What Is Baseball’s Best And Most Valuable Prediction Algorithm?

If you are an avid fantasy baseball player, you have likely wondered how some can predict the performance of players so well? There are so many different stats and numbers to consider when looking at a baseball player and the way they play and are likely to contribute to your fantasy team. Not only do each of the teams have their own numbers and statistics to figure in, but each player has an entire history of stats and numbers that factor in with fields, conditions and all of the other wild variables that can change performance. But somehow the experts are always able to come up with an accurate prediction for the performance of a player and get that statistical edge in the fantasy game. It is due to a predictive algorithm that they use to factor in all the most meaningful bits of information.

I will explain to you how this algorithm works and how they are able to come up with their predictions.

How Does The Predictive Algorithm In Baseball Work?

The algorithm used to predict the performance of a player is known as PECOTA. This is an acronym for Player Empirical Comparison and Optimization Test Algorithm. This method was developed by a man named Nate Silver in 2003. He released the method to the public in his book called Baseball Prospectus that he published in 2003. Predictions made by this algorithm are published in Baseball Prospectus annual books every year as well as in very detailed forms on the website. The website is a subscription based website so if you want those professional predictions, get ready to pay for the information.


There are other algorithms that have been inspired by the PECOTA for other sports as well. Football has the KUBIAK; basketball has CARMELO and Hockey has VUKOTA.

What Does PECOTA Do?

When you are looking for predictions of your players, that are on your fantasy baseball team you want all of the pertinent information. The PECOTA is designed to forecast a player’s performance and analyze each of the major categories that typically factor into the majority of fantasy baseball categories. It even has a few metrics unique to the Baseball Prospectus.

This algorithm also set out predictions and ratings for players. It attempts to predict the rates in which the players will break out, how they will improve over time, how the players will experience attrition and even calculates the players market values for trading and buying purposes.

1651622 (1)

How Does PECOTA Work?

Many have tried to do writeups about the way PECOTA works. Many publications have had descriptions of the logic and methodology. Though they have been explained to some extent, the detailed ins and outs of the formulas are proprietary and have not been given out to the public, let alone the research community.

Though the exact detailed formulas are not shared, there is enough information to get an idea of the methodology that is used. It is said that PECOTA is a marriage of two previous methods. The work that was done by Bill James on similarity scores, and Baseball Prospectus’s former prediction method, Vlad, that was developed by Gary Huckabay.

While the Vlad would ascribe one of thirteen different career paths to players to project performance, the PECOTA goes to the extreme and gives every player in the history of the game a different path and tries to predict a current player’s performance based on the closest path to their trajectory. It takes those comparative scores to find the most appropriate path and applies it to the prediction of the players.


This means that a player’s scores and predictions are based on the most closely related player in the history of the game. The players numbers are compared to historical players and matched with a player with similar numbers. Each player is compared to the database of other players and seasons throughout history. In all, there are over 20,000 major league batting seasons as well as 15,000 minor league batting seasons to be compared to so there is more than likely going to be one that matches up well to the season any particular player is having.

There are four attributes that are primary in the predictions to look at when the predictions of this method come out. First it the production methods. These are things that the player produces. This includes things like batting average and strikeout rate.

The second metric is usage. This is how long and how often the player is used and sees field time.

The third are things that are classified ‘phenotypical attributes’. These attributes are facts about a player that may contribute to physical ability. Which hand they favor, how tall they are, how much they weigh, and how long they have been playing the game.

The fourth is the position of the player. For hitters, this means the position in the field that they play. For pitchers, it means if they are a starting or relief pitcher.

How To Know

Knowing what it expects can make you far more successful when it comes to fantasy baseball. There are plenty of people that have worked to get predictive methods down to a science. The algorithm PECOTA is one that is widely respected and does a great job predicting performance. This method uses the past and the stats of each player to compare them in a historical context. This gives the best idea of how things will turn out for that player and gives you a good idea how that player will work in your fantasy baseball team.