C2 Score Makes No Sense

I’m sure this topic has been covered before, and it seems like the only answer given is that it’s a “secret formula”, but the C2 score makes absolutely no sense. The help desk assures me that it’s working correctly but that it may need some improvement. Both of my strategies are routinely in the 90’s most of the time, but will seemingly randomly drop into the 60’s for no apparent reason; most of these occasions were on days that the strategies were up in value. Both of my strategies are at all time highs, and have been up in value every day this week, but both dropped to from the 90’s to the 60’s. If this is “working correctly”, then there’s some really bad stuff in that “secret formula”, and the C2 score should be removed until it is improved. Thanks, David

I have experienced the exact same thing. Thanks for bringing this up David J.

I am also troubled by the apparent circular logic of the C2 score…a high C2 drives subscribers, which drives a higher C2 score, which drives subscribers. Not sure also why it should be categorized as a Popularity score, unless it’s largely popularity weighted.

How do you know if c2 score attract subscribers?

I’ve never seen c2 score as a criteria for system selection discussed here at c2. Returns, dd, number of trades etc. As a subscriber I don’t even check it for the system.

I suppose I should leave it to Matt to confirm that C2 score does not have any impact on the strategies that new subs are filtered into when they join C2, and that C2 score does not affect the ranking of top strategies on the Grid and elsewhere on the site. Matt?

In the Grid you can sort and rank strategies in different ways including but not limited by c2 scores. Everybody can verify it by themselves.

You are correct Andrey. However, the Popular Strategies listed on the dashboard, and the Leader Board shown under the Find Trading Strategy tab both appear to be ranked by C2 score.

1 Like

I too have been trying to figure this out (seeing fluctuations as described by you all).

I guess they are keeping the nature of the C2 score algorithm secret, I would have to assume because it drives something important (such as the order of the systems on the dashboard or somesuch), otherwise I’m not sure why there would be a concern about it getting “gamed” and therefore the details of how it is derived exposed.

That said, I am quite curious myself as to what drives such changes in the score. Someone suggested in another thread that when you put trades on it varies, which would indicate it is somehow tied to account risk in some way, but I’m not 100% convinced, and if this is a factor it may be (the model results) just getting re-sampled on a trade execution rather than something with the trade itself causing a score change.

Another thought may be it is relative-ranked, in which case as other systems change, your score will change, and possibly by drastic amounts depending on how sensitive the components of the C2 score are in general. I’ve seen this a lot in my work on relative-ranking-based stock trading systems…

I think it is natural to wonder about these things for those of us who build models, it is (figuring out what makes things tick) after all part of the attraction of what we do in the market =)

I see your point. I found these lists useless, so never used them to select a strategy. They look similar.

At first glance looks like these lists reward old strategies, and strategies with high return to dd (recovery or calmar) ratio.

You are correct, C2 scores are relatively-ranked compared to how other strategies are doing at a “moment in time” versus your strategy. The major problem, as others have discussed, is that score does indeed reward older systems and more “active” system (i.e. strategies holding positions overnight–even at an urealized loss or draw down versus those opening and closing out positions each day like a “day trader” would). Another major issue, as the original poster of this thread mentioned, is that during the time frame when your C2 score is being “recalculated” after you have opened and closed out a position, your C2 score is somehow in “limbo” under some “baseline” score (like in the 60s or so) and your new (real) score only appears hours or so later (usually overnight the next morning) after the result and effect the newly closed trade’s “score” (added to your past performance and in comparison to other strategies) is added to that “baseline” value. This phenomenon is usually common and more prevalent in newer systems that are less than 3 months/90 days old. The older a system is the more stable/established that baseline score relative to the difference and changes to the “real score” after each closed out trade. An older system like YZ Income and R Option can gradually go from the 90s to 60s with a continual string of bad trades in time but never experience the drastic overnight changes that a 3-month old system (even with great performance) goes thru after each opened and closed out trade while its true C2 is being recalculated with its performance being compared to more established systems.

My scores are, indeed, back up to 96 and 97 from the 60’s
yesterday, after absolutely no action in either system. Is there any way to define this as other than
a glitch? Here’s how I see it: All of us system owners are paying C2 for
them to provide the infrastructure and the marketplace for subscribers to
follow our strategies, for which we get paid back. C2 publishes a “score” for each strategy –
which I think is great. But if a score
is published, it should be both fair and it should calculate correctly. If I’m enrolled in a class for which I’m
paying, and have a score of 97% (an A), and hand in an A paper for my day’s
work after which the professor tells me I now have a 65% (a D), and won’t tell
me what I’ve done to drop my score from an A to a D in one day during which I
got an A on my paper – I think anyone would agree that this is either unfair or
incorrect. Does it help that my score
goes back up to 97 the next day? Sure –
but what about the potential subscribers who decided against my system because they
happen to look at it in the afternoon instead of the next morning, and didn’t
want to subscribe to a D system? The score
is either unfair or is not calculating correctly – I don’t know that there is any
other explanation. I ask that Collective
stop publishing the score until it can be either fixed or changed to be more