Regarding the error I found yesterday. It is about the "buffering" to avoid unnecessary trading related to signal errors. I found I had run the rounding operation BEFORE the buffering calculation, in calculating target number of contract. Through some exercise, I realized that the rounding needs to be done AFTER, to get the buffering to work effectively. I made this change accordingly yesterday.
I am also considering modifying thresholding process a little bit. I was wondering I might be more benefitted from the thresholding if the flatten part of the line (the forecast - multiplier relationship line) is narrower. The charts below compares target number of conctract comparison, with and without thresholding (with: red, without: blue). I asked related question to Rob yesterday via his website. This charts may explain background why the question (if he sees this). Modifying the slope (red dotted line) to start from (6.67, 0) (instead of current (10, 0)), then it looked that I get bigger benefit. Edit: color tag of the chart was wrong. now corrected.
I am sorry the image contained blank areas...but good that image capturing worked. here is the modified version (slope starts at (6.67, 0)), for reference. You need to look carefully as the diffence vs. the previous charts is small (then why bother? may be a good question!)
then why bother? may be a good question!) It's exacly the question I would ask, that's for sure... GAT
It can work pretty well; I priced up some of the more liquid US ETFs and they were competitive cost with some futures. When I used to work for AHL we used them mostly for EM equity indices for which there were no futures. A nice side effect of that was that the corporate entertainment was much better from ETF providers than from futures brokers. The obvious disadvantage is you can't get leverage except with portfolio margin, and that can push up the holding costs somewhat once you are paying retail level borrowing fees and interest costs. So for example for the 'starter system' (basically EWMAC16_64) in 'Leveraged Trading' the SPY ETF comes in at 0.02 SR units compared to say 0.004 for Corn futures. GAT
So one would have to limit oneself to "long-only" to avoid borrowing cost. And use one of those no-commission brokers. If you then would use an ETF and its inverse ETF (assuming that it exists) then you might be able to reduce the negative impact on SR. It sounds too complicated to me to try out though.
Thanks for the answer Rob. Indeed, i forgot about the interest charged on margin. Using etfs is then beneficial if futures are not possible, and if the increase you gain in sharpe by adding another market is > 0.02. I wonder if you've done a study on the marginal increase in sharpe you gain by adding another market. I suppose the increase won't be the same if you already trade 40 markets as opposed to 3 markets. Thanks!
If you trade on margin (which will most likely be the case for small accounts), i think you'll pay borrowing costs whether you're long or short.