Cattle Crushers

Inspiration

Inspired by Ray Dalio forecasting chicken prices for McDonald's. Cattle markets connect a biological production cycle with financial markets. When feeding margins deteriorate, feedlots may place fewer cattle, potentially tightening future supply. We wanted to know whether that economic mechanism creates a predictable trading opportunity, or whether futures prices already incorporate the information.

Cattle Crushers began with that question, rather than a search for the best-looking backtest.

What it does

Cattle Crushers tests two systematic trading hypotheses:

Cattle futures: Can a feeding-margin signal predict placements and profitable deferred Live Cattle futures returns?
Beef-exposed equities: Can cattle price movements predict sector-relative returns in Tyson Foods and Texas Roadhouse?

The project combines economic tests, transaction-cost-aware backtests, a once-evaluated holdout, and liquidity and funding analysis.

Both primary strategies failed their registered acceptance criteria. The primary cattle strategy achieved a positive holdout Sharpe of 0.63, but that did not override its failed development conditions. Our submission explains these results and their limitations rather than presenting an unsupported trading edge.

How we built it

We built the research pipeline in Python using pandas, NumPy and statsmodels. Our data sources included:

CME futures data through Databento
USDA cattle placements
CFTC positioning reports
Yahoo Finance equity prices

The futures study constructed a projected feeding-margin proxy from live cattle, feeder cattle and corn prices. Signals used trailing information, with trades modeled at the following trading session's settlement. We accounted for contract selection, rolls, transaction costs and specified execution delays.

After the cattle hypothesis failed, we developed a separate equity follow-up using TSN/XLP and TXRH/XLY pairs. We disclosed that sequence and preserved both primary specifications.

We also packaged an artifact-only verifier that checks paper claims against saved results, verifies protected-file hashes and confirms the registered rejection decisions. It allows reviewers to check submission consistency without acquiring market data; it does not replace backtest reproduction.

Challenges we ran into

The hardest work involved deciding what the data could actually support.

Data gaps: Missing settlements and session ranges complicated execution assumptions. Some feeder-cattle contracts lacked cash-final records.
Incomplete funding data: Historical margin requirements and equity borrow availability were incomplete, while settlement-price fills and fractional futures limited practical interpretation.
Careful distinctions: We had to distinguish initial-capital returns from current-NAV leverage, modeled collateral from available cash, and participation-based capacity illustrations from profitable capacity.
Research chronology: We documented later implementation choices and kept post-evaluation corrections and diagnostics separate from the frozen hypothesis decisions.

Accomplishments that we're proud of

Preserving the original acceptance criteria even when later results looked more encouraging.
Connecting economic hypotheses to explicit statistical tests and trading rules.
Reporting unsuccessful results, execution uncertainty and incomplete checks openly.
Delivering a concise research note with traceable saved evidence.
Building an artifact verifier supported by 22 focused tests.
Developing a prospective risk-governance assessment that distinguishes proposed procedures from demonstrated protection.

What we learned

A plausible economic mechanism is not enough to establish a trading edge. The mechanism must be measurable, predictive at the relevant horizon and tradable after costs.
A positive holdout result cannot rescue a strategy that fails its registered development requirements. Conversely, rejecting our particular rules does not prove that no economically useful relationship exists.
Risk management requires more than a drawdown chart: exposure, execution, collateral eligibility and cash-payment timing answer different questions.

What's next for Cattle Crushers

Our next steps would focus on better evidence:

Calibrating the feeding-margin proxy against published feedlot economics
Strengthening execution assumptions
Incorporating historical funding and borrow information
Measuring equity volume capacity

Any revised strategy or risk control would be specified before evaluation on fresh data. The current submission remains a documented research result, not a claim of a deployment-ready trading system.