When forecasting teams, here are some things to consider

Predictions that went wrong for 2021 MLB season

There are projections from almost every major outlet available on players. These are handy when trying to make a fantasy roster, but we are going to try to determine how many wins a team will get. We want to be able to make some quality future wagers on team total wins and playoff odds so individual player projections will come in handy for some insight especially considering the influence we should apply. The major focuses should be utilized elsewhere. Let’s explore.


We could start off by utilizing the team’s won/lost record. If we did this, we would not be accurate in valuing its strength. Let me explain. A better way to determine the strength of a team is by utilizing the Pythagorean records. These are based on runs scored and runs allowed. Baseball teams can be unlucky or lucky, or productive or less productive, hence they will either win games they should have lost or vice versa. Pythagorean records indicate what a team should have done based on its productivity removing the luck factor, thereby creating a more accurate view of its ability to win games. Last year the SF Giants won 107 games. The LA Dodgers won 106 games. It can be said the Giants were lucky to have won 107 games though. Based on their runs scored and allowed, they should have only won 103 games, still outstanding. The Dodgers were unlucky as their true win total should have been 109 wins! The Dodgers are actually 6 games better than the Giants is a more accurate view than if we went off their actual records. This is significant when forecasting this year’s team win totals. We need a starting point and the Pythagorean example is a good one.

Another view we should check into is the Base Runs model. Base Runs is a model similar to Pythagorean Theory (runs scored vs runs allowed) in that it looks at expected runs scored and allowed. It is based on a normal distribution of runs scored or allowed given the type of hits and sequencing of those hits instead of raw data points. The Base Runs model is the most accurate model available to us. Using the same teams from above, the SF Giants should have won 103 wins and the LA Dodgers 107 wins using the Base Runs model. In this case, the Dodgers are 4 games better than the Giants.

Here is a breakdown of the divisions with both Pythagorean and Base Runs models.

Once we get our starting point, we need to look at the make-up of the player performances. This is where the projections can come in handy. I can not tell you which projections to use (I use Steamer) yet the premise should be the same for all projections. We are not exactly trying to know who will hit more HRs than last year. We are trying to know if they are going to be more or less productive than the last year. There is a significant way we need to value these players. We will look at the Giants. Here are their Steamer projections for 2022.

Here are their actual performances from 2021

We are looking for significant variances. For this team, there are many! The 2022 projections have Brandon Crawford with 1.2 OFF WAR and 2.6 WAR (his value is on defense). His 2021 actual performance was 27.8 OFF WAR and 5.5 WAR! He was All-Star worthy in 2021 and is projected to be just an average player in 2022. That is a HUGE drop-off from their leader in producing offense. A further look into this indicates 2021 was a career season for him (his career data is the next graphic). He is not a spring chicken so he should not be expected to back up a career year with another one. He should be expected to have BIG fall off, and he does.

A team such as this one is in for a big fall. It is because their best contributors had career years and are not going to be expected to back them up with equal or better ones this year. The 2021 Giants had four players with double-digit OFF WAR success, but they only have projected in 2022 (Brandon Belt) and he is only projected to be half as productive as he was in 2021! The more regression we see in players is mainly due to their age and if they had career seasons. Younger players tend to build off of what they did and are expected to improve. The make-up of each team is a considerable factor when looking at forecasting. Certainly, we do not just look at the batters either. We need to look at the starting pitching and bullpen in the same way too.

Another factor to consider is the losses and gains a team will have had during the off-season. Free agents come and go in this sport. We need to know if the team needs to overcome losses, or if they are getting stronger because they have added better talent. in some cases like the Giants, they could have had a producer retire. Buster Posey was the team’s 3rd best OFF WAR player in 2021. He has retired and will not play in 2022. This team now has diminished all of its former OFF WAR contributors! Posey retired, Crawford should revert back to a normal season, Belt is only half as good as last year, and Ruff will see a diminished usage role, plus he too is off a career year!

We need to consider the callups as well. Identify the best prospects in the game as they can be big producers right away. They are generally forecasted conservatively, but they can add big value to some teams. The Giants will be using Joey Bart this season for example. He will replace Posey behind the plate. He is not expected to be the hitter Posey was, be he is good on DEF. A team like Tampa Bay has the number one prospect in all of baseball in Wander Franco. He will get his first shot at a full season and should be a big contributor to the team wins.

I like to work from the Base Runs or Pythagorean model to the projected win totals. I like to see what variances are, then see if they are explainable via player regressions/progression, additions/subtractions, or prospects getting their chance to play. SF is projected at 82 wins for 2022. This is a 21 game drop-off from 2021! We did not go over their pitching but if it is like their hitting situation, the variance is explainable. If all things were equal, then we would have to see if the Giants made improvements that would have them at minimumm4 games better so they could catch the Dodgers. In this case, they are much worse. We do this exercise for every team, then we can get some win projections to work with. We must be able to explain variances and to keep the integrity of the true win gap at the start. Teams either improve enough to move up compared to their rivals or they fall back. It is possible to have teams improve or fall and still be in the same slot, meaning their rivals did the same. It is a fun exercise and an important one. The 2022 forecasting cannot be done because Free Agency has yet to be completed! It will be a mad dash at the end of the lockout but if you are ready, you can get the jump on the early odds from the sportsbooks!

Leave a Reply

%d bloggers like this: