# Chapter Eight: Video Roulette

## Yes, you can trust them

In a growing number of areas of the country video roulette has become available. For the most part, these games are pretty much the same to play on as regular LIVE Dealer Roulette games. The payoff rate is the same and the betting options are the same.

They are run by random number generating (or RNG) computer programs. Sometimes I have heard people playing these games grumble that the fix is in, that somehow the computer knows what they have bet on and deliberately produces a losing number. This is utter nonsense. The casinos are not going to risk losing their licenses by rigging the machines in such a way. They don't have to! They make a fine profit just relying on the built-in house advantage that defeats all but the most sophisticated players. It is in their interest that the game be fair and produce random results.

Each and every machine in the casino is individually licensed by the state gaming board in which it is operated. These gaming boards test, or hire others to test, the RNG programs and the output of these games to make sure they are fair and random. So the short answer is that Yes, you can trust video roulette.

Keep in mind that a regular roulette wheel is merely a mechanical random number generator. There is no logical difference between the wheel and a bit of computer software. They do the same thing. So there is no reason to fear video roulette. You won't find biased wheels, but that can be good news if your strategies assume random distribution as most of ours do.

I do have some advice concerning video roulette. First, it is frankly not quite as much fun as regular roulette because you do not have a live dealer and pit crew with whom to interact. In fact, the other people playing the same machine (most have 5 seats) might be quite annoying. The online casinos that have video roulette are not likely to have more than a few machines, so you cannot table hop either. When you play video roulette, get serious and do not let folks distract you.

The most annoying thing that some unhelpful nearby "coaches" may do is claim that they see "patterns" in the results. If 32 hits after a 22 they will not knowingly and say, "see? 32 often follows a 22." Of course they also would have said that if 32 followed a 23 (inverse numbers), or a 32 followed a 16 (16 x 2 = 32). What is happening here is a reflection of the fact that we humans are very good at seeing patterns. You can "see" some sort of pattern in just about any sequence! Ignore these jokers. The only patterns that are important are the ones that we can track statistically that have implications in terms of probability theory. The rest is nonsense.

I was in Phoenix, Arizona, about a year ago for a conference and, of course, found time to go to a nearby casino. To my delight I found the exact same video roulette game that I play in Minneapolis (in fact both machines are manufactured here in the Twin Cities). But as I started to play I suddenly realized that the table would not allow you to place a bet for more than $8 anyplace on the betting area. $8! Needless to say, I got off the machine immediately. My point is a simple one: When you start to play on a video roulette machine, check its limits immediately. Also make sure that 1 unit = $1. Most of the time that is the case, but in some places 1 unit is 50 cents.

The table limits obviously will limit what strategies you can pursue. The tables here in Minnesota stop at a bet of $99. Now, that is fine for inside bets. In fact, I can place $99 on each of two numbers plus another $99 as a split bet. More than enough for any of my strategies. But the outside bets are also limited to $99. That is why, for the most part, I cannot pursue my outside strategies here and can pursue them only in out of state casinos with live roulette.

So, to sum up: Yes, you can trust them. Ignore talkative players or on-lookers. Check out table limits right away and adapt your play accordingly.

## Some Inside Info you can Profit From

Many of the video roulette machines in the country are made by a company located right here in the Twin Cities. I interviewed the chief design engineer two years ago and it is that conversation that led me to research about how these machines work in more detail. As I said before, to guarantee that the programs are fair to customers, they are checked by state gaming boards. In Minnesota, the certification of machines is overseen by the Minnesota Gambling Enforcement Division. Before a given type of machine is certified for use, it is tested by outside companies. I interviewed a supervisor at the Minnesota Gambling Enforcement Division as well as the Director of Engineering and Testing Methods at one of the firms hired to check the fairness of the results of these games. From them I was able to ascertain two key bits of information. First, I learned the exact output measures they use. Second, I learned the statistical tests they use to measure whether that output is reasonably random. Knowing how they do these tests allows me to have a much more sophisticated knowledge of what the results of a program will be--in this case, what the video roulette game will do and will not do.

If you have forgotten the explanation of the Variance Demon, go back to page 2.

As I explained, there are results from roulette that look in the short run like they are terribly unfair and non-random. To be sure about this, you have to analyze the results of many spins. If you have a biased wheel or a flawed program that is generating non-random results, then the results of many spins will be significantly more or less than is predicted by standard probability calculation. How do we know if the variation is "significant"?

In response to the question “How many standard deviations above or below the mean do you have to be before we consider a score to be significantly different from the average?” most social scientists will respond, “about 2,” which marks about the 95th percentile. So, one possible way to test whether a RNG program is producing reasonably random results would be to measure the standard deviation and stipulate that anything plus-or-minus two standard deviations suggests a non-random result.

Why is this important? Before I explain why it is important, I need to make one more point about the importance of the sample size (n) and standard deviations. As I mentioned earlier, what kills most betting strategies are "unusually" long streaks or sequences. In the long run, red or black will come up slightly less than 50% of the time, which means on average it appears every other spin. But the key to success is knowing what the standard deviation is for red or black. That is, how far off of the average of "every other spin" does red or black actually occur? One cannot answer the question without making reference to a sample size. That is, for 100 spins you will get one standard deviation score, whereas with 10,000,000 spins you will get quite a different score. That is simply the results of randomness.

Think about a town of 100 people compared to a million people. While the averages may be similar (in terms of height, weight, income, number of kids, etc.), you will find far more variation from those averages in a large town than in a little one. The larger the sample size (n), the greater chance for really wild variation.

For example, I once ran a program for 1,000 spins and the longest streak of red or black I got was 9. When I ran the program for 100,000 spins I got steaks of 15.

The important thing to note here is that the larger the sample size, the greater the variation one will observe.

The reason this is important goes back to the idea that programs for video roulette must be tested to see if they are fair. Here's the punchline: Once we know the sample size used to test the casino's video roulette program and the tests used for measuring the program's fairness, we can roughly estimate the range of variation we will typically observe as output of the program.

Go back to the formula I mentioned earlier: n/38 + .4 n1/2. The key to the solution, obviously, is n. The larger the sample, the greater the unpredictability and variation. If we apply this concept to even money bets like red and black, then we could say that a streak of 16 reds in a row in a sample of 1000 would suggest that the wheel or computer program is badly biased or flawed, while in a sample of 1 million spins, 16 reds in a row would not be a surprise.

As I said before, I have learned both the size of the sample size (only 38,000) and how they measure its fairness (whether it is "reasonably" random). What is the bottom line for you the roulette player? Basically, you can safely assume that a video roulette game will play very much like a live roulette wheel that is in use for about 100,000 spins. No big secret here that will make you millions. But the comfort is that the strategies described in Roulette 2000 can be used with a great deal of confidence when playing Video Roulette since 100,000 is the typical test I used.