well, more what I was talking about, you have a system which has worked well in the past that uses constant size then it goes through a period where it is wrong so consistently that it becomes comical is it a good idea to increase size since you are confident wins and losses will return to a more nomal distribution? asking mathmaticans more than anecdotal traders and conversely, would it be wise to reduce size when it goes through one of those "just can't miss" phases?
honestly van tharp when I was a newby sounded like the missing key to my trading. Now his stuff reminds me more of a Carney than an asset to trading. YMMV
No, that assumption isn't automatically correct. The trading criteria is broken in your example in that the successful trades only match a particular market condition. If you go from consistent winners to consistent losers under the same criteria, it means that you don't really have an edge. You just coincidentally found yourself using criteria that matched the prevailing conditions at the time. Think of it like this. You have a relative who loses a bunch of money after the dot com crash. He decides he can pick stocks better than his adviser. He transfers his money in mid 2003 and begins to pick his own stocks. From then until about Oct 2007 he has a success rate of 85%. It seems like every time he buys something he is right. Then suddenly almost every trade he makes for the rest of 2007 is a loser. He tells himself, my picks have been so consistently wrong that it is comical. I know my criteria is right because it worked so well before. Therefore, I am now going to increase my size because my system has a proven win rate of 85%, so it has to return to there soon. By the end of 2008 he has almost blown up his account. His actual mistake was in assuming that a high win rate was the same thing as having an edge. In reality all he had was a high beta strategy that happened to coincide with a bull market. He would've realized this if he'd properly analyzed his system, but he just assumed that it was his genius at the root of his success. A string of losses DOES NOT increase the likelihood that the next trade will be a winner. A string of losses simply indicates that your system does not fit current conditions. It says nothing about whether those conditions will persist.
I would ask, why do you only add once? I'm not suggesting that it is wrong to only add once. I'm just wondering, what is the basis for that rule? They way you are throwing it out there sounds like it was just some arbitrary rule that you came up with to limit losses. As opposed to an actual calculated trade that increases the positive expectancy of the system.
good answer the old 50 heads in a row betting on coin tosses has always baffled me, but you would expect it is time to start betting tails until it gets back to about 50/50 heads to tails
It isn't baffling at all from a statistical viewpoint. If the sample size is large enough, the frequency of 50 heads in a row is precisely predictable. Also, the fact that there were 50 heads in a row does not in any way make a tails more likely on the next toss, or the next 10 tosses. That is the gambler's fallacy. If you don't get that out of your head, your trading career is going to be rough. As much as you seem to want to assume that a bettor should switch to betting tails after a string of 50 heads, that change doesn't impact his odds of winning at all.
That's why most quants suck at trading; they refuse to bet heads on the 51st flip. I've never traded in a frictionless nor fair-market. (1/2)^50. When I see 50-days up on the Dow I am going to be long on day 51.
In flipping coins there is never a trend. If the Dow goes up 50 days in a row there is clearly a trend. You cannot compare flipping coins with trading the Dow. I think in trading the sharp ratio is very important. So you now what to do to keep the ratio on a good level.
Which is why trend following systems typically outperform contrarian systems. The most interesting result of backtesting markets is that it turns out that the probability of another positive day in the markets actually increases as the positive streak extends.