prob.of.ret <- function(nsteps = 100, nruns = 1000) {
  rt <- NULL;
  for(i in 1:nruns) {
    rt <- c(rt, match(0, cumsum(sample(c(-1,1), nsteps, repl=TRUE))))
  }
  # Below it is explained how to build the final expression
  # in a typical manner: from the inside out!

  # --- Phase 0
  #rt

  # --- Phase 1
  #split(rt, factor(rt));

  # ---- Phase 2
  #lapply(split(rt, factor(rt)), function (x) { length(x)/nruns });

  # ---- Phase 3
  as.matrix(lapply(split(rt, factor(rt)), function (x) { length(x)/nruns }));
}