Every game on this board gets the same three numbers: a win probability, a projected game total, and a projected run count for each club — all from one model, all posted beside the market's price, all graded against the real final score. Most sites in this space publish moneyline picks and let the losses disappear. We publish the model's whole team-level view and the ledger that scores it, because where the model beats the book and where it doesn't is the actual information.
How the run projections work
Everything on this board starts as runs. We project each club's scoring from the lineup it's actually running against the starter it's actually facing — handedness splits, what that starter and the bullpen behind him tend to allow — with the park and the day's weather folded in. One run projection per side. Sum them and you have our game total. Set them against each other and you have a win probability. The team totals you see are those same two numbers, each measured against its own posted line.
One model, three markets — which is the point. When the win probability and the team-total lean agree, you're seeing the same signal from two angles, not two opinions. The exact weighting is ours and re-tunes weekly against real results (how MatchWiz works).
Why we don't sell moneyline picks
The moneyline is the sharpest market in baseball. It carries the most money, the most attention, and the fastest corrections — by first pitch, the price has already digested nearly everything a good model knows. Our season grade shows exactly that: the model's side has won 53% of 476 graded games — right about where the prices said it should, which lands near break-even once the vig is paid. Anyone claiming a big, durable moneyline edge is describing a market that doesn't exist.
So we treat the moneyline and the game total as transparency: the model's number, the market's number, and the season record between them. That break-even grade isn't a failure — it's calibration proof. A model that can price the sharpest market correctly is a model you can trust in the markets where the pricing gets soft.
Team totals — where the room is
A team total is the over/under on one club's runs, not the game's. It's a corner market: less money, less attention, lines that get set off the game total and left alone. Our run projections are built per-club from the ground up, so this is where the model's work has somewhere to go — when our projected runs gap the posted line by enough, that's a lean worth flagging.
The strongest of those leans — the ones that clear our backtested edge threshold — get surfaced as plays, and they carry their own graded record:
Qualifying team-total leans post to the Plays of the Day board once lineups are final, sized 1–3 units by the size of the gap.
Graded at the price you could have taken
Every lean on this board grades against the line and price that were actually posted when we captured them — not the closing number, not a cleaner price that showed up later. If a team total surfaced at over 4.5 at −115, it grades at over 4.5 at −115. Each game's numbers lock at first pitch, and the archive keeps every past date one click away. What we said, what the book offered, what happened — that's the whole system.
How to actually use it
- Read the gap, not the pick. Convert your book's moneyline to an implied probability (the FAQ below shows the math) and set it against ours. Small gaps mean the market already knows. Big gaps mean someone's wrong — and the season record tells you how often it's been us.
- Let the plays do the filtering. If you only want the spots that clear the backtested bar, the Plays board already applies the threshold, the lineup gate, and the unit sizing.
- Use the totals as context. The projected game total tells you what kind of night the model expects; the NRFI board zooms the same run-scoring math into the first inning.
- Audit us first. Roll the archive back a few weeks before you weight anything here. The record is the product.
The traps
- Betting good teams instead of good prices. The best club in baseball is a bad bet at the wrong number. Every edge on this board is a price disagreement, never a team endorsement.
- Parlaying "safe" favorites. Chaining short-priced legs compounds the book's margin on every leg. The parlay price is always worse than the legs deserve.
- Treating a projection as a script. A 58% win probability loses 42% of the time — that's not the model failing, that's baseball. Judge the process on the season ledger, not on last night.
- Ignoring the vig on totals. Two juiced sides of a total can both be losing bets. Fair-price the market before you like either side.
MLB moneyline & team totals — FAQ
How accurate is MatchWiz's MLB moneyline and run-total model?
We grade every game against the result and post the record right on the board. Moneylines and game totals are efficient markets, so the model grades near break-even there and we treat those as transparency, not plays. Team totals are a softer market — that's where it's finding a real edge.
What's a team total, and why does the model bet it?
A team total is the over/under on one club's runs, not the whole game. It's a softer, less-efficient market than the moneyline, so our run projections find more room there. When our projected runs gap the line enough, we post it as a play with recommended units.
How does the run-total (over/under) projection work?
We project each side's runs from its lineup against the starter and bullpen, plus park and weather, then sum both sides for the game total and set it beside the market line. Over or under is just which side of that line we land on. The exact weighting is proprietary and tuned weekly.
Are these MLB betting picks?
No. Every number here is a graded research signal from our model, shown against the market and honest about its record. Not betting advice, and no outcome is guaranteed. 21+, please play responsibly (1-800-GAMBLER).
How do I compare a win probability to a moneyline?
Convert the moneyline to an implied probability: for a negative line, odds ÷ (odds + 100) — so −150 implies 60%; for a positive line, 100 ÷ (odds + 100) — so +130 implies about 43.5%. Both sides of a game add up to more than 100% because the book's margin (the vig) is baked in. Compare our posted win probability to that implied number and you can see exactly where the model and the market disagree.
Why does the model grade near break-even on moneylines?
Because the moneyline is the sharpest, most liquid market in baseball — the price already reflects nearly everything a good model knows. Grading near break-even against that market is what an honest, well-calibrated model looks like. We publish the record anyway, because the win probabilities are still useful context — and because the same run projections find real room in the softer team-total market.
Win probabilities, projected totals, and grades are model output, updated daily and scored against the real final at the posted price. Research signals, not betting advice — 21+, play responsibly (1-800-GAMBLER).