Originally posted by Funster
I'm sorry but this is typical of the BS so prevalent in this industry. Roulette is a mathematically impossible game to beat. If you played it without time constriction you would eventually lose. Now blackjack and poker are a different kettle of fish.
The subject matter in question was not trying to predict the game of roulette, the mathematics of which are obvious, but predicting the underlying assumptions of that model. In order for the results to be fully random, the wheel has to be perfectly balanced, be spinning at a variety of speeds sufficient to produce random results in combination with the dealer dropping the ball at random locations and flipping it with random speeds.
As it turns out, many dealers when they spin the wheel do so with a similar velocity, many spin the ball with a consistent velocity, and many begin the drop at the ball around the same area on the wheel (e.g. some wait for one of the green zeroes before spinning the ball). In addition to this, no wheel is perfectly balanced! The greater the unbalance, the less random the results. You add all of these variables up and you have a chaotic system that seems random most of the time, but there are definite patterns that can be exploited.
The way they won money was to predict what area of the wheel the ball would land on, and then make the bets accordingly. Betting a particular number is too difficult, but you can identify patterns of what area of the wheel the ball is likely to land. You only need a small advantage to be profitable.
Doyne Farmer did this very thing by calculating the "calibration of the roulette operator's flick of the wheel, the speed of the bouncing ball, and the tilt of the wheel's wobble." Farmer went on to found LTCM which proved that using Chaos to trade the markets works. Their eventual failure was due to poor money-management and over leverage, and nothing to do with their mathematical model.
In fact, because of this the casinos have made several changes to the game, including shifting dealers frequently, stopping the wheel after so many spins and spinning the ball the other direction, using multiple balls with different size and mass, recalibrating wheels periodically, and switching wheels periodically so you cannot handicap a single wheel over time. Now gaining an advantage is too difficult (not to mention illegal to use a device like Farmer used in the casinos).
This is all well documented.
Originally posted by Funster
Before I make my comments please understand I am a real trader putting my own money on the line every day.
...
I have followed both discretionary and fully (months of research) mechanical systems for a number of years. I can say categorically that if you are blindly using a trend following system - no matter how you cut it - multiple entrants, pullbacks, multiple exits etc etc you will end up with a zero sum game. At best you might figure a profit - but the best I could ever manage is a system with a profit factor of around 1.4. This is just too close to the bone and there are safer homes for your capital.
I also put my money on the line trading, and a mechanical method at that. Just because you haven't found a system you have confidence in does not mean that one does not exist.
If you don't believe that mechanical systems based on trend following work, then you have to also believe that discretionary systems based on trend following does not work. The reality is most discretionary systems are just complex mechanical systems where the trader has not been able to fully mechanize the system. Some parts of the trading analysis are too complex to easily model. In addition, I believe there is a huge group of "hybrid traders" whol use a mechanical setup, and then use discretion to execute the trade (Tony Oz scanner as an example). These methods are neither 100% discretionary nor 100% mechanical, but clearly show that a system of some kind is being utilized. There is a lot of evidence of traders here on ET who have been
consistently profitable using a method that trades with the trend.
Regarding Profit Factor, 1.4 is at the low-end of what I consider "tradable systems," but you also need to look at Sharpe ratio, Payoff ratio, and Recovery Factor. Even though at 1.4 PF is low, the equity curve is what is most important, and the other risk factors address that better than just PF, but PF is a good benchmark to start from. I prefer PF above 1.8 myself.
) for the sole purpose of using physics to beat the wheel. It was a solo effort, and my funds and the state of microchip technology at the time were sufficient to only design a very bulky and, shall we say, easily detectable computer, so I abandoned the project in 1980. I can, however, say that in laboratory (home) conditions I could easily beat the wheel and I have no doubt that it was successfully done in live, casino conditions. Suffice it to say that this is a subject very close to my heart, and when you uncategorically call it BS, I know you're full of it.