# Measuring difference between number when negative numbers are present. Use percentage?

I’m having difficulty understanding how to measure the difference between two numbers, when negative numbers are involved.  Say I’m testing Roulette strategies:

Strategy A, gain = \$50, Strategy B, gain = \$100.

In this case, I use percentages, and would calculated that Strategy B is 100% more effective than strategy A. However, I’m running into cases like this:

Strategy A result = (\$100) loss

Strategy B result = (\$50) loss

Strategy C result = \$0, break even.

Comparing Strategy A to B, I would say Strategy A = -\$100%. But comparing strategies A and B to Strategy C, what is the percentage? Strategy B is better than A, but as compared as a percentage to Strategy A, how is this calculated? Likewise, I running into cases like:

Strategy A loss = (\$50), Strategy B gain = \$0

Strategy A loss = (\$100), Strategy B gain = \$0

Strategy A loss = (\$50), Strategy B gain = \$50

Strategy A loss = (\$50), Strategy B gain = \$200

As a second question, is it more relevant to measure gain, or difference? Say I have:

Strategy A gain = \$10, Strategy B gain = \$20

Strategy C gain = \$50, Strategy D gain = \$75

Now, the gain between Strategy A to B is 100%, and from Strategy C to D is 50%. Yet, for money in my pocket, Strategy A to B gains me an extra \$10, while C to D gains me an extra \$25. So, for Strategy C to D, the percentage difference is smaller, but the absolute gain is greater.

Do you have suggestions on how I would properly measure the difference in effectiveness for such comparisons?

Relevance
• Gaining money is better than breaking even or losing money. If you actually find a strategy for roulette that lets you consistently win more than you lose, please publish it. You'll be rich until the casinos ban you like they do blackjack players who count cards.

• Login to reply the answers
• I'm a long term investor, live in Las Vegas and know gaming, educated in fields of math. Roulette is always a long term house win.

There are doubling strategies in betting that win until the big loss run, and a person is limited by table max bet or bankroll.

Statistically, there is a way betting a single number, and quitting on first win, never to play again. That could be a winner.

Gain is percent based on an initial bankroll.

A loss is negative.

The results as a percent or unity from bankroll allows + and -

If you start with \$500 as total bankroll, Quit at \$1000 or \$0, you might compare successes that way.

You are measuring it wrong in your method.

• Login to reply the answers