Algorithm 1.

Super-learner estimation of d_{0, A(1)}

1:	function SuperLearner(o1, …, On, ${\widehat{f}}_{2,1},\dots ,{\widehat{f}}_{2,j}$)
2:	Let F be a randomly ordered vector of length n containing n/U 1s, n/U 2s,…, n/U U’s
3:	Initialize an empty matrix X of dimension n × J
4:	for u = 1 to U do
5:	for j =1 to J do
6:	Fit the estimate f2, u, j by running ${\widehat{f}}_{2,j}$ on the set {Oi : Fi ≠ u}
7:	For all i such that Fi = u, let Xi, j = f2, u, j(A(0)i, V(1)i)
8:	Run an optimization routine to solve: ${\alpha }_n=\arg \underset{\alpha \in {\mathrm{\Delta }}_{J-1}}{min}{\displaystyle \sum _{u=1}^U{\displaystyle \sum _{i=F_i=u}{L}_{2,g_0}\left({\displaystyle \sum _{j=1}^J{\alpha }_j{X}_{i,j}}\right)}}$
9:	for j = 1 to J do
10:	Fit the estimate fj by running ${\widehat{f}}_{2,j}$ on the {Oi : i = 1, …, n}
11:	Let $f_{{\alpha }_n}\equiv {\displaystyle {\sum }_{j=1}^J{\alpha }_{n,j}{f}_j}$
12:	return $d_{n,A(1)}\equiv (a(0),v(1)\mapsto I(f_{{\alpha }_n}(a(0),v(1))>0).$.