But C2 does already give ratings! See the left side of your window, under Features, the link Collective2 Ratings. And if a system vendor posts in a forum, his C2 rating is shown too.
Given that C2 gives such ratings, it is legitimate to ask that these ratings reflect important characteristics such as drag downs. I believe that the C2 ratings are already influenced by this, so the question of the thread starter seems to be that drag downs get more weight in the rating.
I agree with you if you mean that the systems should not be compared by a single rating. They differ on many dimensions, not just one. So if you mean that C2 should delete the ratings, I would not oppose that. But I can imagine that C2 created these ratings because other people asked for it. Well, I’ve no problem with the fact that ratings are given. I just don’t use them.
I totally disagree with the suggestion to report just the trades. It is vitally important to summarize the outcomes of these trades with various statistics such as the cumulative profit, % profitable trades, information about drag downs, etc., and with graphics that illustrate the behavior of the system.
Given that such statistics are reported, many readers will use them to choose a system as if they are ratings. So it is important that the reported statistics provide a fair comparison of the systems, without biases due to age. If a statistic does not satisfy this condition, then it should not be reported or one should be warned against it.
My analysis above suggests that the Annualized % and the maximum Drag down do not satisfy this condition with the present set up of C2, where systems are allowed to use earlier profits. So they should not be reported or the set up should be changed. Exclamation mark!
I think the ‘best systems’ search function should not return the highest annualized % return, but should sort by annual%return divided by max drawdown%. That way the unrealistic gunslinger systems would not automatically rise to the top.
‘Best’ should reflect reward as a function of risk, not max reward regardless of risk.
As I understand it, the Collective2 Ratings (under the Features) are already influenced by drag downs. But on the page Best Systems, the systems are ordered by Annualized return.
The problem of your specific suggestion (Annualized return / max Drag Down) is that for a young system, the max Drag Down is probably low, which will then have an excessive effect on the rating. E.g. a system that has 1% drag down in one month may well have 5% drag down in another month, which makes a factor 5 difference.
Furthermore, I think that Annualized return and max Drag Down are both age dependent, but probably not in the same way. This because, from a statistical point of view, the Annualized return is an sample mean and the max Drag down is a sample extreme, and their limit distributions are different if the sample size (i.e. the number of days) increase. Therefore I think that this statistic has a bias to favor young systems. I.e. if you and I have the same system but you start in January and I start in July, then your ranking will probably be lower just because you had more exposure to the risk of a drag down.
Apart from this, I think the user should be able to choose his own risk level, and get a ranking accordingly. E.g. a 2-dimensional plot with a profit measure on one axis and a risk measure on the other, and the systems as points in the plot. But perhaps there are too many systems to do this conveniently.
I would appreciate it if, in addition to the Best Systems page, there would be a page with for each system more statistics and no figures, and where you can order systems according to each of these statistics. If the ordering function is too difficult, it would already be nice if one can export the statistics to Excel to do the ordering therein. And everyone can then compute the statistics that he likes.
With ‘more statistics’ I mean all the statistics that are now reported on each systems page (e.g. drag down, sharpe, etc.) and perhaps more (e.g. average profit % per trade). These things are reported already, and it would be convenient if you can get them in a single overview table.
