Quote from rcanfiel:
Quote from empee:
A) "Statistically Signficant" -- how do you quantify that. Do you have a sample size of N? I do considerable backtesting, Monte Carlo, and probably the hardest question to answer is -- IS IT STATISTICALLY SIGNIFICANT. To say that as a being a "given" is obtuse viewpoint at best. How many ppl who backtest have seen a system work great and then when say the test period changes it fails miserably?
Backtesting and similar methods is not a sample size. It is someone looking thruiogh a universe fo possibilities, nad wrongly believing that those few that outperform will continue to do so. It is a similar problem to relying on testimonials from vendors. You are seeing a cherry picked set that does not represent the truth.
b) Wrong and wrong. How can you say charts do not definitively provide an edge? As Taleb says (Black Swan) one instance disproves what you're saying. Until you can disprove say, head and shoulders don't work (I don't think they do) but there may be traders who can statistically show they do, you can't make blanket statements like that. I've seen ppl/trading with violation of every trading rule that have historically made money (not saying that it was STATISTICALLY SIGNIFICANT). There many be edges that you have or haven't thought of, once again to say "anything" can't be an edge is not true. Just because YOU haven't figured out an EDGE someone else has, doesn't mean it doesn't exist.
Most studies show little to no value-add for indicators or pattern methods. The problem is those who disagree without any significant proof to the opposite. Traders who can " statistically show they do" rarely do and prefer to provide worthless anecdotal or gut feels instead.