In a previous blog post we went in-depth into the explanation of relative strength using momentum measures. This blog post is a follow-up which includes further research and attempts to get to the bottom of momentum in the Forex market.
The main takeaway is that it’s easy to get carried away with indicators and testing – and it always pays to remember what exactly you’re looking to accomplish in the first place!
Weekly Blended Normalized Momentum
In the prior blog post, we had studies various permutations of momentum scores based on daily data, and in particular a blended 60 day, 22 Day and 5 Day approach. The blending is an attempt to stay in touch with current and medium term trend strength.
This time we decided to use weekly data directly, and we blended 2, 5 and 7 weeks, a timeframe that was suggested by Andrew Abraham and that is strangely similar to our own average time period calculation (23 vs 29 days). The Momentum ATR “normalizes” momentum by dividing the raw momentum score by it’s Average True Range.
More importantly, our Momentum ATR generates an average score of all 7 crosses for each instrument. So in the excel sheet below, when you see AUD, think of an average momentum score of:
AudUsd, EurAud, AudJpy, AudChf, AudCad, AudNzd, GbpAud.
This is an attempt at catching “themes” in the currency market. It’s our best guess for now. Here are the strong/weak combinations that our weekly Momentum ATR spewed out:
Going into April, the #1 pair would have been GBPCHF which proceeded with a decent performance. But we cannot let recency bias affect our judgement of the indicator.
Weekly Blended vs. Basic Macro ACD
Of course the indicator started picking up GBP strength (a recent theme) as of the close of March 25th. To check how reactive the indicator is, we went back and looked at an “old” version of the Macro ACD introduced in the previous article on relative strength. This “basic” Macro ACD only takes into consideration the strength & weakness of a currency vs.USD. Essentially it ranks the majors, period.
Interestingly enough, this basic model (which is however based on daily closes and as such is more reactive) picked up GBP strength as of the 19th of March. Around that time is when we had some members of our Trading Tribe picking up the strength across the board. More recently, it switched from a weak Cad stance to a weak EUR stance vs. GBP.
So far, we are seeing some correlation between the two models: they both picked up Cad weakness and (mostly) Jpy strength for the month of February. In March the Weekly model stuck with a short CADJPY stance and then switched to long GBP towards the beginning of April. The daily basic model switched between GBP and JPY strength in March, going back to a long GBP position in April also.
Basic Macro ACD vs. Average Strenth Macro ACD
The results have been interesting up until now. Before drawing any conclusions however, we need to compare the basic “majors only” Macro ACD with our recently built Macro ACD which averages out the strength of all 7 crosses of each regional currency. Here are the results for the month of March. In the displacement column is the Average Strength Macro ACD. In the Manual Rank column is the basic Macro ACD.
It seems evident that our programmed Macro ACD tended to be more stable during the recent past, at the expense of being somewhat stubborn and less reactive than the basic model.
Momentum is broadly defined as the change in price over a given time period. The farther price goes in that time period, the stronger it looks.
What we’re trying to build, with our momentum research, is a filter. As retail traders we have limited capital and need to deploy our capital wisely. We need to filter out the absolute best situations and ignore all else. One way of doing this is by ranking instruments of a same group based on their momentum, and allowing only the best 1-3 instruments filter through to the execution phase.
You may be asking the question: why do all this research in the first place? Why not use a dumb 10 period Rate of Change and keep it simple?
Actually questions like these keep us on track: when dealing with indicators & programming in general, it’s VERY easy to go astray and lose sight of the objective. So here’s how we’re doing our research:
- we want the indicator to spew out a selection of instruments that have “visually” been appealing in the past;
- we want the indicator to return a similar selection of instruments that we have been selecting manually with our trend/momentum model (because we’re trying to make our own process systematic – not because we think we’re infallible at selecting the best instruments);
- we want the indicator to be universal, and comparable across markets.
This is why we are doing research into this. But if you come up with better ideas, let us know!
Regarding the various indicators we have tested thus far, the discussion is this:
- fundamentally, so long as the indicator is doing a decent job at picking up strength & weakness, you can use whatever you want as long as you stick to it;
- it does appear that just by comparing the majors, we can get a good idea of strength & weakness within the FX market and the additional complexity of averaging all currency crosses doesn’t seem to add too much value at this point in time;
- the basic model (majors only) is more reactive than the other two models and this is at the same time a strength and a weakness.
On the negative side, we’ve been normalizing our indicators (making them universal) by dividing them by their volatility. This means that given equal momentum, the asset with higher volatility will actually show a lower reading. To generalize:
- declining volatility with equal momentum (a market that’s slowing down) will actually present higher readings than a market with higher volatility and this is counter-intuitive;
- ATR expansions and the trend in ATR is actually influential in showing the stregth of a trend. When interest dries up, ATR tends to decline.
So the last bit of research might be done in this sense: what is a good way to normalize momentum across asset classes? Are we diluting our momentum readings too much, by normalizing with ATR?
After these questions are answered, we will release our relative strength indicator – stay tuned!
About the Author
Justin is a Forex trader and Coach. He is co-owner of www.fxrenew.com, a provider of Forex signals from ex-bank and hedge fund traders (get a free trial), or get FREE access to the Advanced Forex Course for Smart Traders. If you like his writing you can subscribe to the newsletter for free.