07.11.2019 - Have you seen the movie Moneyball? It is a movie based on real events where manager of baseball team with very limited budget is building team which won a lot of games using revolutionary approach of team building.
Baseball is collective sport based on individual player performance and almost all aspects of the game can be measured by statistics. Team chemistry is almost irrelevant except communication between pitcher and catcher or infield players. But that is far less important than individual batting and pitching performance. If we assume that player statistics are fixed we can easily simulate game results using simple baseball game generator. It is very similar to roulette games where all the probabilities are fixed and best winning strategies can be found.
So what they did in Moneyball? Baseball team is built from 9 batters and about 9 pitchers plus few replacements in case of fatique and injuries. Also it makes sense to use different rotation against left handed pitchers as it clearly affects the batting statistics. And each player have their statistics from previous seasons like batting average, slugging, on base percentage and so on. Pitchers also have theirs like batting average of opposed batters.
Moneyball
Now the extreme approach is to hire best baseball player available for your offense with more than 30 million per year salary. As your budget is limited you can hire 8 low cost players with 2.5 million salary to fill the lineup and that would cost 50 million annually. Opposite extreme would be to hire 9 average players with salary of 5.6 million salary.
The batting average of the best MLB batter in 2019 was 0.335 while worst regular player batting average was 0.205. If we have one star player with 8 weak cheap players we would get combined batting average of 0.219. In the meantime average batting average in MLB was 0.252 which is the batting average of the team built by just average players. Pretty big difference for about the same money right?
Of course this is just a simplified model to explain the strategy. To be more accurate we would need to assign realistic salaries to players. Top MLB batter earns 35 million annually but salaries of average batters should be also adjusted from 5.6 million we used in our example.
Batting average of course is not the only important batting stat and also it is truth that the first player in batting order have more at bats than the ninth. It is good strategy to place your high on-base percentage guys ahead of your best slugging guys. But this all can be quantified and used in more advanced model. At the end of the day team management can simulate the overal batting order and get the optimal lineup for the budget they have.
Other Sports
There is clear intention to follow this path in other sports mostly in leagues with salary cap like NFL. Of course it is much more complicated as other sports relies more on teamwork, player intelligence and other aspects which are very difficult to measure. it needs much more advanced models and estimations but they get some results too.
