The Operation
APEX currently publishes MLB first-five quantitative forecasts built around disciplined market coverage, flat-unit staking, and transparent model scope.
APEX is not a pick-and-choose tout sheet. Where a market is designated as bet-all, every eligible game is modeled, published, and graded. Where a market is designated as filtered, the filter is part of the model design and is disclosed directly. The objective is consistency: clear model boundaries, repeatable output, and a record that can be evaluated without hidden selectivity.
MLB — F5 Run Line · F5 Totals
F5 Run Line — Bet-All
APEX publishes one first-five-inning run-line position for every eligible MLB game using the current production spread engine, ATS-ELITE-C1-CONDITIONAL-DISTRIBUTION-RESIDUAL.
The engine evaluates the actual posted FanDuel first-five spread and the corresponding price available on each side of that market. The published position is therefore based on the exact F5 spread being offered rather than a moneyline proxy or a generic favorite/underdog classification.
Every eligible MLB game receives one F5 run-line position. The system does not selectively remove games after scoring and does not publish MLB moneyline positions.
The model combines current starting-pitcher state, offensive and lineup information, handedness and matchup context, park and game-environment information, and the authentic pregame market available at issuance. Its final output is a calibrated probability for the posted F5 spread.
F5 Totals — Bet-All
APEX also publishes one first-five total position for every eligible MLB game using the current production totals engine, TOTALS_FINAL_NESTED_JOINT_R4_V1_20260825.
The totals engine is a complete probability-distribution model rather than a simple over/under classifier. It produces an exact discrete probability distribution across possible first-five run totals before the sportsbook line is applied.
That structure preserves the probability of every relevant scoring outcome, including exact push probability when FanDuel posts an integer F5 total. Half-run totals are evaluated directly from the same complete run distribution.
The final over or under position is selected only after the model's full first-five scoring distribution has been produced.
The engine incorporates the current starting-pitcher matchup, offensive environment, lineup and platoon context, park characteristics, roof and weather information where relevant, schedule and game context, and the authentic FanDuel F5 total and prices available at issuance.
Model Framework
APEX MLB is a two-engine pregame forecasting system built specifically for first-five-inning markets.
The two production engines are intentionally separate:
Run Line: ATS-ELITE-C1-CONDITIONAL-DISTRIBUTION-RESIDUAL
Totals: TOTALS_FINAL_NESTED_JOINT_R4_V1_20260825
The run-line engine estimates the probability of covering the actual posted F5 spread.
The totals engine estimates the complete probability distribution of first-five runs and then evaluates the posted F5 total from that distribution.
Both engines use point-in-time pregame information. Player, team, lineup, market and contextual information must be available before the relevant game begins. Later information is not substituted into historical model inputs.
Model development and comparison preserve chronological ordering. Candidate methods are evaluated against the existing production engine on the same historical games using probability-based scoring, calibration, season stability and reproducibility. A production engine is not replaced merely because another method is newer or more complex.
The current engines remain the production models unless a future challenger is separately tested, independently reproduced, demonstrates measurable improvement, and is manually approved for adoption.
Production Process
The public MLB forecast is produced through one daily production process.
Current factual player information, team and lineup information, historical outcomes and authentic market information feed the daily MLB process. The system refreshes the game state before scoring, captures the current FanDuel first-five markets, runs both production engines, and publishes one final slate.
Every published position is preserved with its market, line and price so that the following day's grader evaluates the exact position that was originally issued.
Research and experimental model testing are maintained separately from daily production. Experimental signals or mathematical methods cannot automatically replace either production engine.
Market Philosophy
APEX does not use public betting percentages, sharp-money labels, steam-chasing signals or reverse-line-movement narratives as substitutes for matchup forecasting.
The FanDuel market is used where it is mathematically relevant to the actual wagering decision: the real posted F5 line and the authentic prices available on both sides of that market.
APEX therefore does not attempt to ignore the market, nor does it simply follow it. The objective is to combine an independently modeled view of the baseball matchup with the exact market that must ultimately be evaluated.
Closing-line movement and subsequent market behavior may be studied after publication, but they do not retroactively alter the original APEX position.
Conviction Framework
APEX uses four public conviction tiers:
Stake sizing does not change across tiers. Public conviction describes model strength, not a recommendation to increase bet size.
Operating Principles
APEX is built around five operating principles:
The goal is not to make every position look equally strong. The goal is to publish the model honestly, preserve the record, and let the results compound over time.
Responsible Use
APEX publishes quantitative sports forecasts for informational and research purposes. No model guarantees outcomes, and all betting involves risk. The published slate should be evaluated as a disciplined forecasting product, not as a promise of profit.