Ok guys so some ideas I'd like to coin here:
1. Rapid open and closes
Mathias has helped me with the quick open and closure problem issue. First off I had it completely wrong with setting Min Diff = - 1 as exit condition for the CSS{H4}_Diff_Min. He coded the reverse logic like this:
- Entry for a BUY: CSS.XXX - CSS.YYY > Entry_Min_Diff
- Exit for a BUY: CSS.YYY - CSS.XXX > Exit_Min_Diff
Therefore, the Min_Diff at both places should just be 1 et voila - fixed and ready.
Secondly, it'd also help if we upped the parameter 'Seconds until an Exit signal is stable & valid' from 30 seconds to 600 seconds (i.e. 10 minutes) in order to remedy this rapid open and closure thingy. I've started with setting this to 300 seconds (i.e. 5 minutes) for a start. Currently all is stable and will report soon to see if this helped. Ultimately I'd like to bring these Min_Diff (for both the entry and exit) and the exit stabilization parameters down to the tightest yet solid values. So far all fine and dandy.
2. Risk and money management
In order to stabilize and enhance margin-use and risk, two things come to mind:
- Limit your exposure per currency
- Stable yet dynamic lot-sizing
With regards to the first point, if you think about it: there's really no point in trading all 28 pairs (for 8 currencies, we have 28 combinations). There's a certain degree in unnecessary risk and transaction costs (i.e. commission, swap, etc.) present if we'd be trading all of them. If a move in a currency happens, then all the currency pairs in question should normally exhibit the same move. This HAS to happen, else there would be traditional arbitrage opportunities in the markets all the time and we really wouldn't be having markets at all actually (HFT robots do occasionally pick up super quickly arbitrage opportunities when they establish themselves and only the best and most competitive robots will be able to benefit from them first).
Now firstly we must ensure that we are involved in
all 8 currencies, such that if a move happens for a particular currency we'd be ready and have a chance to benefit from it. Secondly, we must aim to trade a well-balanced set of symbols (trade all currencies in a balanced way). Thirdly in general our approach should work best for
pairs that move relatively a lot. A sensible approach is to work from the top downwards, with every step excluding a currency
at least and at most once!
- full exposure to 8 currencies: every currency is traded 7 times -- trade 28 pairs
- reduce exposure with 1 'unit': every currency is traded 6 times -- trade (28 - 4) = 24 pairs
- reduce exposure with another unit: every currency is traded 5 times -- trade (24 - 4) = 20 pairs
Now to choose which pairs to actually exclude, I made use of Rene's nice VolatilityChart tool. I've set the ATR period to 125 and ordered the pairs on their Daily time frame. Why 125? We have roughly 250 trading days in a year and thus 125 would be exact a half year's worth of trading days. I use this also for the lot sizing component; short periods are a bit too volatile for me and prone to get skewed significantly when things such as brexit and a flash crash happen. Precisely because these ATR-values are the
basis from which lot sizes and TPs and SLs are determined (via respective ratios), it doesn't make sense to me to have this basis value potentially swing too much and have for example 50 pips as a TP one day and 95 pips the next week on the same pair. Using 125 days works really well in that it enables to characterize pairs on a pair-specific (interpair) level, is averaged out nicely and YET is dynamic on a
intrapair-level too by capturing slow but sure changes in a pair's character over time! As we all know, the way a pair behaves doesn't change overnight (nor over a fortnight), but if you'd compare EURUSD today with EURUSD in 2008 you'd see some serious differences in overall volatility.
That being said, I ordered all 28 pairs according to their Daily 125-period ATRs and picked out the lowest values of pairs to be excluded.
- First set of 4 pairs to exclude with the lowest 125-period daily ATR: EURCHF,AUDNZD,USDCAD,GBPJPY
- Second set of 4 pairs to exclude: CADCHF,AUDUSD,EURGBP,NZDJPY
So ultimately you'd have the 20 most volatile pairs to trade and be equally invested (to keep the balance) in all 8 currencies.
Cheers
Aram
You do not have the required permissions to view the files attached to this post.
