An alternative formula for Dennis' Strong/Weak Analysis

In Dennis’ thread Trading the Trend with Strong Weak Analysis, a question was raised regarding the formula used to calculate strong/weak ratings for the 8 major currencies.

Kevin (screen-name k3v1n0s) suggested a correction to Dennis’ formula, and from a purely mathematical perspective, his suggestion is valid, and deserves a look.

As it happens, I considered the same issue a couple of years ago, and found (as Kevin has also noted) that the correction has little effect on the end-result of Dennis’ analysis, — that end-result being to identify the strongest and weakest of the 8 major currencies, for the purpose of considering a trade between the strongest and weakest.

For Kevin, and anyone else who might be interested, here is what I did in 2018, and what I found.

About 2 years ago, I decided to modify Dennis’ formula in two ways, in order to (1) determine whether the correction mentioned above would make a substantial difference in the ratings and rankings generated, and (2) to render the calculated numerical results as dimensionless index numbers, rather than percentages, in order to make them easier to comprehend at a glance.

In the formulas below, dividing by mAJ and mBJ makes the correction described above in (1). And multiplying by 1000, and discarding the %-sign, renders the index number in (2).

I rewrote Dennis’ formulas, as follows:

S(A) = (pAJ - mAJ) ÷ mAJ x 1000

S(B) = (pBJ - mBJ) ÷ mBJ x 1000

S(AB) = S(A) - S(B)

in which

AB denotes a pair in which A is the base currency and B is the quote currency

AJ denotes the A/JPY pair

BJ denotes the B/JPY pair

pAJ = the current day’s closing ASK price for the AJ pair

pBJ = the current day’s closing ASK price for the BJ pair

mAJ = the 200-period simple moving average on the 4-hr AJ chart at the current day’s close

mBJ = the 200-period simple moving average on the 4-hr BJ chart at the current day’s close

S denotes strength

S(A) denotes the strength of currency A

S(B) denotes the strength of currency B

S(AB) denotes the strength of currency pair AB

Then I produced a spreadsheet to crunch the numbers, display the strength index for each of seven major currencies versus the JPY, and display S(AB) for 28 pairs (all possible combinations of the 8 majors).

I compared my ratings and rankings with Dennis’ ratings and rankings each day. My study ran 5 days per week, for the 20 weeks from Monday, May 7, 2018, through Friday, September 21, 2018.

I won’t post all that data here. But, I will post the results for one day, to illustrate typical differences between the two methods of calculation.

On Monday, May 7, 2018, my calculated results versus Dennis’ posted results were as follows:

Over the 20-week study, there were occasions when my calculations produced reversals in the ranking of adjacent currencies (not in the #1 or #8 positions) having close numerical ratings. The NZD and EUR in the comparison above, highlighted with the red box, is a typical example.

There were 3 cases in which the #7 and #8 (weakest) currencies were reversed in our rankings, and one case in which the #1 and #2 (strongest) currencies were reversed. In all of these cases, the currencies switching positions in the rankings had very close numerical ratings – i.e., they were essentially equally tradeable versus the currency at the opposite end of the ranking.

Finally, my spreadsheet calculated the difference between the index value of the #1 currency and the index value of the #8 currency in each day’s results. This is the S(AB) term in the notation given above, wherein A and B are the #1 and #8 currencies (or vice versa) in the daily ranking.

I created a tentative trading plan which called for trading position sizes proportional to the magnitude of these differences. The rules of the trading plan were:

S(AB) is less than -40, then pair AB is a very strong SHORT – trade 4 lots

S(AB) is between -40 and -30, then pair AB is a strong SHORT – trade 3 lots

S(AB) is between -30 and -20, then pair AB is a weak SHORT – trade 2 lots

S(AB) is between -20 and -10, then pair AB is a very weak SHORT – trade 1 lot

S(AB) is between -10 and +10, NO SIGNAL – no trade is indicated

S(AB) is between +10 and +20, then pair AB is a very weak LONG – trade 1 lot

S(AB) is between +20 and +30, then pair AB is a weak LONG – trade 2 lots

S(AB) is between +30 and +40, then pair AB is a strong LONG – trade 3 lots

S(AB) is above +40, then pair AB is a very strong LONG – trade 4 lots

This can be illustrated on a number-line, as follows:

After 20 weeks, I ceased my attempt to improve upon Dennis’ method. Dennis’ formula works — and, if it ain’t broke, don’t fix it. Every trading day, Dennis posts data that any swing trader or position trader can use to earn pips on a regular basis.

Any one of us can write a proprietary formula for analyzing strong/weak relationships. But, we would be hard pressed to write one that works better, and more consistently, than the one Dennis has written.

And he offers his results here, for free, every day.

6 Likes

Hi Clint,

Very interesting! Thank you a lot for sharing the results of your personal study.

I totally agree that Dennis’s formula works in displaying the stronger vs the weaker currencies. If one relies only on the positions in the ranking, “correcting” the formula wouldn’t make any (valuable) difference.

It would only if the strategy relied on the percentage (or index) values themselves.

For example, if the strategy stated that one can only buy/sell EUR against another currency if it is displaying an absolute value higher than 1% in the ranking, there would be some instances when one would buy/sell too soon or too late.

But in the way Dennis is using his rankings, it doesn’t make any difference, really.

Thank you again for this valuable information, it was a very interesting read.

Cheers

When graphing the results after dividing by “close” and “SMA” the lines are identical so I don’t think it matters which one you use for Dennis purpose. Here are 10 rows of data from the USDJPY.

TIME abs(close-SMA)*100/SMA abs(close-SMA)*100/close
1 1.3243 1.3070
2 1.0944 1.0826
3 0.6791 0.6745
4 0.6663 0.6619
5 0.7561 0.7504
6 0.5392 0.5363
7 0.9628 0.9537
8 0.6972 0.6923
9 0.6235 0.6196
10 0.5269 0.5241