Great post! Thanks for sharing your insights with us.
How is strategy working 20 months later?
Also, you mention that all that you need to trade for others is power of attorney. Without a license and being registered, are you opening yourself up to legal liabilities should the trades go poorly?

Cheers,
Jon

First, I’m familiar with Collective2 but the way it works is that they use their own data feeds to determine what the P/L for your trade calls would have been. Unfortunately, they don’t take interest into account! So I’d basically come out looking like a breakeven system. So “no go” there unless they change things. Personally, I think I’ll start with fund management. All it requires are willing funds and a power-of-attorney to act as the trader on the account. The customers would just need to open an trading account, name me as the trader and sign a power of attorney to that effect. I would only be allowed to trade on the account and not withdraw money or anything. The broker would automatically withdraw my incentive fee and transfer it to my account. This model prevents the system from being known by the public or any potential market manipulators. But in any case, we can certainly talk about options.

About the drawdowns and expected return, let me paint a picture of how to think of the dynamics of this approach. Imagine a perfectly hedged portfolio where every movement is cancelled out by the movement of something else in the portfolio. What does the equity curve look like? A flat line. Now throw in some volatility due to some directional exposure or to temporary deviations in the correlations. The result now looks something like a sawtooth pattern with ups and downs but still following the flat line most of the time. I say most of the time because now we have probabilistic distribution of outcomes. Now consider the interest that the portfolio is expected to earn assuming no directional market movement. This is the expected return number I mentioned. If you drew it on a chart, it would be a line moving up and to the right. This is the same as saying that the flat line we originally had is now tilted up by the interest. Bring the volatility in and you have a tilted sawtooth. It zigs and zags but because it has an inherent upward pull, it will follow that line upwards like a staircase. This of course requires that the volatility is such that it doesn’t drown the interest rate. That’s the goal of finding the right mix. The 10% drawdown is describing one of those zags alone the line and isn’t the same as the volatility of the portfolio. When I say that I’m attempting to keep drawdowns in equity to 10%, it means that given a certain volatilty and expected return, I feel confident that a run of consequtive down weeks would not be likely to continue past 10%. In other words, if I put the return and volatility into a Monte Carlo machine, it would have maybe 5% of the results come back with drawdowns in excess of 10%.

On to rebalancing. No, I don’t care what pairs are best in terms of rate but what combination will lead to the best “value” of rate to volatility. This way I can use leverage to its best advantage. The flaw in most attempts at carry trading is that people go right for the highest yield pair (such as AUD/JPY or NZD/JPY) and make it the focus with no consideration about how well it can be hedged or how volatile it might be. They ignore the pair that’s “only” earning 75% on margin instead of 225% not realizing that one can often significantly lower the volatility of the “lesser” pairs through hedging and can then ratchet the return up with leverage to have returns that are just as good as AUD/JPY with less likely risk. There is another approach which is to ignore return all together and pick the “minimal volatility” portfolio. You’d still want to make sure the return was worthwhile relative to the RFR. I personally think this approach might actually beat an “optimal” approach because the higher volatility of optimal baskets might trigger more “no-trade” signals that could kill any advantage it might have. This portfolio I mentioned was, in fact, the minimal volatility portfolio and not the optimal one because the optimal one was on a “no-trade” signal. Can’t make money if your always on the sideline waiting. I think of the optimal as maybe being like T.O. He may be the best playmaker in the NFL, but if he’s always throwing tantrums getting him kicked off teams, he can’t very well score touchdowns.

The rebalance comes as a result of finding the basket that was optimal (or minimal) over the past so many months excluding the most recent few weeks. This basket is then validated on the most recent few weeks to see that it continued to “hold up”. This prevents one from over-fitting and falling into the optimization trap. Price changes are not looked at except in the since that they will affect the correlations that go into the model. This is a direction-neutral strategy.

I currently don’t have plans to use stop-loss orders beyond a very far off emergency switch. I don’t like them because I think they are the source of frustration in trading and careful risk management can be done without them. Brokers love stop-losses and I don’t like things that my broker likes. Watching the leverage and keeping it sane is my primary focus.

The result of the basket selection is by definition a portfolio with minimal directional exposure. I don’t want to make any money from market moves. Some might wish to enhance this style by doing so, but not me. If you look at the portfolio I mentioned, I’m acutally short EUR/GBP and long EUR/CHF (and to a small degree EUR/SEK). So there is hedging on the EUR going on there. Nonetheless, there will always be some directional exposure or the expected return would be zero. The key is to make it minimal enough that the pull of the interest rate is enough to swamp the directional gains/losses.

My only “don’t trade” indicator is the validation period. If it tells me that the new rebalanced portfolio I’m about to trade was a loser over the last few weeks, I sit in cash until it says otherwise. I’m sure you could try to time things but my goal was to make this non-discretionary and not try to beat the news game. The big thing to watch as I try this out is how volatile the rebalancing is. I don’t think it will vary that much from week to week so spread costs should be low. But if it goes to cash too often, I’ll be dropping whole portfolios too much. Another reason to go with minimal volatiliy rather than the more volatile optimal solution.

The spreads do vary. They increase on weekends for instance. The plan is to adjust on Sundays after the Asian market reopens for the week and spreads tighten again.

I hope these answers were clear. Just chat me up if you have more.

What will you be doing when you re-balance? Just find the best pairs based on interest rates and volatility, then move your money? Will you be looking at price changes over the last week (i.e., the USD is down, I’m more interested in using it as one of the pairs)?

Would you have stop loss orders in the market to limit your exposure to moves against your positions?

And is there anything specific you’re planning to use to avoid the risk of correlated positions? In your example above you’re long three different EUR spreads (with CHF, GBP, and SEK). How would those positions react if the EUR dropped quite a bit?

Any plans for a “don’t trade” indicator? For example, if volatility is over a certain threshhold in the USD (e.g., because Congress is debating some protectionist measures) you don’t take any spreads with USD as one of the currencies… Or if news is pending from the FOMC or BOJ on Tuesday, you avoid taking a position that week?

Any specific tactics for entering/exiting positions? The few times I’ve looked, the spreads on some of the currency pairs vary quite a bit during the week. Would you just enter/exit at the market?

You figured out the leverage you wanted based on the potential for drawdowns… with a target of a 10% maximum drawdown. What would the upside be from currency movements?

Let me see if I can re-phrase the question… if I could have a 10% drawdown, and the system would typically yield 10% a year from the interest payments, what’s the total expected return from the system? Are we expecting the currency pairs to appreciate in value as well as provide interest? What is the risk/reward ratio (aka expectancy) of the system as a whole?

As far as carry trades go, this seems like a good way to approach it. I read a quote from a hedge fund manager that 80% of his funds are allocated to carry trades (and similar cash generating systems), while the last 20% is used for macro trades. Get the basics for a carry trade down and you’ve got a base for a lot of other potential activity.

On the topic of selling a trading service (i.e., publishing trade recommendations to paying subscribers):

Setting up an online business and accepting credit cards is relatively trivial as long as you don’t mind using PayPal or other entry-level services. You can set up a password protected blog or something similar very easily — let me know if you want to make a conversation out of this and I can give you some pointers to resources.

Similarly, there are a couple of places like http://www.collective2.com that will do that for you (for a hefty fee, but hey, there’s no up front cost or cost of getting your system in front of potential customers). Collective2 caters more towards day trading systems, but I did see a few medium term services.

It will be interesting to see how well this plan works over the next few years because Japan may soon end its longstanding policy of 0% interest which will narrow a lot of the really lucrative carry spreads and who knows how long the USD will remain at such a high rate. But with the multitude of pairs available, there will always be a carry trade to make and so I’m not worried about longevity.

And yes, if this or any of my ideas works, I will likely offer it up for subscription or manage funds for those who don’t want to do the trading themselves or both. I also am looking into starting an incubator hedge fund, which is essentially a hedge fund with only one investor (me) that can build up an officially sanctioned track record before opening to investors. These are three very different approaches (giving trading ideas vs. trading for someone vs. being a trading vehicle that can be bought into) and I’m not sure which road to take yet. The first requires setting up an online business with the ability to accept credit card payments etc. The second option is really easy and if anyone I know ever wanst to begin investing, they’d just need to tell Oanda that I’m the listed trader for the account and what incentive fee to pay. I’m allowed to trade for up to 15 people I know without seeking sancition from the government. The hedge fund option would require a significant investment in legal fees but it goes after the really big money. In fact, by law, I would only be able to solicit investors with net worths in excess of $1 million. Hefty stuff indeed.

Assuming your plan achieves close to your expected yield, do you plan on posting (for a fee, of course) a weekly investment guide with suggested % allocations?