2013 cards - estimating Chris Davis' 2013 card
Posted: Thu Nov 14, 2013 12:41 am2013 cards - estimating Chris Davis' 2013 card
In one of the keeper leagues that I'm in - a new one, with a small ($70M) salary cap - I kind of backed into a trade where I acquired Chris Davis for 2012, and the right to declare him as a keeper for 2013.
I've got quite a few pitchers and batters where their 2013 baseball years were better than their 2012 years. I was emailing back and forth with one of the other managers in the league. We were chatting about what kind of issues my keeper team might have with the salary cap for the 2013 card year. He speculated that Chris Davis would have a monster card - "like Moss on steroids", and that the salary for Chris Davis in 2013 might be $12.2 million.
Well, it got me thinking...what will Chris Davis 2013 card look like???
So I did some number crunching. I found league stats for 2012 and 2013 for the AL, and I compared them to see what the overall stats looked like for rates of walks, singles, doubles, triples and homeruns. The rates were relatively constant from year to year, but I made notes on the averages and the changes.
I found split stats (vs. LHP and vs. RHP) for Chris Davis for 2012 and 2013. I also got the split stats for Brandon Moss for 2012.
I developed a methodology:
1. Calculate the rate of HBP in relation to total plate appearances. Check to see whether the distribution of HBP vs left/right is relatively constant. (It was).
2. Take the individual player's 2012 stats vs. LHP, and calculate the rate of occurrence for each of HBP, walk, single, double triple and homerun ("the events"). Then express the occurrence as a number over 216. I.e. multiply the rate of occurrence times 216. (216, because that is the number of die roll combinations with 3 dice - 6 times 6 times 6).
3. Calculate the average pitcher card results, by taking the league average for each event, and multiplying that times 108. Please note that it shouldn't matter whether the real life results occur from fielding (X chart) or from pitchers' effectiveness, since it's all "baked" into the same league average statistics.
4. For each type of event, subtract the pitcher card result (3) from the total result (2). The net difference is the theoretical number of die rolls for each event for the batter card vs LHP.
5. Add up the actual die roll chances for each type of event vs. LHP from his 2012 card.
6. Compare 4 and 5 - theoretically they should be the same. In fact, they were very close. However, I noted some trends for small differences that make common sense - more on that later.
7. Now for 2013 vs LHP - repeat the above process for 2013 exactly as it was done for 2012. Calculate the theoretical batter card totals for 2013 as though it was a new 2013 card (as I did for 2013 in step 4 above, except this time using 2013 stats).
8. Calculate the change in the theoretical batter card totals from 2012 to 2013 - i.e. step 7 result minus step 4 result. This number can be positive if 2013 was a better year for that event, or worse if 2013 was a worse year than 2012 for that type of event.
9. For each event, add the delta change (2013 minus 2012) from step 8 to the actual 2012 die roll card totals for each event from the 2012 card (from step 5). This is the theoretical estimated batter card totals for 2013, with adjustments for the "constant" differences (from step 6 above).
10. Calculate the estimated ballpark singles and ballpark home runs for the particular batter for 2013. For a batter like Chris Davis, this isn't so hard, since it's almost certain that he'll have normal ballpark singles vs. LHP and RHP, and also 8 ballpark homerun chances vs each of LHP as well as RHP. The trick with the ballpark chances is that you have to take an average of the "road" ballparks, and then take a simple average of the road parks together with the home park - in this case Camden Yards for Chris Davis. I am assuming that Camden (and other AL ballparks) are the same in 2013 as they were in 2012. Calculate the total singles and homeruns from ballpark chances for 2013 for Chris Davis vs. LHP.
11. Subtract the ballpark singles and home runs (10 above) from the total batter card singles and home runs (9 above). This net amount is the actual "pure" singles and homerun die roll chances for Chris Davis for 2013 vs LHP.
12. Repeat steps 2 through 11 using stats vs RHP for 2012 and 2013.
The results for 2012 were very, very close to his actual 2012 card -
HBP - calculated - 3.24 - actual 2012 card - 3
vs LHP
W vs LHP - calc - 0 actual - 0
SI vs LHP - calc - 16.5, actual 16.5 (includes est. BP singles)
2B vs LHP - calc 4.4, actual 4.2
3B vs LHP- calc - 0, actual 0
HR vs LHP - calc - 9.7, actual 11.2 (includes est. ballpark HRs using average BP HR rate).
Note the difference in HRs vs LHP - about 1.5 - card homeruns exceed estimated card. My explanation of this difference is that the average LH pitcher card will yield significantly fewer HRs to LH batters than the overall league average. Therefore, the strat batter card compensates for this by increasing the home runs for the power hitting lefty hitter. Hence the 1.5 HR difference compared to the calcs using overall league averages.
2012 - vs RHP:
W vs RHP - calc - 7.5 actual - 4
SI vs RHP - calc - 16.9, actual 17.75 (includes est. BP singles)
2B vs RHP - calc 2.55, actual 1.5
3B vs RHP- calc - 0, actual 0
HR vs RHP - calc - 9.6, actual 9.6 (includes est. ballpark HRs using average BP HR rate).
There are differences here in doubles and walks. The walks make sense to me - that a lefty hitter is more likely to draw relatively more walks off the pitcher card vs a RHP - i.e. a righty pitcher will give up relatively more walks to a lefty hitter on average. Therefore, there is less of a need for walks to appear on the batter card vs RHP, based on a simple league average. The difference in doubles is relatively small, and I haven't tried to explain it.
The main point is that, overall, the model produces estimates of the batter card that are pretty darn close to the actual strat card.
Now, when I complete the remaining steps to project Chris Davis' 2013 card, I get the following projected card die roll totals:
HBP - 3 die roll chances vs both LHP and RHP
vs. LHP:
W- 4
SI - total SI - 6.3, est. ballpark SI effects- 2.8, net "pure" single die roll chances -- 3.6
2b - 6.9
3b - 0
HR - total 10.6, est. ballpark HR's 4.4 (i.e. 8 times .55 average home and away), net "pure" HR's - 6.15
vs RHP:
W- 17
SI - total SI - 7.3, est. ballpark SI effects- 2.8, net "pure" single die roll chances -- 4.5
2b - 8.6
3b - .1
HR - total 17.15, est. ballpark HR's 4.4 (i.e. 8 times .55 average home and away), net "pure" HR's - 12.75
Estimated balance - 6R
In one of the keeper leagues that I'm in - a new one, with a small ($70M) salary cap - I kind of backed into a trade where I acquired Chris Davis for 2012, and the right to declare him as a keeper for 2013.
I've got quite a few pitchers and batters where their 2013 baseball years were better than their 2012 years. I was emailing back and forth with one of the other managers in the league. We were chatting about what kind of issues my keeper team might have with the salary cap for the 2013 card year. He speculated that Chris Davis would have a monster card - "like Moss on steroids", and that the salary for Chris Davis in 2013 might be $12.2 million.
Well, it got me thinking...what will Chris Davis 2013 card look like???
So I did some number crunching. I found league stats for 2012 and 2013 for the AL, and I compared them to see what the overall stats looked like for rates of walks, singles, doubles, triples and homeruns. The rates were relatively constant from year to year, but I made notes on the averages and the changes.
I found split stats (vs. LHP and vs. RHP) for Chris Davis for 2012 and 2013. I also got the split stats for Brandon Moss for 2012.
I developed a methodology:
1. Calculate the rate of HBP in relation to total plate appearances. Check to see whether the distribution of HBP vs left/right is relatively constant. (It was).
2. Take the individual player's 2012 stats vs. LHP, and calculate the rate of occurrence for each of HBP, walk, single, double triple and homerun ("the events"). Then express the occurrence as a number over 216. I.e. multiply the rate of occurrence times 216. (216, because that is the number of die roll combinations with 3 dice - 6 times 6 times 6).
3. Calculate the average pitcher card results, by taking the league average for each event, and multiplying that times 108. Please note that it shouldn't matter whether the real life results occur from fielding (X chart) or from pitchers' effectiveness, since it's all "baked" into the same league average statistics.
4. For each type of event, subtract the pitcher card result (3) from the total result (2). The net difference is the theoretical number of die rolls for each event for the batter card vs LHP.
5. Add up the actual die roll chances for each type of event vs. LHP from his 2012 card.
6. Compare 4 and 5 - theoretically they should be the same. In fact, they were very close. However, I noted some trends for small differences that make common sense - more on that later.
7. Now for 2013 vs LHP - repeat the above process for 2013 exactly as it was done for 2012. Calculate the theoretical batter card totals for 2013 as though it was a new 2013 card (as I did for 2013 in step 4 above, except this time using 2013 stats).
8. Calculate the change in the theoretical batter card totals from 2012 to 2013 - i.e. step 7 result minus step 4 result. This number can be positive if 2013 was a better year for that event, or worse if 2013 was a worse year than 2012 for that type of event.
9. For each event, add the delta change (2013 minus 2012) from step 8 to the actual 2012 die roll card totals for each event from the 2012 card (from step 5). This is the theoretical estimated batter card totals for 2013, with adjustments for the "constant" differences (from step 6 above).
10. Calculate the estimated ballpark singles and ballpark home runs for the particular batter for 2013. For a batter like Chris Davis, this isn't so hard, since it's almost certain that he'll have normal ballpark singles vs. LHP and RHP, and also 8 ballpark homerun chances vs each of LHP as well as RHP. The trick with the ballpark chances is that you have to take an average of the "road" ballparks, and then take a simple average of the road parks together with the home park - in this case Camden Yards for Chris Davis. I am assuming that Camden (and other AL ballparks) are the same in 2013 as they were in 2012. Calculate the total singles and homeruns from ballpark chances for 2013 for Chris Davis vs. LHP.
11. Subtract the ballpark singles and home runs (10 above) from the total batter card singles and home runs (9 above). This net amount is the actual "pure" singles and homerun die roll chances for Chris Davis for 2013 vs LHP.
12. Repeat steps 2 through 11 using stats vs RHP for 2012 and 2013.
The results for 2012 were very, very close to his actual 2012 card -
HBP - calculated - 3.24 - actual 2012 card - 3
vs LHP
W vs LHP - calc - 0 actual - 0
SI vs LHP - calc - 16.5, actual 16.5 (includes est. BP singles)
2B vs LHP - calc 4.4, actual 4.2
3B vs LHP- calc - 0, actual 0
HR vs LHP - calc - 9.7, actual 11.2 (includes est. ballpark HRs using average BP HR rate).
Note the difference in HRs vs LHP - about 1.5 - card homeruns exceed estimated card. My explanation of this difference is that the average LH pitcher card will yield significantly fewer HRs to LH batters than the overall league average. Therefore, the strat batter card compensates for this by increasing the home runs for the power hitting lefty hitter. Hence the 1.5 HR difference compared to the calcs using overall league averages.
2012 - vs RHP:
W vs RHP - calc - 7.5 actual - 4
SI vs RHP - calc - 16.9, actual 17.75 (includes est. BP singles)
2B vs RHP - calc 2.55, actual 1.5
3B vs RHP- calc - 0, actual 0
HR vs RHP - calc - 9.6, actual 9.6 (includes est. ballpark HRs using average BP HR rate).
There are differences here in doubles and walks. The walks make sense to me - that a lefty hitter is more likely to draw relatively more walks off the pitcher card vs a RHP - i.e. a righty pitcher will give up relatively more walks to a lefty hitter on average. Therefore, there is less of a need for walks to appear on the batter card vs RHP, based on a simple league average. The difference in doubles is relatively small, and I haven't tried to explain it.
The main point is that, overall, the model produces estimates of the batter card that are pretty darn close to the actual strat card.
Now, when I complete the remaining steps to project Chris Davis' 2013 card, I get the following projected card die roll totals:
HBP - 3 die roll chances vs both LHP and RHP
vs. LHP:
W- 4
SI - total SI - 6.3, est. ballpark SI effects- 2.8, net "pure" single die roll chances -- 3.6
2b - 6.9
3b - 0
HR - total 10.6, est. ballpark HR's 4.4 (i.e. 8 times .55 average home and away), net "pure" HR's - 6.15
vs RHP:
W- 17
SI - total SI - 7.3, est. ballpark SI effects- 2.8, net "pure" single die roll chances -- 4.5
2b - 8.6
3b - .1
HR - total 17.15, est. ballpark HR's 4.4 (i.e. 8 times .55 average home and away), net "pure" HR's - 12.75
Estimated balance - 6R
