Frequently Asked QuestionsFAQ

Ballpark Pal is an MLB research platform that simulates every game 3,000 times before first pitch. The simulations produce detailed projections, probabilities, and matchup ratings that cover just about every angle of each day's slate - from park factors and weather effects to player props and DFS projections.

Full access to everything on the site. That includes game simulations, daily park factors, matchup ratings, starting pitcher projections, player profiles, the Matchup Machine, Home Run Zone, Strikeout Center, Hits and Bases pages, First Inning and First Five Innings tools, the Odds Screen, Positive EV tool, Parlay Calculator, PrizePicks & Underdog projections, DFS projections, Stacks, lineup tools, the Export Center, and more. There are no tiers or feature locks - one subscription gets you all of it.

Absolutely. You don't need to be a baseball encyclopedia to use the site effectively. Ballpark Pal distills complex factors like stadium dimensions, weather, and pitching matchups into simple numbers you can act on. The park factors page, for example, tells you in plain percentages how much the conditions favor or hurt home runs for a given game. You don't need to know the fence distances at Yankee Stadium or what a launch angle is - the models handle that for you.

No. Ballpark Pal is a research platform, not a picks service. It provides simulation-based probabilities and data tools to help you make your own decisions.

No. Ballpark Pal doesn't take bets, sell picks, or have any affiliation with sportsbooks. You're the customer, not the product.

There are two options: $10 per month or $60 for the full season. The full-season pass is the better value if you know you'll be using the site through the playoffs. Both plans give you the same full access to everything.

Yes. Monthly subscriptions automatically renew each month on the date of your initial signup. You can cancel at any time and you'll keep access through the end of your current billing cycle.

Yes. There are no contracts and no cancellation fees. If you cancel, you keep access until the end of the period you've already paid for.

When you're logged in, click "Subscription" in the top bar. This will take you to the billing portal where you can view, update, or cancel your subscription.

Payments are processed through Whop, which supports credit cards, debit cards, and other standard online payment methods.

Subscription payments and billing are handled through Whop. For any billing-related issues please visit Whop Support where you can chat with a support agent.

Trial availability can vary. The best way to check is to visit the signup page to see current options. Regardless, you can cancel your subscription at any time if it's not for you.

Ballpark Pal is an in-season product with the subscription season running through the end of October. You won't be charged during the offseason as projections and daily data aren't produced without any games.

Yes. Subscriptions from previous seasons will not carry over to 2026, so you will need to create a new account and subscribe for the current year. You'll complete everything directly through the Ballpark Pal site - no need to leave the platform.

Login credentials from previous seasons do not carry over. Ballpark Pal accounts are now native to the site itself, so you'll need to create a new account and subscribe for the current season. The signup process takes place entirely on the Ballpark Pal site.

You can log in with your email and password at the login page. There's also a passwordless option - enter your email and you'll receive a 6-digit code to log in without needing your password. The code expires after 10 minutes.

Click "Forgot Password" on the login page, enter your email, and you'll receive a reset link. The link expires in 60 minutes. Your new password needs to be at least 8 characters with at least one uppercase letter, one lowercase letter, and one number.

Yes. The My Preferences page (accessible from the top bar when logged in) lets you configure default views for various pages, choose between letter grades and numeric scoring, set your preferred sportsbooks (and their display order), and adjust other display settings like play log formats, split views, and column visibility.

Go to My Preferences and you'll find a sportsbook ranking section. You can pick up to 6 books and arrange them in your preferred order. These preferences carry across the Odds Screen, Positive EV, Home Run Zone, Player Props, and other tools that display odds.

For billing and payment questions, visit Whop Support - they have an FAQ and live chat. For questions about the Ballpark Pal website itself, email BallparkPal.site@gmail.com and someone will get back to you.

The Methods page has a detailed breakdown of the entire projection system - expectation models, park factor methodology, the matchup model, and how game simulations work. It's a thorough read but well worth it if you want to understand what's happening under the hood.

Ballpark Pal projections are generated through a system of models. At a high level, the approach focuses on separating player abilities from outside influences like stadium variation, weather, and batted ball luck - largely through machine learning and simulations. The projections come from a proprietary game simulator that pre-executes each game thousands of times. You can read the full breakdown on the Methods page.

No. Ballpark Pal has always used machine learning to make predictions. Model inputs and outputs will never be polluted with AI language models, which are unreliable for projections.

Large Language Models can easily produce data and information that appears reasonable, but they don't model underlying systems or generate results from structured statistical processes. Their outputs are not constrained by real-world distributions, making them unsuitable for quantitative forecasting.

Ballpark Pal projections are built on deterministic models trained on historical data, with explicit assumptions, validated inputs, and measurable performance - rather than generated text that only looks correct.

Each game is simulated 3,000 times. In every iteration, a full game is played out plate appearance by plate appearance. For each at-bat, a probability model (the Matchup Model) assigns likelihoods for every possible outcome - single, double, home run, walk, strikeout, etc. A random number is generated and mapped against those probabilities to determine what happens. The sim also models baserunning events like stolen bases. After all 3,000 iterations are done, the results are aggregated into the projections and probability distributions you see on the site.

Traditional "top-down" models predict one thing at a time - say, total runs or whether a guy hits a homer. The problem is they're rigid and can't account for unexpected changes like a lineup shuffle or a key player being scratched. Ballpark Pal's "bottom-up" simulation approach builds the entire game from scratch, play by play. This naturally captures things that are hard to model explicitly - like home teams having fewer plate appearances, or a weak bottom of the order limiting run-scoring opportunities. One simulation handles every market in a single pass: runs, hits, home runs, strikeouts, YRFI, F5, player props - all of it.

Each game's projections update whenever the most recent simulation for that game completes. Generally, each game simulates at least once per day. Once official lineups are posted, a final sim runs with those lineups accounted for. You can see the "Last Updated" time on relevant pages.

Each simulation runs a finite number of iterations (3,000), so there's a small amount of natural randomness between runs. Think of it like flipping a coin 3,000 times - you'll get close to 50/50 every time, but not exactly the same number. Differences can also come from updated inputs like weather forecasts or recent player performance data.

Simulations still run using projected lineups. Once the official lineups come out (usually a few hours before game time), the sim re-runs with the confirmed batting orders and any late scratches or additions.

Yes. Park factors are built directly into every game simulation. Each plate appearance in the sim uses batter-specific park and weather adjustments to modify outcome probabilities, so the effects of the stadium and conditions are reflected in all projections, probabilities, and derived stats across the site.

Ballpark Pal park factors start at the individual batter level. For each hitter, the system establishes a baseline expected performance using just the physical characteristics of batted balls (exit velocity, launch angle, spray direction) without considering the stadium. Then it runs a second model that adds in the specific park dimensions and the day's weather forecast. The difference between the two models is the park factor. This means park factors are personalized - a pull-heavy lefty will be affected differently than a guy who sprays the ball the other way. Game-level park factors are then calculated by aggregating batter-level effects across both lineups, weighted by expected plate appearances.

Park factors are measured against the MLB average. So a +10% for home runs means that the combination of park and weather is expected to produce 10% more home runs compared to an average environment.

Yes. Weather is central to the park factors model. Temperature, humidity, wind speed, wind direction, and barometric pressure all factor in. The system uses hourly forecasts for each game so it can account for conditions changing during the game. For retractable-roof stadiums, it also considers whether the roof is expected to be open or closed.

Park factors reflect the expected environmental impact on the small subset of plays that are susceptible to conditions - mostly fly balls near the fence. A 450-foot bomb is a home run in every stadium. A game with a bunch of pull-side line drives won't be affected much by wind. Park factors are real, but they're secondary effects that show up over many games. On any single day, the skill level and randomness of the players involved will usually be a bigger factor.

Combined shows the full park factor including both the stadium's structural characteristics and the day's weather conditions. Stadium Only isolates the effect of the park itself (dimensions, altitude, outfield configuration) without weather. Weather Only isolates the impact of the day's specific conditions (wind, temperature, humidity) without the park structure. This breakdown helps you see whether a favorable park factor is driven by the venue itself or just a particularly good weather day.

The same weather conditions can play very differently depending on the stadium. For example, a 15 mph wind blowing out at Wrigley Field is a significant factor - the open grandstand design lets wind flow right through the park and carry fly balls well beyond the fence. That same 15 mph wind at Citi Field has far less impact because the tall stadium walls block most of it before it reaches the field. The model captures these interactions between weather and each park's unique structure - which stadiums are wind-sensitive, how altitude changes ball flight, how temperature and humidity affect carry depending on field dimensions, and how all of that varies based on the direction the ball is hit.

This is one of the most common park factor questions we get. The weather forecast at Oracle Park almost always shows wind blowing out, often at double-digit speeds - which sounds like it should help home runs. But the reality is that the wind at Oracle Park is highly unpredictable. The forecast direction and what actually happens on the field during the game don't line up consistently. The wind changes direction rapidly and unpredictably throughout the game, which means the "blowing out" forecast doesn't give you reliable information about what hitters will actually experience at the plate. On top of that, the fence is deep through the right-center gap and most of left field, and the cool marine air suppresses carry. The result is that Oracle Park consistently grades as one of the bottom-five parks for home runs each season despite that always-out forecast.

Each metric shows how the park and weather conditions affect that specific type of outcome. HR is the effect on home runs. 2B/3B covers extra-base hits (doubles and triples). 1B covers singles. Runs is the overall effect on run scoring. A park can be great for home runs but bad for singles, or vice versa, depending on the outfield layout and conditions.

The matchup model is a machine learning system trained on over 1 million plate appearances since 2016. It uses more than 100 proprietary features to estimate a full probability distribution of outcomes for any batter/pitcher combination - the likelihood of a single, double, home run, walk, strikeout, and so on. The inputs aren't traditional stats you'd find on FanGraphs. They're engineered signals derived from expectation models and pitch-level data that capture how different batter and pitcher traits interact.

The Matchup Machine lets you look up any batter vs. any pitcher - not just today's scheduled matchups. You can explore the expected outcome distribution, release point data, pitch types, contact quality, and more.

Every metric on player pages is standardized to a 0–125 scale relative to all active MLB players (including non-starters). A score of 50 is roughly average, and scores above 100 are reserved for the very best. Here's the full breakdown:

• 101–125 - Best of the Elite (less than 2% of players)
• 81–100 - Top Tier (6%)
• 67–80 - Excellent (10%)
• 56–66 - Above Average (13%)
• 46–55 - Average (14%)
• 32–45 - Below Average (25%)
• 20–31 - Poor (16%)
• 0–19 - Bottom Tier (14%)

You'll notice there are more players rated below average than above. That's not a quirk of the scale - it reflects how talent is actually distributed across the league. A handful of elite players separate themselves significantly from the pack, while the majority of roster spots are filled by players who cluster closer together. This is especially true when you factor in non-starters and bench players who see limited action.

One of the key benefits of this standardized scale is that it puts completely different metrics on the same playing field. When you're looking at a trends graph with swing speed, contact, zone control, and power all plotted together, they're all speaking the same language. A score of 70 means the same relative standing whether it's measuring chase rate or exit velocity.

These ratings are based on rolling pitch count windows, not fixed calendar periods. By default, metrics reflect the last 500 pitches seen (for batters) or thrown (for pitchers), but you can adjust this on the page to 50, 100, 250, 500, or 1,000 pitches using the dropdown in the controls. If you're logged in, you can also set your preferred default pitch count in My Preferences so it carries across sessions.

View the full Ratings Key

Letter grades are directly derived from the 0–125 numeric ratings - they're just a different way to display the same underlying data. Some people find it easier to quickly scan a page full of letter grades rather than numbers. You can toggle between numeric scores and letter grades using the Scores/Grades buttons on any player page, and if you're logged in you can set your preferred default in My Preferences so the site always loads the way you like it.

All probabilities on Ballpark Pal are derived directly from the game simulations. Since each game is simulated 3,000 times and every plate appearance within each iteration is tracked, the system can count how often any given outcome occurred. If a player recorded at least one hit in 2,100 out of 3,000 simulations, that's a 70% probability. This applies to everything - home runs, strikeouts, total runs, first-inning scoring, win probability, and so on. Nothing is estimated with a separate formula; the probabilities are observed frequencies from the simulations.

Ballpark Pal's simulations are completely independent from the sportsbooks. Book odds are not used as inputs to the model in any way - the probabilities come entirely from the simulation engine. Sportsbooks tend to cluster together because they're operating in the same market, adjusting to each other and to the money coming in. Ballpark Pal doesn't have that gravitational pull toward the herd. On any given day the site produces thousands of individual probability estimates across players and markets, so it's not unusual to see a handful of cases where the sim odds and the book odds are significantly apart. That's a natural byproduct of running an independent model.

Sportsbook odds are influenced by each other, by the money coming in, and by the need to balance risk. That's why they tend to cluster together - they're all reacting to the same market forces. Ballpark Pal's probabilities are generated entirely on its own with no awareness of what any book is showing. The reason that's important is that independent sources of information add incremental predictive value. When multiple forecasts are driven by the same inputs, they mostly repeat the same signal. But when a model is built independently, it can capture different information and different relationships in the data. That means it has the potential to improve overall predictions when combined with market odds or other models, rather than simply echoing them. In other words, it's about contributing new signal instead of recycled consensus.

The Odds Screen displays Ballpark Pal's simulation-derived odds alongside live odds from a variety of books, viewable across different stats and markets. The page offers several views: raw odds, probabilities, and matchup-based views. It makes it easy to see where the simulation odds and book odds line up and where they diverge.

The Positive EV page filters for outcomes where the Ballpark Pal simulation probability is higher than the implied probability from the sportsbook odds. You can filter by your preferred books and browse across markets.

The Parlay Calculator estimates the combined probability of multiple outcomes happening together. Since the simulations track every outcome at the iteration level, it accounts for correlation between outcomes in the same game. For example, if a pitcher gives up a lot of hits in a simulation iteration, that same iteration is also more likely to produce more runs and total bases. Standard parlay math assumes outcomes are independent and just multiplies probabilities together - the Parlay Calculator uses the actual simulation data to give a more realistic combined probability.

YRFI stands for "Yes Run First Inning" and NRFI stands for "No Run First Inning" - whether or not either team scores in the first inning. Because the simulations play out every inning of every game, first-inning run probabilities come naturally from the data. You'll find YRFI/NRFI numbers on the Game Simulations and First Inning pages.

The First Five Innings (F5) view isolates the first five innings of each game, which effectively covers only the starting pitchers and removes bullpen variability. Ballpark Pal breaks out F5 runs, totals, and moneylines separately from full-game numbers on the First Five Innings page and within individual game pages.

The Home Run Zone is a dedicated page for viewing the day's slate through a home run lens. It brings together simulation-based HR probabilities, park factors, and matchup data so you can quickly scan which players and games project for the most home run activity.

The Strikeout Center is the strikeout equivalent of the Home Run Zone - a dedicated view of the day's slate focused entirely on strikeouts. It shows simulation-derived strikeout probabilities for both batters and pitchers alongside relevant matchup and park data.

This page displays Ballpark Pal's simulation probabilities alongside the prop lines listed on PrizePicks and Underdog Fantasy. It covers categories like hits, bases, hits+runs+RBIs, singles, and batting strikeouts, making it straightforward to compare where the model's numbers sit relative to the listed lines.

Ballpark Pal supports DraftKings, FanDuel, and Yahoo for DFS projections, Ballpark DFS, and the Stacks tool. Salary caps and scoring systems are specific to each platform, so make sure you're looking at the right one for your contests.

DFS projections are simply the average results from the game simulations, converted into each platform's scoring system. Since the sim plays out every plate appearance, it naturally generates the stat categories that DFS platforms score on - hits, home runs, RBIs, runs, stolen bases, strikeouts (for pitchers), and so on. The projections aren't a separate model - they're a direct output of the same simulations that power everything else on the site.

Stacks are groups of 2–5 players from the same team that you'd slot into a DFS lineup together. The Stacks tool shows you the best combinations, including their total salary, points-per-dollar value, average projection, median, and upside (90th percentile outcome). Stacking is a popular DFS strategy because players on the same team benefit from correlated outcomes - if the team has a big inning, multiple players in your lineup benefit.

Yes. The Ballpark DFS tool lets you build lineups by adding players to position slots, tracking your salary usage and projected points as you go. You can filter by slate, game, position, and stack count. Lineups are saved locally so you won't lose your work if you navigate away.

Yes. The Export Center lets you download simulation results and projections. Several pages also have Excel export options. This is useful if you want to plug Ballpark Pal data into your own spreadsheets, models, or optimizer tools.

The system is built on pitch-level and batted-ball-level data going back to 2016. This includes physical characteristics of every pitch (velocity, spin, movement, location) and every batted ball (exit velocity, launch angle, spray angle). This data powers the expectation models, matchup model, and park factor calculations.

Weather data is sourced from WeatherStack. The system pulls hourly forecasts for each game venue, covering temperature, wind speed, wind direction, humidity, and barometric pressure.

Ballpark Pal has a dedicated Accuracy section where you can compare simulated results to actual outcomes. You can slice accuracy by month, day, park, team, individual batter, or individual pitcher across metrics like runs, hits, home runs, doubles/triples, singles, strikeouts, and walks. The models are typically left alone or lightly tweaked during the season with more heavy adjustments taking place during the offseason.

Ballpark Pal is a completely proprietary and independent system - the models, the data pipeline, the simulation engine, and the features that feed into them are all built in-house. It doesn't rely on publicly available projections, third-party models, or traditional advanced stats. The differences run deep, but to name a few: it uses a true bottom-up simulation rather than top-down formulas - every game is played out plate appearance by plate appearance, not estimated from team-level stats. The park factors are calculated at the individual batter level using batted-ball physics, not historical home/away splits. The matchup model uses 100+ proprietary features that don't resemble a standard stat line. And the simulation approach means all predictions - runs, props, DFS points, YRFI, F5 - come from a single internally consistent system rather than separate disconnected models.

It means each game is essentially played out 3,000 times in a computer before the real game starts. Each of those 3,000 iterations produces a complete game - every at-bat, every baserunner, every run scored. By the end, you have 3,000 possible versions of the game. Averaging those results gives you the projections; looking at the distribution gives you the probabilities. For example, if a player hit at least one home run in 750 out of 3,000 simulations, that's a 25% HR probability.