good point. I still think the $ is good, as 60% being from the Euro is really a weighted average of over a dozen countries.
I think the original request remains.
Since most C2 traders are from USA, the US $ index is the best benchmark. Other countries hold huge amount of dollars, such as China (a trillion $ I have heard??). It remains the world currency
We are discussing Forex. Forex profits are best compared to the US $, as we are discussing currency trading, which often involves the US $ as one of the pairs.
The question is, how does a system offer outperformance for an American, given they hold dollars?
I do not buy the gold benchmark at all.
>The question is, how does a system offer outperformance for an American, given they hold dollars?
Most Americans don’t hold just dollars. Real estate for
example, and/or stocks/mutual funds, etc. might well
be a larger % of net worth than cash. Gold might better
correlate to a Cash + Real Estate + Stocks benchmark.
Divided by the Swiss we have a neutral currency evaluation
of gold.
>I do not buy the gold benchmark at all.
If you are looking at mostly performance vs the EC than
the DX makes sense. Gold/SF is a much broader benchmark
(and an interesting chart…;-)). GC/SF x 123 gives you a nice
$100K starting point. The DX is more or less just an inversion of
the EC.
I still think the best solution is to track the average account value of a bunch of C2 forex systems, possibly adjusted for differences in leverage. That tells you how the system performs in comparison to its competitors. What else is a benchmark? The same can be done for futures systems.
Nice idea.
I think there are (at least) two different ways of using a correlation with a "benchmark", depending on the question you want to answer.
1. You might want to know: "How does <system X> do compared to similar systems on C2". In that case the benchmark should be constructed from a (sub)set of C2 systems. Correlation is interesting, but even more interesting is the degree of outperformance ("alpha").
2. Alternatively, you might want to know: "Does <system X> do well, primarily because it rides a multiple year trend in <index Y>, and how would it do if <index Y> would suddenly drop/rise by 30%, or reverse its trend over the next couple of years". To answer this second question you need an existing index, e.g. S&P, Nasdaq, Gold, or Dollar-index with as long a history as possible, so you can get some inspiration from that history to develop different scenarios you want to consider. Now correlation (and beta) are more important.
With respect to point 2: I’m not sure of this, but is it not the case that the S&P is used as benchmark for American stocks because
(1) it is close to the optimal portfolio according to the efficient market hypothesis?
(2) it corresponds to a passive investment strategy that is profitable in the long run?
I don’t know (yet?) what a similar portfolio would be for currency pairs. The SCB benchmark seems to have to objective of being stable, not of being profitable.
I think I would try to base such a portfolio on a factor analysis of long term correlations between currency pairs, and try to select a subset of pairs that all correlated positively over a very long period and that were increasing in that period. I don’t have these correlations, but based on some shorter term correlations and trends I suspect that CHFUSD would play an important role in it.
To use correlations for stress-testing (my second point) it is not necessary (or perhaps even desirable) to have a single index model. You could in fact enter all US stocks on the right-hand side of the CAPM equation and come up with some algorithm that would select the ones that give you the most explanatory power. Although this might be clumsy for stocks, for currencies it shouldn’t be a big problem as the number of pairs is much smaller. So, I don’t think there’s really a need to construct a currency index, if you can just have a 4 index model that would give you 4 betas for the major currencies.
So basically you define the benchmark for each system separately as the system’s worst nightmare? 
Sounds fun to me, but I would not call it a benchmark. I think it should be the same for all systems within a class. Moreover, you can construct that index only in hindsight, after the system has traded. A benchmark should be defined in advance. IMHO it should be a simple, almost blind and passive strategy, such as “go long in this set of currency pairs” or “subscribe to all systems older than x months” or “buy gold” or “buy houses” . Basically something that Ross’ grandma or Bono the ape could have done and preferably such that they will still outsmart most active forex traders.
No, if you want to compare systems the benchmark(s) should be equal. So, for currencies, you would regress the system’s returns on the returns of the 4 or 6 main currency pairs, and keep these pairs the same for all (forex) systems. My point is that for stress testing you don’t have to construct a single index from these 4 or 6 pairs.
Yes they hold other things, but they are all priced in dollars. If S&P 500 is the benchmark for futures or stocks, then the US $ seems a better benchmark for currencies.
Since the US dollar is central in most traded pairs, and currency futures are priced in dollars, I still fail to see why it is not the most appropriate benchmark
Agreed, although the dollar index weightings leave a lot to be desired it is at least more representative than anything else in terms of it’s components and is also more practical in terms of ease of understanding together with a decent price history. The more I look for an alternative the more I keep coming back to the Dollar Index. The bottom line is whilst it’s not ideal and not for everyone (like most measures on here I suspect) it at least makes a lot more sense than the current S&P.
> The more I look for an alternative the more I keep coming back to the Dollar Index.
The more I think about it the more I like your SCB suggestion.
I assume it wouldn’t be too much of an effort to show both.
lol, well I think Jules made a good point about the main purpose of the SCB is as the name suggests to provide stability, Prechter uses it to accurately assess true global purchasing power or a market’s real performance, ie the Dow could be 20k but if the Dollar has dumped in the meantime then no-one is better off, but try explaining that to the man on the street and you get a blank stare, anyway the point is whilst I obviously think the SCB has a lot of merit for various applications I’m not now as convinced comparing FX systems on C2 is one of them, but I am happy to have brought it to other’s attention.
I agree that the Dollar Index has a lot of advantages. Quote from http://www.akmos.com/main/forex/usdx.html:
"Just as the Dow Jones Industrial Average provides a general indication of the value of the US stock market, the US Dollar Index (USDX®) provides a general indication of the international value of the US Dollar."
What convinces me even more is that the formula that they use is of the same form as what I had in mind, except that they use only USD based currency pairs of course.
Only one suggestion: Take the logarithm of the index. Because that will correspond to a portfolio with the used currency pairs. (See the formula at the website that I quoted: the index is computed by multiplying the currency pairs. The logarithm will change that in addition, which corresponds to buying a fixed portfolio of the underlying currencies.)
One question: Will the benchmark be long or short in this index?
Related to this question, there is one disadvantage though. The index was defined as 100 at March 1973, and in May 2007 it was 106. It seems that, unlike the S&P, it does not show a long term tendency to increase. That would be a rather friendly benchmark for forex systems! For the same reason I don’t see why it would be reasonable to believe that this index is a profitable long run investment. But that is what a benchmark should be IMHO.
(To be fair I must quote this from the same website:
“Since that time, the Dollar Index has traded as high as the mid-160’s and as low as the high-70’s.”)
Should a benchmark be something that is essentially stable or something that can be expected to be profitable on the long run? I prefer the latter. It should be hard to beat the benchmark.
I suppose that the defenders of the Dollar index mean that benchmark would be long in the Dollar index (USDX). I conclude that from the fact that Ross wrote
The question is, how does a system offer outperformance for an American, given they hold dollars?
But:
- The S&P tells you this: If I have a bag with dollars, and I buy a bunch of stocks from it, what will happen with the dollar value of my investment?
- A similar forex index should tell you this: If I have with bag of dollars, and I buy foreign currencies from it, what will happen with the dollar value of my investment?
But buying foreign currencies corresponds to shorting the Dollar Index.
It is just as easy to argue that the benchmark should be short USDX as to argue that it should be long USDX. So whenever you say to a vendor that is he is outperformed by the benchmark, he might argue that you take the index in the wrong direction and that he is the one who is outperforming.
The more I think about it the more I like Sams Gold / CHF suggestion. Actually at this moment I think that even the S&P is better than USDX, because at least it tells you what would happen with a reasonable alternative long term investment method.
A Forex benchmark should mimic what an average investor might do other than actively "trade" currency pairs. Why not construct a benchmark that reinvests in the strongest currency pair after some holding period, like a week or a month. This is a simple form of longer-term, trend-following which should generate long-term growth with short-term hiccups, like mutual funds.
> - A similar forex index should tell you this: If I have with bag of dollars, and I buy foreign currencies from it, what will happen with the dollar value of my investment?
>But buying foreign currencies corresponds to shorting the Dollar Index.
> It is just as easy to argue that the benchmark should be short USDX as to argue that it should be long USDX. So whenever you say to a vendor that is he is outperformed by the benchmark, he might argue that you take the index in the wrong direction and that he is the one who is outperforming.
Exactly. Thanks for taking the time to read my mind better than I could.
That would work for me too (although I would not know how to do it exactly). My main point is that, besides being passive and blind, the benchmark should be profitable in the long run.
