XGBoost Algorithm predicting too high or too low probablity

xgboost
probability

#1

I have run xgboost with objective function binary logistic. After that I have ranked the data. The first 10 percent in Test data is covering 85% events.
The overall data had 3% events
However the Probability range is from 0.99 to 0.04 in the first decile. Is there anything that needs to be corrected.


#2

I am also running xgboost on horse data to retrieve probabilities, with binary logistic.
I turn them into odds and calculate the rank.
It works niceley in the train data, but when I apply the result of training on the test set I have an error :
Error in predict.xgb.Booster(xgb1, dtest) :
Feature names stored in object and newdata are different.
so I momentarily put it aside , as I could not find where it’s wrong.

for your problem, what do you mean by
" The first 10 percent in Test data is covering 85% events." ?