Logistic Regression giving incorrect results

I am working on a website where I collect the results of chess games that people have played. Looking at the ratings of the player and the difference between their rating and that of their opponent, I plot a graph with dots representing win (green), draw (blue), and loss (red).

With this information, I also implemented a logistic regression algorithm to categorize the cutoffs for winning and win/drawing. Using the rating and the difference as my two features, I get a classifier, and then draw the boundaries on the chart for where the classifier changes its prediction.

My code for gradient descent, the cost function, and the sigmoid function are below.

  def gradient_descent()
    oldJ = 0    
    newJ = J()
    alpha = 1.0     # Learning rate
    run = 0
    while (run < 100) do
      tmpTheta = Array.new
      for j in 0...numFeatures do
        sum = 0
        for i in 0...m do
          sum += ((h(training_data[:x][i]) - training_data[:y][i][0]) * training_data[:x][i][j])
        end
        tmpTheta[j] = Array.new
        tmpTheta[j][0] = theta[j, 0] - (alpha / m) * sum  # Alpha * partial derivative of J with respect to theta_j
      end
      self.theta = Matrix.rows(tmpTheta)
      oldJ = newJ
      newJ = J()
      run += 1
      if (run == 100 && (oldJ - newJ > 0.001)) then run -= 20 end   # Do 20 more if the error is still going down a fair amount.
      if (oldJ < newJ)
        alpha /= 10
      end
    end
  end

  def J()
    sum = 0
    for i in 0...m
      sum += ((training_data[:y][i][0] * Math.log(h(training_data[:x][i]))) 
          + ((1 - training_data[:y][i][0]) * Math.log(1 - h(training_data[:x][i]))))
    end
    return (-1.0 / m) * sum
  end

  def h(x)
    if (x.class != 'Matrix')    # In case it comes in as a row vector or an array
      x = Matrix.rows([x])      # [x] because if it's a row vector we want [[a, b]] to get an array whose first row is x.
    end
    x = x.transpose   # x is