All sorts of crazy shit comes out of my mouth, but never once have I said "he's really not a great options trader, he just has a better model." Heck, I could trade options all day without any model. Bet I'd be dear near flat at the end of that day as well. I'm usually quite generous when I say options trading is 5% about the valuation. Probably more like 1%.
Your comment regarding pullbacks after large moves is not accurate for statically distributed price returns. A large move one direction will not result in large moves the other way (if this is what you mean by pullback). Such behavior is indicative of autocorrelation in returns which forces us to leave the binomial pricing model you mentioned in later comments.
This may not be the best metaphor, since throwing that ball into a basket is almost completely a matter of skill. Which supports the point opposite the one you're making. I lean far more toward the (non-Gaussian, heh) probability distribution side, and consider the typical candlestick chart to be mostly voodoo; the fact that it's an abstract representation with imaginary static points in a continuous process, and that people treat those "points" as meaningful, gives me the willies ("OMG, this candle opened at foo and closed at blarg! This MUST mean..." Time markers are imaginary, people!!!) However, having said that: if you have a million stacked sell orders and no buyers, the probability of that price going in one specific direction is nearly 100% - and the chart will show that movement once it begins. Ditto for several other obvious, nearly binary situations. The problem is that many traders desperately want that certainty - and smoking that hopium makes them pretend it exists in areas that are non-binary. As you say, apophenia (or perhaps more properly to chart reading, pareidolia) is the term. Given good risk management, though, even random entries can make a profit for quite a while - perhaps even for a lifetime. If we bet on coin flips at a buck a pop, but I can "cheat" by stopping my losses at a dime each, my P&L will be positive over the long term.
The log return series is stationary, which implies reversion to the mean in extreme moves. While it's a "stylized" fact take the log return of any index and you will find reversion to it's mean. In fact, I have no evidence for this but it seems like stationarity is necessary and sufficient for calling a series mean reverting. A pullback to the mean is not indicative of autocorrelation it is a property of a random variable undergoing an extreme move. Are you claiming that if I run a study where I measure the height of people in America, and I measure 3 people who are 7 feet tall, that the next person being 5 foot 9 inches indicates there is autocorrelation in that series? That's preposterous. Any random variable will behave this way whether or not it's autocorrelated. The binomial option pricing model operates under the same assumptions as the black scholes model. I am not even sure what you are talking about. Whereas @GRULSTMRNN had a point that the SABR model is probably what professional market makers use - you are actually fundamentally wrong. The reason you leave the BSM/Binomial model universe is to allow volatility to vary not some autocorrelation.
Clearly, the Earth is all one color. There's also no other variation, no movement, and everything that happens there is totally predictable. Or it could be that scaling too far out of context elides all meaning and makes everything look simple. Hmm, must give that some thought.
This thread is turning into an arguing match. No the markets are not random. I have proof. If the non randomness of markets were perfectly accurate, everybody would know it and the markets couldn't exist because everyone would know the exact timing of highs and lows. It is non random, but in an imperfect manner. I will attach a chart predicting future highs and lows to this post that is timing high and low prices in the FUTURE. It is of SP500.
Financial markets can be modeled under any number random variable models and their assumed distributions (ie normal, lognormal, pareto etc.) The problem is that none of these models explains reality significantly better than any of the others. The reason is that the underlying processes are more complex than the statistical models can capture. Gee, there's a revelation, human behavior is complex and highly unpredictable. So the real answer is yes and no. The task is to discern when it is and when it isn't, and have the proper trading plan in place to deal with regime change from random to non-random.
Absolutely not. Financial non interest rate asset time series are not stationary, they do not imply reversion to the mean in extreme moves. Through differencing though, for example, you can transform a random/stochastic process into a stationary process. Indexes never revert back to any mean because of any log return properties or for any other mathematical reason. No random process found in non-interest rate financial time series or distributions that define financial time series well have mean reverting properties that are dictated by mathematical principles. Interest rates do. Not others.