Costly Oak Markets
PhD Research in Financial Market Price Discovery Mechanisms
Live Trading
5 Years of Experience in Live Markets - Applying ideas on pockets of inefficient price delivery
Market Ideas
The Efficient Market Hypothesis (EMH), dominant since the 1960s, holds that asset prices fully reflect available information, leaving little scope for systematic edge. Yet this view largely abstracts from how prices are actually delivered in modern markets.In practice, price discovery is not instantaneous or frictionless. Markets often leave behind “signatures” — recurring patterns where price accelerates, overshoots, and then returns to re-test specific levels before continuing. These signatures may reflect information being incorporated into prices, but they also reveal the mechanics of the price delivery process: liquidity gaps, order-matching rules, and algorithmic feedback loops.What emerges is a picture where markets may be informationally efficient in the EMH sense, yet structurally inefficient in the short run. The delivery of prices can be uneven, creating temporary inefficiencies that repeat with surprising regularity. These dynamics blur the line between EMH and microstructure, suggesting that efficiency is not a static state but a process — one in which short-term rebounds into “unfinished” price pockets are an essential part of how markets reach equilibrium.
EMH has dominated for 60+ years
Explains information efficiency, but not how prices are delivered
Repeated signatures show efficiency is a process, not an instant fact
Structural frictions can create daily, observable inefficiencies
Contact and CV
Theories
This project re-examines high-frequency identification of monetary policy shocks. Standard designs rely on fixed event windows around announcements to isolate the “surprise” component, but they implicitly assume the market’s adjustment completes within that preset clock. In practice, execution frictions and liquidity conditions can stage the response, so a fixed window can capture only part of the move or pick up unrelated noise, distorting the measured shock.I propose a liquidity-aware dynamic event window that adapts to market conditions in real time. Operationally, the window closes when bid–ask spreads and depth normalise towards pre-event baselines and when the initial impulse region is revisited—signals that the mechanical delivery of prices has largely run its course. A simple Time-to-Delivery (TTD) metric captures how long this takes, ensuring the identification reflects both the informational move and the short-run delivery phase.By aligning the window to observable market mechanics, the resulting shock series is cleaner and more comparable across regimes, improving downstream impulse responses and transmission estimates in rates, equities and FX. The framework is deliberately practical—implementable with standard L1 microstructure data—and readily extends to other macro shocks (e.g., oil and carbon), where the relevant window may range from seconds to hours depending on liquidity and uncertainty conditions.Selected references: Gürkaynak, Sack & Swanson (2005); Jarociński & Karadi (2020); Miranda-Agrippino & Ricco (2021); Veronesi (2002).
My Projects
Costly Oak Markets All Rights Reserved
Aidan Yates
PhD Student
Proprietary and Live Futures Trader
🇬🇧 UK DISCLAIMER (FCA COMPLIANCE)Educational Content Only – No Personal Advice
This material is provided for educational and demonstration purposes only. It is not intended to be, nor should it be construed as, financial advice or a personal recommendation to buy, sell or hold any financial instrument. Nothing in this material takes into account your individual objectives, financial situation or needs. You should not rely on it for making any investment decisions.Regulatory Status
We are not authorised by the Financial Conduct Authority to provide personal advice or recommendations. All content is produced independently of any broker or product provider, and we do not receive commissions or other remuneration linked to any trades you may undertake.Risk Warnings
• Trading and investing carry a high degree of risk. You may incur losses, including the loss of your entire capital.
• Past performance is not a reliable indicator of future results. The value of investments and the income derived from them can fall as well as rise.
• You may get back less than you invested, and in extreme cases, receive nothing at all.
• You should only trade with funds you can afford to lose.Suitability & Independent Advice
This content is not tailored to your personal circumstances. Before acting on any information, you should consider seeking independent financial advice from an FCA-authorised adviser.Demo Account Simulations
Any examples shown using real-time demo (simulated) trading accounts are for illustration only. Simulated performance does not involve real money and cannot replicate actual market conditions, slippage, liquidity constraints or emotional pressures. Consequently, demo results may differ materially from the results you achieve with real funds.⸻🇺🇸 US DISCLAIMER (CFTC / NFA COMPLIANCE)Educational Purposes Only – Not Investment Advice
All content, including live demonstrations using simulated (“demo”) accounts, is provided solely for educational and illustrative purposes and does not constitute an offer or solicitation to buy or sell any commodity interest, security or other financial instrument. It is not intended as, nor should it be construed as, personalised investment advice.Risk Disclosure
• Trading futures, forex, options and other leveraged instruments entails substantial risk and is not suitable for all investors.
• You may lose some or all of your invested capital.
• Past performance (actual or hypothetical) is not indicative of future results.Hypothetical Performance Disclaimer
The following statement is provided in accordance with CFTC Regulation 4.41 and NFA Compliance Rule 2-29, and must be displayed prominently alongside any hypothetical or simulated performance data:“HYPOTHETICAL PERFORMANCE RESULTS HAVE MANY INHERENT LIMITATIONS, SOME OF WHICH ARE DESCRIBED BELOW. NO REPRESENTATION IS BEING MADE THAT ANY ACCOUNT WILL OR IS LIKELY TO ACHIEVE PROFITS OR LOSSES SIMILAR TO THOSE SHOWN.ONE OF THE LIMITATIONS OF HYPOTHETICAL PERFORMANCE RESULTS IS THAT THEY ARE GENERALLY PREPARED WITH THE BENEFIT OF HINDSIGHT. IN ADDITION, HYPOTHETICAL TRADING DOES NOT INVOLVE FINANCIAL RISK, AND NO HYPOTHETICAL TRADING RECORD CAN COMPLETELY ACCOUNT FOR THE IMPACT OF FINANCIAL RISK IN ACTUAL TRADING. FOR EXAMPLE, THE ABILITY TO WITHSTAND LOSSES OR ADHERE TO A PARTICULAR TRADING PROGRAM IN SPITE OF ADVERSE MARKET CONDITIONS OR LACK OF LIQUIDITY MAY ADVERSELY AFFECT ACTUAL TRADING RESULTS.FURTHERMORE, HYPOTHETICAL TRADING PROGRAMS IN GENERAL ARE ALSO SUBJECT TO THE FACT THAT THEY ARE DESIGNED WITH THE BENEFIT OF HINDSIGHT. THERE ARE NUMEROUS OTHER FACTORS RELATED TO THE MARKETS IN GENERAL OR TO THE IMPLEMENTATION OF ANY SPECIFIC TRADING PROGRAM WHICH CANNOT BE FULLY ACCOUNTED FOR IN THE PREPARATION OF HYPOTHETICAL RESULTS AND ALL OF WHICH CAN ADVERSELY AFFECT ACTUAL TRADING RESULTS.NO REPRESENTATION IS BEING MADE THAT ANY ACCOUNT WILL OR IS LIKELY TO ACHIEVE PROFITS OR LOSSES SIMILAR TO THOSE SHOWN. PAST RESULTS ARE NOT NECESSARILY INDICATIVE OF FUTURE RESULTS AND ACTUAL PERFORMANCE MAY VARY MATERIALLY.”Important Notes on Hypothetical / Simulated Trading
• Simulations are performed on demo accounts and do not involve real financial exposure.
• They cannot replicate market slippage, latency, order rejections, liquidity constraints or emotional factors.
• No simulation can fully account for the impact of risk management under live conditions.
• Performance figures—whether actual or hypothetical—should not be interpreted as guarantees of future financial results.Independent Responsibility
You alone are responsible for verifying and understanding the risks associated with any strategy. Consult a registered financial professional if you have questions about your personal financial circumstances.⸻🌐 GLOBAL / UNIVERSAL DISCLAIMERGeneral Educational Notice
This content is provided for general information and educational purposes only, to audiences in the UK, the US and elsewhere. It does not constitute financial advice, a personal recommendation or an offer to buy or sell any financial instrument.Risk & Performance Warnings
• Trading carries risk: you could lose all your invested capital.
• Past performance (actual or simulated) is not indicative of future results.
• Demo account examples are purely illustrative; they do not reflect real-world trading conditions or emotional pressures.
• You should only trade with funds you can afford to lose and seek independent professional advice if uncertain.By accessing or using any of our materials, you acknowledge your understanding of these disclaimers and agree that we bear no responsibility for any trading decisions you make.⸻References / Links
• FCA Conduct of Business Sourcebook (COBS)
https://www.handbook.fca.org.uk/handbook/COBS/
• FCA Finalised Guidance FG15/1 (Past Performance)
https://www.fca.org.uk/publications/finalised-guidance/fg15-01
• FCA Policy Statement PS22/10 (Suitability Guidance)
https://www.fca.org.uk/publications/policy-statements/ps22-10
• CFTC Regulation 4.41 (Hypothetical Performance Disclosures)
https://www.ecfr.gov/current/title-17/chapter-I/part-4/subpart-B/section-4.41
• NFA Compliance Rule 2-29 (Hypothetical Performance Disclosure)
https://www.nfa.futures.org/rulebook/rule2-29.html
Do markets “deliver” on a schedule? Evidence from NASDAQ futuresTL;DR: Using 15-second NQ futures data, I find that raids of relative equal highs/lows (REH/REL) occur disproportionately in a recurring 20-minute “macro window” each hour: :50 → :10. After adjusting for the window length (20 minutes vs 40 minutes), raids per hour are ~24% higher inside the window; lows in particular are ~38% higher.Why this mattersIf certain minutes of the hour consistently host the strongest “delivery” moves (the tape runs obvious resting liquidity like equal highs/lows), then the market’s microstructure is time-structured, not purely random. That’s a practical edge (timing + location) and a theoretical bridge to a liquidity-aware view of price adjustment.DataInstrument: CME Globex NASDAQ futures (NQ), DataBento OHLCV-1s exportPeriod: 31 days (Aug–Sep 2025)Time zone: All analysis in UTC, New York conversions only for session filtersPre-processing:Resampled to 15-second bars (proper OHLC aggregation; volumes summed)Excluded the opening bell skew: removed 09:25–09:35 New York every day(No other filtering; RTH vs ETH left as configured)DefinitionsRelative Equal Highs/Lows (REH/REL): two swing highs (or lows) within 1 tick (0.25) of each other over a rolling window; recorded as candidate liquidity pools.Raid: first break through a recorded REH/REL level with a quick return (within 8×15s bars = 2 minutes).Macro window: the 20-minute slice from HH:50 to (HH+1):10. Everything else in the hour is the non-macro window (40 minutes).Because the macro window is half the length of the non-macro window, raw counts are time-normalised to a raids-per-hour rate:Macro counts × 3 (60/20)Non-macro counts × 1.5 (60/40)ResultsRaw raid counts (unadjusted):non-macro macro
High raids 448 253
Low raids 352 242Time-adjusted raid rates (per hour):non-macro macro
High raids/hr 672.0 759.0 (+12.9% inside macro)
Low raids/hr 528.0 726.0 (+37.5% inside macro)
Total/hr 1200.0 1485.0 (+23.8% overall)Interpretation. After correcting for the window lengths, the market raids resting liquidity more often inside :50 → :10 than outside it. The effect is strongest for lows (+37.5%). Practically, this backs the idea that delivery (runs to obvious liquidity) is time-clustered rather than spread uniformly over the hour.What exactly I did (replicable steps)Load & resample. Read DataBento OHLCV-1s for NQ → 15-second OHLC, sum volume.Remove opening skew. Drop all bars 09:25–09:35 NY daily.Detect levels. Mark REH/REL if two highs/lows are within 1 tick over a rolling window (≈30 minutes worth of 15s bars).Detect raids. When price first trades through a recorded level and returns within two minutes, record the break timestamp.Label the clock time. For each raid, mark in macro if break time falls in :50 → :10; otherwise out of macro.Aggregate fairly. Compute raw counts and raids per hour (macro ×3, non-macro ×1.5).Report. Present high and low raids separately and combined.Why the adjustment mattersA naïve count would always favour the non-macro window (it’s twice as long). Scaling to per-hour rates gives an apples-to-apples comparison. On that basis the macro slice is busier, which is exactly the claim we wanted to test.Caveats (and how I handled them)Opening bell distortion: excluded 09:25–09:35 NY.OHLCV not quotes: this analysis counts raids (price behaviour), which only needs OHLCV. (For spread-based equilibrium timing I use MBP/MBO; not required here.)Parameter choices: 1 tick tolerance, 2-minute return, 15-second bars. These are pre-set and sensible for NQ; robustness below.Session mix: results use your chosen RTH/ETH setting from config. If you want RTH-only, enable it in the loader and rerun.Robustness I’ll add next (and you can too)Alternative tolerances: 0- to 2-tick REH/REL tolerance and 1–3 minute return windows.Different cadence: test :20 → :40 and :00 → :20 as placebos.Shuffle test: randomly permute raid timestamps day-by-day; genuine time structure should vanish.Proportion tests: formalise with a two-proportion z-test or bootstrap on the adjusted rates.What this supports, theoreticallyThe tape does not deliver uniformly. It targets obvious liquidity in repeating time windows inside the hour. That is consistent with a time-structured execution/clearing pattern: short delivery bursts (often terminating at REH/REL), followed by tail-off as liquidity normalises.Figures (drop into the article)outputs/macro/macroadjustedrates.png — bar chart: adjusted raids per hour (macro vs non-macro, highs & lows).outputs/nqminutehist.png — raids by minute-of-hour (should show visible clustering).outputs/nq_periodogram.png — frequency view; look for a bump near ~3 per hour.
Select out of 2 mentorship options to suit different levels of commitment and focus. Select the option that best aligns with your goals.
• Core Mentorship – £2000
• Advanced Mentorship – £2000
📘 What You’ll Receive:Core Mentorship: Designed for beginners - goal is to be able to learn enough to be able to create and trade profitable personal trading models without help
Please include in the form:
• Full Name
• Email Address
• Telegram Username
• Trading Experience (briefly describe)
• What do you want to gain from mentorship?
• Preferred Start Date