Live performance
Actual trading for the current macro strategy, which entered live trading in 2024.
Open research
Research Database
Explore research published by Ames Investment Systems. Open a paper to read its abstract and download its PDF.
Research downloads453downloads
Does a Time-Series Foundation Model Add Economic Value to Systematic Trading? A Six-Study Falsification Test of TimesFM
Publication date: 2026-09-24
Contributors: Ogdn Ames, Ames Investment Systems
Abstract
Time-series foundation models promise strong zero-shot forecasts, but their value to systematic trading depends on incremental information beyond conventional models and on whether forecast gains survive implementation costs. This paper evaluates Google's TimesFM 2.5 across five distinct designs and one deployment study spanning 2000-2026, using walk-forward estimation, same-slot controls, explicit trading costs, and later-date evaluation windows. TimesFM improves some volatility forecasts but provides no robust evidence of incremental economic value in these designs.
Keywords: TimesFM, time-series foundation models, volatility forecasting, systematic trading, volatility targeting, backtest overfitting, deflated Sharpe ratio
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
How High Can Interest Rates Go Before Markets Break?
Publication date: 2026-09-23
Contributors: Ogdn Ames, Ames Investment Systems
Abstract
Using evidence through September 23, 2026, historical episodes, balance-sheet stress arithmetic, valuation analysis, and conditional Monte Carlo scenarios, this study assesses whether U.S. markets have a universal interest-rate breaking point. It finds no defensible single threshold: stress depends on a shock's source, speed, curve shape, cash-flow response, leverage, liquidity, and refinancing structure. In an illustrative fiscal and term-premium shock with weaker earnings and higher risk premiums, a 50 to 100 basis-point increase in the 10-year Treasury yield sustained for 12 months is associated with conditional scenario frequencies of 40% to 89% for a 20% or larger month-end equity loss. Growth-led rate increases produce materially lower stress. The results support conditional stress analysis and risk monitoring, not point forecasts of market failure.
Keywords: interest rates, Treasury yields, financial stability, Monte Carlo simulation, market stress testing, liquidity, leverage, refinancing risk
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Regime-Conditioned Information Routing in Institutional Quantitative Trading
Publication date: 2026-09-21
Contributors: Ogdn Ames, Ames Investment Systems
Abstract
This technical research paper develops Regime-Conditioned Information Routing (RCIR), a falsifiable theory for when transformer-like adaptive information routing may justify its statistical and economic complexity in institutional quantitative trading. It identifies state-dependent variation in predictive relationships, rather than architectural novelty, as the potential source of routing value. The paper sets out a formal conditional-linear result, a target-free Regime Reallocation Index, prespecified hypotheses, and a frozen forward-validation standard using point-in-time data, realistic costs, and multiple-comparison controls. It concludes that transformers should be treated as optional adaptive routing components that must demonstrate incremental net out-of-sample value before deployment.
Keywords: transformers, quantitative trading, financial time series, regime change, attention, foundation models, systematic macro, model selection, backtest overfitting, point-in-time data
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Corporate Bond Information and Equity Risk Management: An Exploratory Multi-Stage Study of Credit Warnings, Yield Timing, and Adaptive Asset Selection
Publication date: 2026-09-14
Contributors: Ogdn Ames
Abstract
This study evaluates whether corporate bond information can improve equity investment decisions through credit-event analysis, yield-return visualization, fixed-rule backtesting, moving-average optimization, and adaptive portfolio construction. Bloomberg data through August 31, 2026 support an initial 54-bond sample and an expanded 320-bond discovery sample covering 30 equities. Credit-stress months showed a higher observed frequency of six-month peak corrections, but dependence-aware resampling included zero; a 30-session yield moving-average strategy reduced volatility and drawdowns while lowering Sharpe ratios; and an adaptive replay earned 3.49% annualized with a 0.42 Sharpe and 22.14% maximum drawdown. The evidence supports a contemporaneous yield-equity association and lower risk with reduced exposure, but does not establish dependable incremental credit-timing skill.
Keywords: corporate credit, equity timing, bond yields, moving averages, drawdown, backtest overfitting, walk-forward evaluation, credit markets
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Human Day Trading Versus Fly-Derived Reservoir Trading: A Deep Comparative Study of Returns, Return Structure, Risk Appetite, and a Prospective Single-Trader Test
Publication date: 2026-09-14
Contributors: Ogdn Ames
Abstract
This paper compares a fly-connectome-derived artificial recurrent trading system with empirical evidence on human day traders, focusing on returns, return structure, and risk appetite. The prior fly evidence does not establish a benchmark-beating trading advantage, while human day-trader research generally finds losses after costs alongside a small persistent-skill right tail. Human risk taking is endogenous and path dependent, whereas the fly system implements an engineered state-to-risk mapping. The paper therefore does not claim a winner and instead specifies a 252-session prospective intraday protocol with matched data, constraints, transaction costs, and preregistered outcomes.
Keywords: day trading, reservoir computing, Drosophila connectome, behavioral finance, risk appetite, return structure, transaction costs, prospective validation
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Fly-Derived Reservoir Computing for Momentum Trading: Retrospective Evidence, Engineering Strengths, and Empirical Failures
Publication date: 2026-09-13
Contributors: Ogdn Ames
Abstract
This paper evaluates whether a fixed recurrent reservoir derived from the male Drosophila central nervous system can improve momentum-based allocation decisions. Across 2,520 scenario and model runs, the fly-derived family generally trails buy-and-hold on the assigned mean fold objective for SPY, GLD, and QQQ; a small AAPL mean-of-ratios advantage reverses under equal-capital ensemble comparison. No multiplicity-adjusted comparison establishes superiority, and strength-preserving graph nulls are generally similar or slightly better. The study demonstrates a disciplined engineering pipeline but does not establish a trading advantage attributable to fly anatomy because the evidence is retrospective, incomplete, and exposed to data-snooping and execution-assumption risks.
Keywords: reservoir computing, Drosophila connectome, momentum, quantitative trading, backtesting, data snooping, Sharpe ratio, Sortino ratio
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Temporal Counting in MaleCNS-Derived Artificial Recurrent Networks: An Independent Replication, Memory-Capacity Extension, and Synchronized Blackjack Simulation
Publication date: 2026-09-13
Contributors: Ogdn Ames
Abstract
Structural connectomes constrain anatomical topology but do not specify functional neural dynamics. An initial MaleCNS v1.0 study found that a primary fixed reservoir lost to random graphs, whereas a postsynaptic-normalized comparison favored native topology on average across only three computational seeds. We conducted a prospectively specified independent replication using ten new input/readout/null seeds on the same 5,000-neuron anatomical subset, 2,048 fresh training shoes, 512 validation shoes, and 1,024 locked test shoes. Native, random, and degree/raw-strength-preserving families received identical training budgets. Fixed baseline true-count root-mean-square error (RMSE) averaged 0.8085 for native models and 0.9990 for random models. The primary seed-generalized comparison favored native topology. A validation-only search over longer-memory linear dynamics and 64, 128, or 256 readout states yielded mean RMSEs of 0.0097, 0.0150, and 0.0190 for native, random, and raw-strength-null families. An original GRU transferred to fresh shoes achieved 0.0928; exact accumulation remained error-free. A separate paired simulation covered 4,096 fresh shoes and 173,440 initial rounds. No estimator-versus-exact paired interval excluded zero. A synchronized video maps frozen artificial states to real soma coordinates while showing exposed cards, count errors, and same-stream settled equity. Results concern constrained artificial computation, not biological counting or profitability. Single-specimen anatomy, effective signed-strength confounds, and restricted dynamics limit generalization.
Keywords: connectomics, reservoir computing, computational replication, temporal memory, blackjack simulation
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Copper: Tariffs, Inventories and Investment Implications
Publication date: 2026-09-12
Contributors: Chip
Abstract
This conditional commodity report assesses copper after a sharp reversal in regional scarcity pricing and uncertainty over a prospective U.S. refined-copper tariff. It separates policy, physical-flow, and fundamental questions; compares LME, COMEX, and SHFE observations; and explains how inventory accumulation in the United States can tighten availability elsewhere without proving a global shortage. The report is tactically cautious at the September 11 price anchor while structurally constructive on electrification demand, and it does not establish an executable arbitrage, implementation-ready short, or unconditional buy recommendation.
Keywords: copper, tariffs, inventories, LME, COMEX, SHFE, electrification, commodity markets, investment implications
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
U.S.-Canada Tariffs: Incidence, Economic Exposure, and Conditional Policy Scenarios
Publication date: 2026-09-11
Contributors: Ogdn Ames
Abstract
This study examines U.S.-Canada trade restrictions using government instruments, Census trade values, central-bank publications, academic evidence, market observations, and archived institutional research. It distinguishes verified current measures from announced future actions, describes importer-side incidence and limits on retail pass-through, and presents transparent sensitivity calculations for a hypothetical tariff. The scenarios are conditional calculations rather than contemporary estimates, and the paper identifies where the evidence cannot support causal GDP, household-burden, employment, or political-probability conclusions.
Keywords: tariffs, Canada, United States, trade incidence, USMCA, retaliation, conditional scenarios, economic exposure
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Wisconsin’s $15B+ AI Campus Is Riding on an Unsettled Power Timeline
Publication date: 2026-09-11
Contributors: Ogdn Ames
Abstract
Wisconsin has more than $15 billion of AI infrastructure development riding on a power schedule that is no longer settled. In Port Washington, Vantage Data Centers is building the Lighthouse campus across 672 acres, with four data centers and 902 megawatts of critical IT load. All four buildings are under construction, and Oracle has said customer delivery is expected to begin in the second half of 2027. But on Aug. 6, the Public Service Commission of Wisconsin revoked its earlier determination that American Transmission Co.’s transmission application was complete. As of Sept. 9, the public record did not show a revised energization date following that regulatory reset. That gap matters because a campus can be physically built and commercially committed while still producing a much weaker investment return if the electricity arrives late
Keywords: Wisconsin, AI, Campus
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
U.S. Treasury Buybacks Market impact, counterfactual simulations, and policy effectiveness
Publication date: 2026-09-11
Contributors: Ogdn Ames
Abstract
The September 10 operation is verified. Treasury accepted $5.187 billion in face value against a $6 billion ceiling. The accepted basket implies approximately $3.684 billion in settlement cash, including accrued interest. The 10-year benchmark yield nevertheless rose 12.20 basis points on operation day. That combination does not establish failure. Treasury’s stated objective is regular, predictable off-the-run liquidity support—not a benchmark-yield target. Daily quotations show a relative liquidity signal, but selection bias, imperfect matching, spillovers, and a failing yield pretrend prevent a clean causal verdict. The executed Monte Carlo model illustrates how yields can rise even when a buyback helps relative to the no-buyback counterfactual. Its intervention coefficients are assumptions, not estimates of the September operation’s causal impact
Keywords: U.S. Treasury Buybacks Market impact, counterfactual simulations, and policy effectiveness
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Virginia Data Center Investors Have Billions at Stake - and Approved Megawatts Can Still Disappear
Publication date: 2026-09-11
Contributors: Ogdn Ames
Abstract
This analysis explains why announced data-center capacity in Virginia is not equivalent to an investable operating asset. Drawing on Northern Virginia transactions, the Digital Gateway zoning reversal, land values, utility obligations, tenant commitments, and power-delivery risk, it shows how project value depends on durable entitlements, contracted electricity, executable construction plans, financing, and long-term leases.
Keywords: Virginia, Northern Virginia, data centers, AI infrastructure, megawatts, Digital Gateway, zoning risk, power delivery, utility contracts, project finance, Digital Realty, Blackstone, investment risk
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Arizona Retirees Have $556 Million Tied to Data Centers - and the Risk Is Getting Harder to Ignore
Publication date: 2026-09-11
Contributors: Ogdn Ames
Abstract
This information paper examines the Arizona State Retirement System reported $556.2 million exposure to a data-center investment strategy amid rapid infrastructure growth, constrained power availability, financing risk, and construction uncertainty. It explains how leverage can amplify equity losses, provides clearly labeled stress-test illustrations, and recommends improved disclosure, realistic downside testing, and stronger underwriting protections for retirement-system beneficiaries.
Keywords: Arizona, data centers, AI infrastructure, Arizona State Retirement System, ASRS, retirement assets, institutional investing, power constraints, leverage, construction risk, Menlo Equities, risk management
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
A $300 Million-Plus California Data Center Was Finished Before Its Power Path Was
Publication date: 2026-09-10
Contributors: Ogdn Ames
Abstract
This analysis examines a Santa Clara data-center project where more than $300 million was tied up in a completed building, land, and interest expense while utility power remained uncertain. It explains how a revised capacity path to 77 MVA through July 2031, added load-development fees, curtailment rights, and changed legal protections can affect lease-up timing, carrying costs, and investment risk. The reported project costs and leasing-delay figures are identified as company claims or illustrative scenarios, not adjudicated losses.
Keywords: California, Santa Clara, data centers, AI infrastructure, power risk, utility capacity, lease-up risk, project finance
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Kansas Farmers Have $284 Million at Risk — And $6 Diesel Could Turn Harvest Into a Cash-Flow Squeeze
Publication date: 2026-09-10
Contributors: Ogdn Ames, Michael Schuler
Abstract
This joint research brief estimates that Kansas farms could face roughly $189.5 million to $379.0 million in additional annualized fuel costs if Midwest diesel prices remain near early-September levels. The 75% base case implies approximately $284.3 million of gross annualized cost exposure as harvest begins. The analysis reconstructs diesel-equivalent fuel volume from USDA agricultural fuel and oil purchases, tests low, base, and high exposure scenarios, and examines the contribution of crude oil and refined-diesel market conditions. The estimates describe cash-flow exposure scenarios rather than realized losses or predictions of farm profitability.
Keywords: Kansas agriculture, diesel prices, harvest cash flow, fuel cost exposure, energy markets, rural businesses
Published by Ames Investment Systems
License: All rights reserved
Permission is required for reuse beyond applicable legal exceptions.
Research discipline
A deliberate path from research to live trading.
Explicit development periods and separate evaluation stages help assess whether an investment idea holds up beyond the data used to build it.
- 01
Develop
Historical research with an explicit development period.
- 02
Out-of-sample
Evaluate unseen data without tuning to the evaluation period.
- 03
Final holdout validation
Reserve a final blind subset for deployment readiness.
- 04
Live incubation
Review actual fills, slippage, drawdowns and signal stability.
- 05
Deploy & monitor
Scale gradually, monitor drift and re-estimate capacity.
Evidence hierarchy
Know what each result represents.
Historical managed-account composite
Multiple accounts and multiple strategy regimes, including discontinued strategies. Not the current macro strategy alone.
Out-of-sample research
Evaluation on unseen historical data. Research evidence, not live trading.
In-sample development
The model-development period. Context for the research process, not a live record.
Modeled / simulated results
Hypothetical scenario and risk estimates, dependent on assumptions. Actual outcomes may differ materially.
Monitoring
Validation continues after deployment.
Execution and slippage review, signal monitoring, gradual scaling and capacity re-estimation inform ongoing oversight. Historical regime shocks and scenario simulations supplement this process.
Past, live, incubated, historical managed-account composite, backtested, out-of-sample, hypothetical, simulated and modeled results are not indicative of future results. These evidence categories are distinct. The historical managed-account composite spans multiple accounts and strategy regimes and is not the current macro strategy alone. Objectives, targets and estimates are not guarantees, promises or predictions. Assumptions may prove incorrect, and models cannot capture every structural break, liquidity crisis, operational failure or nonlinear event.
Allocator diligence
Start with the evidence.
Discuss the strategy, research process and operational readiness with AIS.