clear	
clear matrix
set matsize 800
set mem 500m
cd "[your directory]"
use Lanham_team_hitting_pitching



*** merge Lahman HBP/sf/sh patch

merge 1:1 yearID teamID lgID using "lahman_hbp_sf_sh_patch.dta", update


drop _merge

//drop non AL/NL

drop if lgID != "AL" & lgID != "NL"


// drop pre-modern

drop if yearID <1900


//rename hit variables



rename B doub

rename S trip


// add 2024 bb_ref_team_patch

merge m:m yearID teamID using "bb_ref_2024_patch.dta"


	drop _merge

// generate singles 

gen sing = H-(doub+trip)

// runs per game 

gen rpg=(R/IPouts)*27


*** standardize runs per game by seasons


bysort yearID: egen mean_rpg = mean(rpg)
bysort yearID: egen sd_rpg = sd(rpg)
gen z_rpg = (rpg - mean_rpg) / sd_rpg

 


*** generate SLG ***

gen slg=(sing+2*doub+3*trip+4*HR)/AB


*** generate opb ***

gen obp =(H+BB+HBP)/(AB+BB+HBP+SF)

*** generate ops ***

gen ops=obp+slg



*** generate TLD wOBA ***

gen wOBA=(HR*1.7+1.37*trip+1.08*doub+.77*sing+.62*BB+.65*HBP)/(AB+H+BB+HBP+SF)




/// standardize ops  woba by season


bysort yearID: egen mean_ops = mean(ops)
bysort yearID: egen sd_ops = sd(ops)
gen z_ops = (ops - mean_ops) / sd_ops

bysort yearID: egen mean_wOBA = mean(wOBA)
bysort yearID: egen sd_wOBA = sd(wOBA)
gen z_woba = (wOBA - mean_wOBA) / sd_wOBA










	
	

	
	//merge BBREF rbat
	
rename teamID teamid

	merge m:m yearID teamid using "bb_ref_team_rbat_coded.dta"


drop _merge

*** standardize rbat by season  

bysort yearID: egen mean_rbat = mean(rbat)
bysort yearID: egen sd_rbat = sd(rbat)
gen z_rbat = (rbat - mean_rbat) / sd_rbat


//// create swoba --regression derived wOBA

* Generate variables to store the coefficients 
generate b_hrPA  = .
generate b_sPA   = .
generate b_dPA   = .
generate b_tPA   = .
generate b_bbPA  = .
generate b_hbpPA = .
generate b_cons  = .


*** generate regression variables for regression-derived wOBA: swoba

gen sPA =sing/(AB+H+BB+HBP+SF)
gen dPA= doub/(AB+H+BB+HBP+SF)
gen tPA= trip/(AB+H+BB+HBP+SF)
gen hrPA=HR/(AB+H+BB+HBP+SF)
gen bbPA=BB/(AB+H+BB+HBP+SF)
gen hbpPA=HBP/(AB+H+BB+HBP+SF)


* Generate variable to store regression-derived woba
generate swoba = .
* Loop over the years from 1900 to 2024
forvalues y = 1900/2024 {

    * Run the regression for the current yearID
    regress rpg hrPA sPA dPA tPA bbPA hbpPA if yearID == `y'

    * Store the coefficients in the dataset for the current yearID
    replace b_hrPA  = _b[hrPA]   if yearID == `y'
    replace b_sPA   = _b[sPA]    if yearID == `y'
    replace b_dPA   = _b[dPA]    if yearID == `y'
    replace b_tPA   = _b[tPA]    if yearID == `y'
    replace b_bbPA  = _b[bbPA]   if yearID == `y'
    replace b_hbpPA = _b[hbpPA]  if yearID == `y'
    replace b_cons  = _b[_cons]  if yearID == `y'

    * calculate team swoba values 
    tempvar temp_pred

    
    predict `temp_pred' if yearID == `y', xb

  
    replace swoba = `temp_pred' if yearID == `y'

    
}

**** save swoba coefficients for predictive testing

preserve
keep yearID b_hrPA b_sPA b_dPA b_tPA b_bbPA b_hbpPA b_cons
save "swoba_coefficients.dta", replace
restore


* standardizing swoba


	 bysort yearID: egen mean_swoba = mean(swoba)
	bysort yearID: egen sd_swoba  = sd(swoba)
	gen z_swoba = (swoba - mean_swoba) / sd_swoba
	
	
	
*** create fangraphs woba: fwoba 


merge m:m yearID using "fg_woba_wts.dta"
drop _merge

gen fwoba=(hrPA*fwhr+tPA*fw3b+dPA*fw2b+sPA*fw1b+bbPA*fwbb+hbpPA*fwhbp)

	 bysort yearID: egen mean_fwoba = mean(fwoba)
	bysort yearID: egen sd_fwoba  = sd(fwoba)
	gen z_fwoba = (fwoba - mean_fwoba) / sd_fwoba

	



/// perform regressions of runs per game explained


regress z_rpg z_swoba
regress z_rpg z_ops
regress z_rpg z_woba
regress z_rpg z_fwoba

regress z_rpg z_rbat


su rpg ops swoba,d


twoway (scatter z_rpg z_ops, mcolor(ltblue)) ///
       (lfit z_rpg z_ops, lcolor(black)) ///
       , xlabel(-2.5  ".580" -1.5 ".640"  -.5 ".700" .5 "760" 1.5 ".820" 2.5 ".880", nogrid  ) ///
         ylabel( -2.8 "3" -1.3 "4" .2 "5" 1.7 "6" 3.2 "7", nogrid) ///
          xtitle("z_ops") ///
         graphregion(color(white)) ///
         plotregion(style(none)) ///
         xscale(r(-3 3 )) ///
		  legend(off)
		  
		  
	
twoway (scatter z_rpg z_swoba, mcolor(ltblue)) ///
       (lfit z_rpg z_swoba, lcolor(black)) ///
       , xlabel(-2.3 "3" -.6 "4" 1 "5" 2.6 "6"  4.2 "7", nogrid  ) ///
         ylabel( -2.8 "3" -1.3 "4" .2 "5" 1.7 "6" 3.2 "7", nogrid) ///
          xtitle("swoba") ///
         graphregion(color(white)) ///
         plotregion(style(none)) ///
         xscale(r(-3.8 3.2)) ///
		  legend(off)
		  

	  

gen Rfg = .
gen Rwr = .
gen Rops = .
gen Rbrb =.

foreach yr of numlist 1900/2024 { // Adjust the range to your data
    // Run the regression with z_FIP only for the current year
    regress rpg ops if year == `yr'

    // Store season r2s for ops
    local r2_1 = e(r2)
    replace Rops = `r2_1' if year == `yr'
	
	 // Run the regression with z_FIP only for the current year
    regress rpg fwoba if year == `yr'

    // Store R2 from the first model in R21 for the current year
    local r2_1 = e(r2)
    replace Rfg = `r2_1' if year == `yr'

	
	 // Run the regression with z_FIP only for the current year
    regress rpg swoba if year == `yr'

    // Store R2 from the first model in R21 for the current year
    local r2_1 = e(r2)
    replace Rwr = `r2_1' if year == `yr'
	
	
	 // Run the regression with z_FIP only for the current year
    regress rpg rbat if year == `yr'

    // Store R2 from the first model in R21 for the current year
    local r2_1 = e(r2)
    replace Rbrb = `r2_1' if year == `yr'


   
}


preserve

keep yearID Rfg Rwr Rops Rbrb

   
	  gen Rfg_100=Rfg*100
	  gen Rwr_100=Rwr*100
	  gen Rops_100=Rops*100
	  gen Rbrb_100=Rbrb*100  
	  
	      twoway (lpolyci Rfg_100 year, lwidth(none) bwidth(10)) ///  // Gray area for smoothed Rfg
       (lpoly Rfg_100 year, lcolor(black) lpattern(shortdash) lwidth(thin) bwidth(10)) ///  // Black dashed lpoly line for smoothed Rfg
       (lpolyci Rops_100 year, lwidth(none) bwidth(10)) ///  // Gray area for ci
       (lpoly Rops_100 year, lcolor(black) lpattern(shortdash) lwidth(thin) bwidth(10)) ///  // Black dashed lpoly line for smoothed Rops
       (lpolyci Rwr_100 year, lwidth(none) bwidth(10)) ///  // Gray area for ci
       (lpoly Rwr_100 year, lcolor(black) lpattern(shortdash) lwidth(thin) bwidth(10)) ///  // Black dashed lpoly line for smoothed Rwr
       (lpolyci Rbrb_100 year, lwidth(none) bwidth(10)) ///  // Gray area for ci
       (lpoly Rbrb_100 year, lcolor(black) lpattern(shortdash) lwidth(thin) bwidth(10)), ///  // Black dashed lpoly line for smoothed Rbrb
       ytitle("R-squared (scaled x100)") ///
       xtitle("Year") ///
       ylabel(50(10)100, labsize(medium) nogrid) ///
       xlabel(1900(10)2024, angle(45) labsize(medium) nogrid) ///
       legend(off) ///
       graphregion(color(white)) ///
       plotregion(margin(zero))




restore


//// Now examine individual hitter WOBAr


clear	

use Lahman hitting data 
		  
		  
///rename some variables


rename AB  ab
rename R r
rename H h
rename B b2
rename L b3
rename HR hr
rename RBI rbi
rename SB sb
rename CS cs
rename BB bb
rename IBB ibb
rename HBP hbp
rename SH sh
rename SF sf
rename G g
rename SO so









* drop pitchers

merge m:1 playerID using  "bb_ref_lahman_pitcher_names_match.dta", update replace





keep if _merge==1 | playerID=="ruthba01" // drop all pitchers except Ruth


drop _merge


// drop seasons prior to 1900 



drop if yearID<1900


//drop non AL/NL

drop if lgID != "AL" & lgID != "NL"



// aggregate players season performances variables over multiple stints

collapse (sum) g ab r h b2 b3 hr rbi sb cs bb so ibb hbp sh sf , by(playerID yearID lgID)

drop if ab==0




*** rename for consistency

rename yearID yearid

rename playerID playerid

** eliminate select steroid users



drop if (playerid=="bondsba01" & yearid > 1996)
drop if (playerid == "mcgwima01")
drop if (playerid == "sosasa01")
drop if (playerid == "cansejo01")
drop if (playerid == "giambja01")
drop if (playerid == "palmera01")
drop if (playerid == "rodrial01")

*** 


/// add 2024 patch


merge m:m yearid playerid  using  "bb_ref_2024_season_hitting_coded"
drop if ab ==.
drop if ab==0

drop _merge 



/// merge swoba coefficients

rename yearid yearID

merge m:m yearID using "C:\Users\dmk38\Documents\x5\swoba_coefficients.dta"

drop _merge




rename yearID yearid

// Calculate the obs for each player-year combination


gen slg=(((h-(b2+b3+hr))+2*b2+3*b3+4*hr))/ab
gen obp=(h+bb+hbp)/(ab+bb+hbp+sf)
gen ops=slg+obp

drop if ab ==.
drop if ab==0





// calculate  swoba for each player-year combo

gen doub =b2
gen trip=b3
gen sing= h-(hr+doub+trip)

gen hrPA=hr/(ab+bb+hbp+sf)
gen tPA=trip/(ab+bb+hbp+sf)
gen dPA=doub/(ab+bb+hbp+sf)
gen sPA=sing/(ab+bb+hbp+sf)
gen bbPA=bb/(ab+bb+hbp+sf)
gen hbpPA=hbp/(ab+bb+hbp+sf)

gen swoba=(hrPA*b_hrPA+tPA*b_tPA+dPA*b_dPA+sPA*b_sPA+bbPA*b_bbPA+hbpPA*b_hbpPA+b_cons)




/// calculate weighted mean and SD for each season's    swoba irrespective of league ; weighting by  PAs

egen total_pa = sum(ab+bb+hbp+sf), by(yearid)









generate weighted_swoba_product = swoba * (ab+bb+hbp+sf)
egen total_weighted_swoba = sum(weighted_swoba_product), by(yearid)
generate weighted_mean_swoba = total_weighted_swoba / total_pa

generate squared_diff_swoba = (swoba - weighted_mean_swoba)^2
egen total_squared_diff_swoba = sum(squared_diff_swoba *(ab+bb+hbp+sf)), by(yearid)
generate weighted_variance_swoba = (total_squared_diff_swoba / (total_pa - 1))
generate weighted_sd_swoba = sqrt(weighted_variance_swoba)














// Calculate the real PA (Plate Appearances) as the sum of the relevant totals
gen PA = ab+bb+sh+sf+hbp



// adjusted Gwynn 96

replace ab=495 if (yearid==1996 & playerid=="gwynnto01")





// Determine batting title eligibility including ad hoc adjustments made by MLB...
gen ba_elig = 0
replace ba_elig = 1 if (yearid < 1949 & g > 99) | ///
                    (yearid >= 1950 & yearid <= 1956 & ab > 399) | ///
                    (yearid >= 1957 & yearid <= 1960 & PA > 477) | ///
                    (yearid > 1960 & PA > 502) | ///
                    (playerid == "hargrbu01" & yearid == 1926) | ///
                    (playerid == "lombaer01" & yearid == 1942) | ///
                    (playerid == "madlobi01" & yearid == 1981) | ///
					(yearid==1996 & playerid=="gwynnto01") | ///
					(yearid==1969 & playerid=="carewro01")

replace ba_elig = 1 if (yearid == 1918 & PA > 419 | yearid ==1919 & PA > 419)
replace ba_elig	= 1 if (yearid == 1972 & PA > 483)	
replace ba_elig = 1 if (yearid == 1981 & PA > 320 )
	replace ba_elig = 1 if (yearid == 1981 & PA > 356 )
	replace ba_elig=1 if (yearid==1994 & PA >356)
	replace ba_elig = 1 if (yearid== 2020 & PA > 185)
	



				   
// Calculate z-scores 


gen z_woba= .



replace z_ops = (ops - weighted_mean_ops) / weighted_sd_ops if ba_elig == 1

1


replace z_swoba = (swoba - weighted_mean_swoba) / weighted_sd_swoba if ba_elig == 1








//delete batters not eligible for batting title

drop if ba_elig == 0


/// center z_scores at 0






su z_swoba
replace z_swoba=z_swoba- r(mean)




/// create a scaled measure of swoba


su z_swoba,d
gen normal_swoba=((z_swoba-r(min))/(r(max)-r(min)))*100 

/// sops+
su normal_swoba
gen swoba_plus = 100*(normal_swoba/r(mean))  

*** generate swoba performance report
*** fix names

rename (playerid yearid) (playerID yearID)

merge m:m playerID using "lahman_names.dta"


* Keep only the relevant observations where yearID is 2024 and nameFirst and nameLast are missing
gen byte update_needed = (yearID == 2024 & missing(nameFirst) & missing(nameLast))

* Split the playerID variable into words (assuming a maximum of 3 words)
split playerID, parse(" ") gen(word)

* Update nameFirst and nameLast based on the split components for the relevant cases
replace nameFirst = word1 if update_needed
replace nameLast = word2 if !missing(word3) & update_needed
replace nameLast = word2 + " " + word3 if !missing(word3) & update_needed
replace nameLast = word2 if missing(word3) & update_needed









//dump extraneous variables

keep nameFirst nameLast yearID lgID swoba z_swoba swoba_plus 
drop if swoba=.

export excel using "C:\Users\dmk38\Documents\x5\swoba_report.xls", firstrow(variables) replace