LafataCorp

An automated trading agent deployed on Kalshi event contract markets. The system uses a proprietary position-sizing model with data-driven entry criteria across multiple market categories. The agent was live from March 17 through June 22, 2026 before being paused for retooling.

The agent's early performance was driven by two factors: a high-conviction currency strategy (93.8% win rate) and disciplined position sizing. Both eroded over time. Currency trades stopped executing in late April due to a configuration issue, removing the strategy's highest-edge category entirely. Without currency exposure, the remaining index trades were unable to sustain profitability on their own.

The index strategy underperformed during a period of sustained directional momentum in equity markets -- a regime where the system's core assumptions were consistently violated. What had been reliable setups in range-bound conditions broke down as the market trended strongly in one direction for weeks.

Position sizing compounded the damage. Average position size grew well beyond the original parameters, and since every losing trade settled at a total loss, a single oversized position could erase the gains from 8-10 winning trades.

The system is now paused while three improvements are implemented: restoring currency trade execution, adding a market-regime filter to reduce exposure during trending conditions, and enforcing hard position-size caps. The core model remains sound -- the failures were operational and environmental, not structural.

A live in-game trading system built on a proprietary model trained extensively on historical NHL data. The system identifies high-confidence situations during games and executes only when both model output and market conditions meet a defined threshold. Paper traded across the full 2025-26 NHL regular season.

A pre-game model that builds edge predictions from a proprietary combination of advanced hockey metrics, scheduling context, and historical matchup data. The system was profitable in three of its four active months during the 2025-26 season. January, February, and April all produced positive returns, with February posting a 73% win rate -- the highest of any month across all LafataCorp systems.

March was the sole losing month, and the cause was specific and well-understood. The NHL's mid-season Olympic break created a regime shift: teams returning from the break rested key players, managed workloads ahead of the playoff push, and disrupted the head-to-head matchup patterns the model relied on. The model's H2H factor went from 83% accuracy pre-break to 42% post-break. Critically, the position-sizing model compounded the damage by increasing stake sizes as the bankroll had peaked entering March.

Trading was not paused during this period. The decision to continue tracking through the downturn was intentional -- documenting how the model fails under regime change produced the data needed to fix it.

Three targeted improvements are being implemented for the upcoming season based on lessons from the March drawdown. First, stake sizing will be capped at a flat amount after February 15 regardless of bankroll -- eliminating the compounding effect that amplified March losses. Second, head-to-head records will decay after February 1, using only current-season data to avoid stale matchup signals. Third, a playoff-contention filter will skip picks involving teams that have already clinched or been eliminated, removing the low-effort games that degraded model accuracy. The system resumes with the 2026-27 NHL season in October.

This section tracks Cam's personal event contract trading on Robinhood, benchmarking manual trading intuition against the automated systems. Over 10 months and 245+ bets, the portfolio doubled its deposited capital, with 7 of 9 active months producing positive returns.

The data tells a clear story about where the edge lives: NHL. Hockey contracts represent the vast majority of trading volume and nearly all of the portfolio's net positive performance. Every other sport -- NFL, NBA, MLB, NCAA -- was either breakeven or net negative. This isn't surprising given Cam's deep background in hockey analytics, but it's a useful signal: the manual trading edge is sport-specific, not generalizable.

Trading is paused until the 2026-27 NHL season begins in October. The offseason offers no hockey markets to trade, and the data strongly suggests that non-hockey event contracts are not worth the capital risk. When trading resumes, the focus will be exclusively on NHL markets.

Profits from event contract trading are swept into a long-term investment portfolio on a bi-weekly schedule. The portfolio is diversified across index ETFs, tech positions, and speculative plays. 67% of total portfolio funding has been sourced directly from prediction market returns, with the remaining 33% from direct deposits. Individual holdings are not disclosed.