These systems are being incubated for World Cup for trading futures. They aren't public, and the referenced returns are for a futures portfolio, not stock portfolio. We're posting the composite returns to certain CTA tracking websites. Our public ETF trading model relies solely on the pairs system, which hasn't been that robust during its initial 6 month period, but I expect that to change.
I have one which would probably not qualify as "Holy Grail" with other traders for various reasons, but does the trick for me at this time ... ... developed it on 16 months data, out of sample test on the prior 18 months was pretty reasonable, I have been trading it live for 8 months now (since mid-July), fully automated, with a current P/F on that period slightly above 1.3 (90+ trades) (was about 1.7 for the combined 16+18 months backtesting period). This is far from impressive I know ... still looking for a "Better Holy Grail"
You can't solve it, but you can find "best-fit" results using the efficient market frontier, assuming you've been able to design robust, adaptive models. Most people don't have models like that, so they never make it to worry about the efficient frontier their models lay on. You have to have methods first, then do your frontier analysis, but that is only an approximation and you're left with the closest fitting curve if you've done your analyses correctly. This is as precise an answer as anyone can give once they've created their trading systems. The Napsack problem cannot be solved, only approximated. It can be impossible to solve for most, but even for some who've made it possible to find solutions they still know it is only an approximation if you combine the models' risk/return characteristics with efficient frontier analyses. (There's also an element of subjectivity and choice since you do eventually decide which frontiner is most optimal to use in which inexact proportion).
Good p/f for daytrading, but not compared to anything else. I bet it's only for one market, too, so that does not fall under "holy grail" where universal applicability is probably a pre-condition for that qualification. Is it for multiple markets, or are your results for use on just one instrument?
So basically there is no evidence of the HG, kind of gets old to read fabulous claims and no backup time and time again on these places, yet the only verifiable stuff available is mediocre at best. FoN
We just need another 20% to make the top 5. Top's at 67%, top 5's at 37%. We'll make the top 5 within the month if our next trade is a winner.
Mediocrity is part of the reason there is no "stock's world cup." You just can't get decent returns trading only in stocks. I'm sure the mediocrity got to them since very few are happy with the slightly greater than 50% chance after someone wins a stock world cup that they'll make the same 20-50% return the next year, and if the difference isn't obvious, futures are a lot different risk instruments than stocks. I don't expect those types of returns on my ETF trading. Those returns I quoted were what we submitted to World Cup to compete in the futures and forex trading championships.
Dude .. I'm pretty sure I solved that as an example of dynamic programming a long time ago. From memory, it's a good example of why greedy algorithms can shoot you in the foot .. and so useful in undergrad computer science algorithmics courses. It's not exactly anything new though.