Thanks, this gives me some stuff to look into. At a first glance though, they seem not to have statistics on newbie traders. Kahneman did research in the 1990s about the performance of professional traders, using data he got from trading firms he consulted. He found that top traders achieved an average annual return of 3% over market. More remarkably, the correlation coefficient between the yearly performances of individual traders was zero. No trader was found who consistently performed above average. This seems to suggest that a "trading talent" does not exist. Kahnemans suggestion to the trading firms was to remove the bonus system, as it had no effect on trader performance.
Did they count the losers who are no longer trading , i.e the 95% , are only 20 to 50 % of the 5 % profitable currently?
SSRN has thousands of papers on trading, although obviously if you are looking for something very specific, it might not be available. If you want to look at a slightly different approach to analyzing trader performance and performance drivers, you can look at this paper http://www.pnas.org/content/106/2/623.full The findings suggest that short-term trading talent is innate and a result of hormonal influences on the ability to "read" the market. I like this because I generally believe that biology pretty much determines everything in this life. This paper tackles the "luck vs. skill" question over the longer term and concludes that the empirical data don't support the "luck" hypothesis. http://www.lmcm.com/868299.pdf Frankly, I think anyone still thinking trading is luck is living in the past. The main reason that I say this is because I've come to the conclusion that trading is best thought of as process analysis. The market is a machine for creating fluctuations. Those fluctuations unfold according to a specific process. The process has some practical limitations, but is ultimately very flexible. The ability to analyze and document processes is most definitely a skill, not luck. Therefore, the trader who can most accurately and comprehensively document the market's fluctuation process must be the most skillful trader. I use the example of Elliott Wave as an example of skill in analyzing and documenting the fluctuation process. This is not because I believe that Elliott Wave is the correct analysis of the process, but because it gives an indication of the form a correct analysis would take. I have found (and corrected) 4 fundamental errors in Elliott Wave and created a "superset" of wave theory at a much more generic level. This new variation of wave theory has enabled me to identify over 50 logical (not visual, this is not "pattern recognition"; one of Elliott's fundamental errors was the emphasis he placed on the visual) structures which are present in any market. There is one logical structure which is actually "illogical", which I call my "wildcard" structure. Because its logic is unclear, I avoid it. But the key point is that I can identify it. I've basically turned everything the market does into one giant algorithm for generating trade signals. Elliott Wave could be considered a "subset" of my theory, so I've actually inverted the relationship between the earlier theory and my own. Each logical structure has a expectancy associated with its corresponding trade trigger. Some expectancies are positive and some are negative. I take the trade triggers with positive expectancies and avoid the ones with negative expectancy. Interestingly, there are more negative than positive expectancies, which means that I spend less time in the market than out of it. Which is fine by me because I want my time in the market to be as efficient as possible. As a result of all this, I can tell you at any given minute whether you should be long, short, flat or (and this is where my strategy differs from most) waiting for a negative expectancy trade to play out in order to take the next positive expectancy trade, in a given market and what your odds would be in that trade. I can tell you to the minute and the tick what needs to happen in any given situation in order for a trade to trigger. What I can't tell you is whether a trade will trigger or not. That's for the market to decide, not me. Now, I've either built the trading world's largest monument to being "fooled by randomness" or I have some level of skill in analyzing the market's fluctuations. I like to think it's the latter. At least I think you'd have to agree that the way I've approached trading makes it possible to consider it a skill vs. just trying to get lucky.
What if the 20-50% quit trading while they were ahead? I'm not sure why you are fixated on the 95% number, nor do I care, I'm only pointing out that it's probably too high an estimate and that it's probably closer to 70%. I know that doesn't make for dramatic statements about trading, but the world isn't always dramatic. At least, it isn't for those of us who aren't drama queens.
There are so many statistical and fundamental issues with that research, I would not even go there. For starters, while evaluating any performance vs the market, it's worth looking at risk adjusted returns, which he did not. Personally, I think he has fallen a long way since his Nobel-prize winning research (I am not a specialist, by no means, but I do read a lot scientific papers and you spot a bad fish when you see one). Overall, I have a problem with any form of cognitive studies - they take a group of idiots (obviously, because why else you would participate in a study for $10 dollars an hour instead of doing productive work), give them a few quick-ass problems to solve and draw deep conclusions based on this. While some of the conclusions are more correct then not (e.g. you should mistrust your intuition and instead learn to analyze data numerically), many are total crap and will probably get debunked some time in the future. On the topic I can only say - without any barriers to entry (e.g. structured training or any sort of selection process) majority of humans would fail in any activity. I am pretty sure 95% of football players never make it to big leagues and 95% of girls in the local ballet school will never set a fut on the stage of Kirov Theater.
Come on sle, it's both simpler, and worse than that. Because the majority of traders by sheer numbers are retail, and every synthetic retail product has an edge on it, the average case is a small loss for the trader. So .. given enough time, 100% lose, just by the law of large numbers. Even if they're trading cash, the spread and any commissions have the same effect. This is pretty simple stuff. If you're not arbing something, making a spread, or collecting some kind of rebate / commission .. you pay into the system.
This seems to me more plausible than the 95%. When traders base their decision on throwing a coin, you would expect that about 50% are profitable in any observed period, disregarding commission. As the real number is lower, the intelligence and skill of coins seems to be highly underestimated.
Kahneman has mainly looked into the strange failures and misconceptions of human thinking. Discovering "statistical and fundamental issues" in a research without knowing anything of it is certainly a prime example.
Kahnemans prospect theory is a great contribution to behavioral economics and gives you great insight (an alternative one, too) into the concept of risk premium. But once you past that, you realize that the research of a Nobel prize winner is no different then any other scientific paper. I read a lot of them and in this particular case it's relatively easy for me to make an opinion since I have a PhD in a related field. I would not make similar comments on, for example, string theory research. Here, I am specifically talking about his research on trading performance (only one paper), which in my view actually compounds the shortcomings of the cognitive research methods (small sample sizes, inherent participant selection bias and the low utility of the test result to the participants) with his lack of understanding of what trading "performance" actually is.