Telecommut
- SPY
Public work on the S&P 500 ETF under a shared standard: causal features, walk-forward evaluation, baselines, and a willingness to stop when the edge is not there. The live production line is short-horizon realized volatility. Experiments that fail those standards are documented as null results and stay out of production, including future work that does not clear the bar.
Vision
- proposed SPY suite
End goal: a unified SPY market-state system built from modular models that each answer a different question (how large is next-horizon vol? is the session range expanding? which regime? which way?), sharing data and ops discipline, shown together with honest track records. Not one black-box “all of SPY” model.
What “suite” means: one data spine (SPY OHLCV, later bars; optional VIX), one ops spine (registry, candidate vs active, promotion gates, health, ledgers), one product surface (panels for vol, range, regime, direction, track). Shared culture: walk-forward only, label-native baselines, freeze vs promote separated, clean fails published.
What it does not mean: the same head predicting vol and direction; reopening the live vol freeze to chase a sibling problem; or treating research pass as live promote.
- Vol (live freeze) - next-5d annualized realized vol vs naive / GARCH; weekday pipeline and public board.
- Stability / promotion - PSI tiers, golden master, shadow before any active model flip (no auto-unfreeze).
- Range expansion (research) - next-session high-low vs recent ranges; own design and scorecard; no live promote until freeze pass/fail.
- Regime (planned) - low / normal / high vol or trend vs chop labels.
- Direction (documented null for open-to-close) - parked; not revived without a new design.
- SPYForge (ops verify) - product-law checks on pipeline / gates / write isolation; not a forecast.
- Unified app (end product) - one place for modules with shared freshness and honest status cards.
Live
- production research board
Walk-forward forecasts of next-five-trading-day annualized realized volatility for SPY, compared to naive persistence and GARCH(1,1). Architecture is frozen (monthly LightGBM, nine baseline features). Monthly weight updates, when run, are gated; the modeling ladder is not reopened for routine ops. This is the system that remains active.
Open vol forecast board Methodology PDF
Infrastructure
- SPYForge
GraphForge is a standalone product (any workflow): encode product law as typed state graphs and verify pipelines. SPYForge is the folder inside the SPY research repo that applies it here (freeze, candidate-only retrain, promote gates, agent ACL). Simulate board verifies that application. Not forecast skill. Production stays on vol pipeline scripts until an explicit cutover.
Open verification board SPYForge technical note (PDF)
Parked
- research reference only
We tested a strict causal daily-bar model that predicts whether SPY will close higher than it opened, using only information available at the open. Out-of-sample (2019-2026) neither logistic nor tree models beat the always-up baseline on accuracy or Brier score. A weak ranking of open-to-close returns exists, but it does not survive realistic transaction costs.
Null-result write-up
Negative results are part of the research process. The same evaluation standards apply across experiments: causal features, baselines, walk-forward evaluation, and an explicit stop when the bar is not cleared.
Martial Systems LLC
- SPY research suite
- not investment advice.
Experimental research summary. Historical forecast error is not a guarantee of future accuracy and does not constitute investment advice, a solicitation to buy or sell any security or product, or a performance track record for live trading. Not a trading signal service. Past results do not predict future results. You can lose money.
Trademarks / no affiliation. Martial Systems LLC is not affiliated with, endorsed by, sponsored by, or partnered with State Street Global Advisors, SPDR, SPY, the S&P 500, S&P Dow Jones Indices, any securities exchange, or any index provider or ETF issuer. Names and tickers are used only to describe publicly discussed instruments for non-commercial research. All trademarks are the property of their respective owners.
Data sources. Research inputs were obtained from publicly available market data and public APIs for offline analysis. This site does not redistribute raw proprietary market-data feeds, full historical dumps, or a commercial data product. Third-party data may be delayed, incomplete, restated, or erroneous; we do not warrant completeness or accuracy.
Martial Systems LLC is an independent research company and is not affiliated with Kalshi Inc. or its affiliates. Kalshi and related marks are trademarks of their respective owners.
PROPRIETARY AND CONFIDENTIAL. Copyright (c) 2026 Martial Systems LLC. All rights reserved. Unauthorized copying or redistribution of site software or materials is prohibited except as permitted by a separate written agreement.
Copyright (c) 2026 Martial Systems LLC. All rights reserved.
