Reproducing the results from A Bayesian analysis of the time through the order penalty in baseball (TTO):

Get data:

• download retro_final_PA_1990-2020d.csv (which was compiled in TTO_/data_wrangling), and TTO_dataset_410.csv and TTO_dataset_510.csv (which were compiled in TTO_/data/create_design_matrix4_10.R and TTO_/data/create_design_matrix5_10.R), from https://upenn.app.box.com/v/retrosheet-pa-1990-2000, and store these datasets inTTO_/data

Fit the bayesian model used in the TTO paper to observed data:

• run the Bayesian model (the Stan file TTO_/model_positive_slope_prior/tto9_PF_BL_linePosSlope.stan) for multiple years on a cluster via the array job TTO_/model_positive_slope_prior/run_obs_fitModel_obs_AJ.sh, which runs TTO_/model_positive_slope_prior/obs_fitModel.R
• then, run TTO_/model_positive_slope_prior/results_obs.R and TTO_/model_positive_slope_prior/results_obs2.R to obtain the plots and results from the fitted models
• same process but in the folder TTO_/model_pitcher_varying_decay to use the model with pitcher-varying slopes
• same process but in the folder TTO_/model_indicators to use the indicator model (indicator at each batter sequence number 1,...,27)
• simulation study in TTO_/model_positive_slope_prior/sims