library(PerformanceAnalytics)
library(xts)

test_that("KellyRatio matches legacy expectations", {
  # Mock data
  set.seed(123)
  dates <- as.Date("2020-01-01") + 0:11
  R1 <- xts(rnorm(12, 0.01, 0.02), dates)
  R2 <- xts(rnorm(12, 0.005, 0.015), dates)
  R_multi <- cbind(R1, R2)
  colnames(R_multi) <- c("Asset1", "Asset2")
  
  Rf_const <- 0.001   # constant risk-free rate per period
  Rf_xts   <- xts(rep(0.001, 12), dates)  # time series risk-free
  names(Rf_xts) <- "RiskFree"
  
  # 1. Default (half-Kelly) single column
  res1 <- KellyRatio(R1, Rf = Rf_const)
  expect_equal(nrow(res1), 1)
  expect_equal(rownames(res1), "Kelly Ratio")
  
  # 2. Full Kelly (method = "full")
  res2 <- KellyRatio(R1, Rf = Rf_const, method = "full")
  res_half <- KellyRatio(R1, Rf = Rf_const, method = "half")
  expect_equal(as.numeric(res2), as.numeric(res_half * 2))
  
  # 3. Multiple assets
  res3 <- KellyRatio(R_multi, Rf = Rf_const)
  expect_equal(ncol(res3), 2)
  expect_equal(colnames(res3), c("Asset1", "Asset2"))
  
  # 4. Time series Rf input
  res4 <- KellyRatio(R_multi, Rf = Rf_xts)
  expect_equal(dim(res4), c(1, 2))
  
  # 5. NA handling
  R1_withNA <- R1
  R1_withNA[3] <- NA
  res5 <- KellyRatio(R1_withNA, Rf = Rf_const)
  expect_false(any(is.na(res5)))   # still computes because na.rm=TRUE
})