitant?

[07:00 KST] Generating Plausible Stress Scenarios via Large Deviations

  • URL: https://arxiv.org/abs/2606.31122v1
  • Source Type: paper
  • Domain: risk-sizing, macro-liquidity, capital-markets
  • Trading Relevance: 8
  • Time Horizon: cycle
  • Actionability: study
  • Henry Risk Link: leverage, sizing, correlation
  • Why it matters: Handpicked stress scenarios can be emotionally satisfying but structurally wrong. This paper argues for plausible stress configurations rather than arbitrary disaster stories. For Henry, this matters because 3x leverage risk is often hidden in correlated tail moves that have few historical examples.
  • Raw excerpt / abstract: The paper develops a method for generating plausible stress scenarios using a large-deviations principle. Conditional on a large loss, risk factors concentrate near the most likely stress configurations. The method extrapolates observed samples into more extreme scenarios while preserving relative plausibility of stress mechanisms, even when historical data have few stressed observations.
  • Cold read: Tail risk is not just bigger versions of normal days. The useful question is which combination of factors most plausibly creates a large loss.
  • Trading insight: Stress tests should be scenario families, not single anecdotes. For leveraged semiconductor exposure, test simultaneous real-rate rise, liquidity drain, USD strength, and volatility expansion rather than only a price gap.
  • Change sensor: Monitor correlation convergence, volatility regime shifts, credit spread widening, and reserve/TGA/RRP liquidity changes.
  • Follow-up question: What minimal stress-scenario template should be applied before any large add to SOXL or MSTU?

[07:00 KST] Split-Session Cluster GARCH for Overnight and Intraday Returns: The Role of Tail Heterogeneity

  • URL: https://arxiv.org/abs/2607.03669v1
  • Source Type: paper
  • Domain: risk-sizing, execution, capital-markets
  • Trading Relevance: 8
  • Time Horizon: swing
  • Actionability: study
  • Henry Risk Link: leverage, regime, sizing
  • Why it matters: The paper distinguishes overnight and intraday tail behavior. That matters for Henry because US equities held from Korea carry overnight gap exposure in local time and cannot be managed the same way as intraday risk. Leveraged ETFs compound this gap/tail asymmetry.
  • Raw excerpt / abstract: The model decomposes asset returns into overnight and intraday components and allows tail behavior to differ by trading session and sector. Results on US equity returns reveal pronounced tail heterogeneity between overnight and intraday returns. Session-specific tail parameters improve fit, and asset-level tail heterogeneity produces the strongest out-of-sample likelihood and global minimum-variance portfolio performance.
  • Cold read: The market has two risk machines: intraday liquidity risk and overnight discontinuity risk. Combining them into one volatility number hides the real exposure.
  • Trading insight: Stops do not fully control overnight tail risk. Sizing must assume the exit price can skip the stop during US after-hours or opening gaps.
  • Change sensor: Track overnight-versus-intraday return contribution, sector-specific gap frequency, and premarket liquidity after major macro releases.
  • Follow-up question: How much of Henry’s portfolio volatility comes from unmanageable overnight gaps versus tradable intraday moves?

[07:00 KST] Stablecoins and Safe Asset Prices

  • URL: https://www.bis.org/publ/work1270.htm
  • Source Type: research
  • Domain: macro-liquidity, crypto-structure, capital-markets
  • Trading Relevance: 9
  • Time Horizon: cycle
  • Actionability: monitor
  • Henry Risk Link: liquidity, correlation, regime
  • Why it matters: Stablecoins are no longer only crypto plumbing; BIS frames them as meaningful buyers of short-term US Treasury bills. This links crypto liquidity, Treasury bill yields, repo conditions, and broader dollar liquidity. Henry should treat stablecoin expansion/contraction as a cross-market liquidity signal, not a crypto-only metric.
  • Raw excerpt / abstract: BIS says dollar-backed stablecoins had combined AUM above 35 billion of US Treasury bills in 2025. The paper estimates that a $3.5 billion stablecoin inflow lowers 3-month Treasury bill yields by 0.71 basis points on impact and up to 4 basis points within 10 days, with stronger effects under Treasury market stress and as the stablecoin sector scales.
  • Cold read: Crypto dollar liquidity is becoming part of the safe-asset ecosystem. The boundary between crypto leverage and Treasury market plumbing is blurring.
  • Trading insight: BTC and high-beta crypto equities may respond not only to ETF flows but to stablecoin reserve allocation and T-bill market stress. Watch stablecoin supply as dry powder and as a Treasury demand channel.
  • Change sensor: Stablecoin AUM, T-bill holdings by issuers, 3-month T-bill yield pressure, repo stress, and regulatory reserve rules.
  • Follow-up question: Does stablecoin supply growth lead BTC risk appetite after controlling for ETF flows and broad dollar liquidity?

[07:00 KST] Following the Flows: ETFs, Stablecoins, and Where Capital Actually Went

  • URL: https://blog.amberdata.io/following-the-flows-etfs-stablecoins-and-where-capital-actually-went
  • Source Type: research
  • Domain: crypto-structure, reflexivity, capital-markets
  • Trading Relevance: 8
  • Time Horizon: cycle
  • Actionability: monitor
  • Henry Risk Link: liquidity, narrative, FOMO
  • Why it matters: Amberdata argues that a large ETF outflow episode was a basis-arbitrage unwind, not institutional capitulation. This is exactly the kind of narrative misread that creates emotional trading. Henry needs to classify flows by mechanism before reacting to headlines.
  • Raw excerpt / abstract: The report states that spot Bitcoin ETFs saw 269 billion after $77.3 billion YTD expansion. It frames October outflows as mechanical arbitrage unwinds after carry compressed from above 15% to under 5%, not fundamental selling.
  • Cold read: Flow data are not self-explanatory. The same outflow can mean capitulation, arbitrage closure, tax rotation, fee rotation, or hedging.
  • Trading insight: Before using ETF flows as bullish or bearish input, identify whether the flow is directional demand or basis trade mechanics. Do not trade headlines without the carry and funding context.
  • Change sensor: ETF net flows by issuer, CME basis, perp funding, stablecoin supply, GBTC-style fee rotation, and custody concentration.
  • Follow-up question: What dashboard can distinguish directional BTC ETF demand from hedge/basis unwind flows in near real time?

[07:00 KST] Cryptocurrencies in the Balance Sheet: Insights from (Micro)Strategy and Bitcoin Interactions

  • URL: https://arxiv.org/abs/2505.14655
  • Source Type: paper
  • Domain: crypto-structure, reflexivity, capital-markets
  • Trading Relevance: 9
  • Time Horizon: cycle
  • Actionability: monitor
  • Henry Risk Link: correlation, dilution, narrative, leverage
  • Why it matters: The paper documents that corporate BTC treasury firms can have material BTC beta, with BTC usually driving information flow. For MSTU/MSTR-style exposure, the core risk is not only company-specific leverage but reflexive BTC-equity coupling. Dynamic hedge ratios matter because the information driver can shift around financing announcements.
  • Raw excerpt / abstract: The authors assemble 39 public firms holding BTC through April 2025. They find significant positive co-movements with an average BTC beta of 0.62 and identify 12 companies, including Strategy/MSTR, with beta above 1. Transfer entropy identifies BTC as the dominant information driver, with brief announcement-driven feedback from stocks to BTC during major financial events.
  • Cold read: Corporate BTC treasury strategy converts equity into a levered, financing-sensitive BTC wrapper. The wrapper can feed back into BTC only during specific financing or announcement windows.
  • Trading insight: MSTU exposure should not be sized like normal equity beta. Treat it as BTC beta plus financing/reflexivity risk plus ETF-style leverage decay.
  • Change sensor: MSTR premium to NAV, ATM issuance, convertible terms, BTC beta drift, transfer of leadership between BTC and MSTR, and post-announcement liquidity.
  • Follow-up question: At what MSTR premium and BTC beta does MSTU become a poor way to express a BTC view relative to direct BTC or BTC ETF exposure?

[07:00 KST] Aggregating Orderbook Depth to Create Liquidity Metrics

  • URL: https://gitbook-docs.coinmetrics.io/tutorials-and-examples/tutorials/aggregating-orderbook-depth-to-create-liquidity-metrics
  • Source Type: official
  • Domain: microstructure, execution, crypto-structure
  • Trading Relevance: 8
  • Time Horizon: intraday
  • Actionability: now
  • Henry Risk Link: liquidity, execution, sizing
  • Why it matters: Coin Metrics frames order book depth as necessary for mature institutional markets and for estimating slippage. This is practical, not theoretical: Henry can turn liquidity from a vague word into measured depth around intended order size. For crypto-linked trades, depth fragmentation across venues is part of the risk.
  • Raw excerpt / abstract: Coin Metrics states that a liquid order book is necessary for a mature financial asset, and that the ability to quickly enter and exit large positions with small price impact or slippage is desirable for institutional market participants. The tutorial aggregates BTC spot order book depth across actively traded markets using Coin Metrics Market Data Feed.
  • Cold read: Liquidity is a measurable inventory of exit capacity, not a background assumption. Fragmented depth means headline volume can overstate executable liquidity.
  • Trading insight: Before a crypto or crypto-equity trade, define the expected exit size and compare it with visible depth and spread. If available depth cannot absorb the exit, the position is too large regardless of conviction.
  • Change sensor: Aggregated depth within 10 bps and 50 bps, exchange concentration, spread regime, and depth recovery after shocks.
  • Follow-up question: Can Henry build a simple liquidity heatmap using public exchange APIs before entering crypto-sensitive positions?

Run Summary [07:00 KST]

  • collected_count: 10
  • skipped_duplicates: 0
  • domain_mix: microstructure/execution 40%, risk-sizing 25%, macro/capital-markets 20%, crypto-structure 35%, AI-history 0%. Several items map to multiple domains; today’s run intentionally leaned into execution and crypto liquidity because q-fin.TR produced unusually relevant new papers.
  • top_theme: Liquidity state is the hidden variable. Across crypto futures, order-book herding, stablecoins, ETF flows, and MSTR beta, the key lesson is to classify the mechanism before taking directional exposure.
  • sharpest_insight: The same observed flow or price move can mean different things depending on microstructure state: quarter-hour algo bursts, basis unwind, stablecoin T-bill demand, and one-sided herding all look like momentum on a chart but imply different risk.
  • danger_of_misuse: Treating these papers as immediate buy/sell signals would be dangerous. They are process upgrades: execution timing, state detection, stress testing, and sizing discipline. Using them to justify bigger leverage without measuring liquidity would invert the lesson.
  • next_probe: Build a retail-accessible pre-trade liquidity checklist: spread, depth, timing window, funding/basis, ETF flow mechanism, and overnight gap risk.

Daily Cold Synthesis

  • Overnight Thesis: The edge is not predicting the next headline; it is identifying the liquidity state and flow mechanism before committing size.
  • World model update:
    • Stablecoins have crossed from crypto infrastructure into Treasury bill demand and short-rate microstructure.
    • Crypto futures show clock-time algorithmic behavior that can affect both execution quality and apparent short-horizon predictability.
    • Corporate BTC treasury equities behave like dynamic BTC wrappers whose beta and feedback loops change around financing events.
  • Trading-process lessons:
    • Always separate liquidity state from directional thesis before sizing.
    • Stress tests should search for plausible joint configurations, not dramatic one-off stories.
    • Overnight and intraday risk are different risk machines; a stop order cannot fully solve gap exposure.
  • Henry-specific caution:
    • Do not use ETF inflow/outflow headlines as emotional confirmation without basis and funding context.
    • Do not average into leveraged exposure when the underlying book is one-sided or depth is deteriorating.
    • Do not confuse risk control with hesitation; define in advance when execution should be passive, active, or avoided.
  • Signals to monitor:
    • Top-20 L2 depth and spread regime in BTC and ETH before macro windows.
    • Quarter-hour crypto perp volume and order imbalance persistence.
    • Stablecoin AUM and issuer T-bill allocation.
    • BTC ETF flow composition by issuer plus CME basis and perp funding.
    • MSTR premium to NAV, BTC beta drift, ATM issuance, and convertible financing terms.
  • Do not trade directly from this: These clippings describe mechanisms and research directions, not validated strategies for Henry’s account. Execution costs, tax/account constraints, overnight gap risk, and position concentration must be checked before any order.

Sources

Keywords

  • ai-research
  • overnight-briefing
  • papers
  • ai-community
  • market microstructure
  • crypto futures
  • algorithmic trading
  • order flow
  • L2 order book
  • liquidity state
  • herding
  • robustness
  • high-frequency market making
  • stress scenarios
  • large deviations
  • overnight returns
  • intraday returns
  • stablecoins
  • Treasury bills
  • ETF flows
  • Bitcoin
  • MicroStrategy
  • MSTR
  • MSTU
  • orderbook depth
  • execution risk
  • risk sizing

그래프 뷰