I have two recent articles on moving average strategies on SeekingAlpha.

Specter: A Finite State Trading Strategy
and
Peak and Trough Analysis

Peak and Trough analyzes price above or below EMA performance since 1930 with the Dow Indexes.

Needless to say, a holy grail wasn't discovered.

One of the technical issues that my work tries to deal with is how to express the current state of an equity with respect to indicator levels.

The analysis looks at 4 different indicators of 8 different lengths. 4 x 8 = 32.
The lengths can be variable but are currently set to: 3,7,16,32,57,93,142, and 206. All lengths are days.
The indicators are: E=EMA, M=SMA, P=Price percent in range, and R=Rate of change.

The differences in the lengths are based on the square of integers starting with 2. That is, 7-3 = 4 (square of 2), 16-7 = 9 (square of 3)...

The current price can be above an indicator level or not. The indicator level can be above the previous reading or not.

This is how SPY, QQQ look currently:



HexA shows current above indicator status and HexU shows the indicator ticking up status.

The characters are arranged by length. The coding scheme for each character is E=8, M=4, P=2, and R=1.

For SPY, HexA = xFFF00000. The first character "F" means that price is above all 3 day indicators. The second character F means the same for 7 day indicators and the third 16 day indicators. The 5 "0" mean that all the longer indicators are below.

HexU = xDDDFFFFF. "D" is hex for 8 + 4 + 1. That means that all 3,7, and 16 day indicators ticked higher yesterday except for 2. 2 is the status P.

The **Avg line shows statistics for a list of swing/day trading stocks that I have - more or less the usual suspects.

FFF00000 did pretty well on the long side yesterday.

Major ETF numbers for the first 67 trading days of the year are below:



The LogRet column is the key one to look at as opposed to the Return. YTD losses are down to single digits in many cases now.

This is how the major ETFs looked on 2/21/20.



HexX tries to simplify the readout. The lengths are 3,7,13, and 19 and each length corresponds to a character in HexX. I want to test this length combination against Specter which is best with short term averages.

The lights get scored: 8 = xDayEMA above, 4 = xDaySMA above, 2 = xDayEMA ticked up, 1 = xDaySMA ticked up.

The HexX readings are quite weak the Friday before the crash started.

Here is the current period.



Some relative weakness of QQQ, SMH, XLK the last few days and strength in IWM.

Somewhere between 285 - 290 in SPY is the 61.8 fib retracement from the February high to the March low. Maybe getting a bit elevated.

It seems important to express the hexadecimal characters in a meaningful way. This is my latest attempt using yesterday's data:



HexA and HexU have morphed into xEM and xPR. E=EMA,P=SMA,P=Price in Range, R=Rate of Change.

For both columns, each character represents a length. The first is 3, then it goes 7,16,32,57,93,142, and 206.

The high order bit (= 8) for each character is ExA - Price is above EMA, the second (=4) is MxA, the third (=2) is ExU - EMA is higher than the previous day's reading, and the fourth (=1) is MxU.

F (Fox - 1111) means everything is positive in the given length. 0 (Zero - 0000) means everything is negative, the opposite of F. These are the most important readings.

C (Charlie - 1100) is a critical reading, that means that prices are above both averages but the current day's reading is below the previous day for E and M.

3 (Three - 0011) is the opposite of Charlie, it indicates that price is below both averages but both reading have improved from the previous day.

Note how many of the readings changed from F on Thursday to C after the loss on Monday. Monday's action was quite bullish even though the market finished down. The decline was not enough to change the position of price to the averages, but the Up readings slipped.

Needless to say it is a bit dangerous here. 290 SPY is the 61.8 fib retracement. A decline to 260 is quite possible but that level is not one where an investor would want to exit. That would be a lost off $3000 on 100 shares and once it gets there one could start worrying about 230. Sometimes you have to risk getting bitch slapped.

I guess this confirms what I've read from Larry Williams that MA's aren't a reliable strategy over the long term. Holy grail search continues lol

I sold Wednesday 276 ATM covered calls on SPY at the close on Monday for 3.41. Figured the market would go up yesterday but wanted the ITM protection for Wednesday. The plan worked for a change, had I been only long yesterday, God knows what I would have done to protect my position.

Here is the status of the major ETFs as of yesterday.



I'm a fan of Larry Williams. He made a StockCharts youtube video about range expansion a few weeks ago, where he asserted that the Fed was strong enough to prop up markets. That impressed me enough to be comfortable long during the bounce. The 61.8% fib retracement of SPY to about 290 has been on a lot of people's radar.

I'm planning to look at range expansion soon. Range is Tmm in my table. That is from a fixed timeframe (of 70 days) though. One needs other time frames (like the 8 I look at) to make it work.

Larry was a big supporter of the gold standard when he ran for the Senate and mentioned in the video that he now thinks he was wrong (duh). It's OK to believe in the gold standard, it is not the stupidest thing in the world; but it fascinated me that he changed his mind about something he must have been more sure of than the presence or absence of significance in moving average numbers.

Personally, I'm not searching for the holy grail, which is why I mentioned that in my OP as a courtesy to any gallant knights searching for it; so that they would not waste time reading my stuff.

The numbers deteriorated a little today. I played the little rally yesterday at 2:30 EDT to the long side but that got pushed back around 280 SPY, the opening 30 minute bar went back up there today but that also looks like it will be rejected. I am back to modestly long covered by ITM calls expiring tomorrow.



The deterioration is not necessarily bearish. The next step with the hex codes is to modify my strategy algorithm to analyze various values in the columns, specifically F, 0, C and 3. Needless to say, the current algorithm is pretty good and can also handle those combinations but this current exercise seems to be worth the time investment.

The prototype strategy engine works. A sample listing is below:



xEM between the two x's are the readings by length. After the second x, the 2 hex characters show ema above sma for each length. For example the 4 for SPY shows that the 7 day ENA is higher than the 7 day SMA. The 3 day, 16, and 32 day are all lower. That isn't used in the study listed here.

I call the algorithm FoxForceCharlie. Columns starting with X are log returns for the five different strategies analyzed.
XFL7 goes long when the 7 day indicators are all positive - 1111. X0L7 all indicators are negative - 0000. XCL7 price is above both E and M but the amount has ticked down compared to the previous day - 1100. X3L7 prices are below averages but they are ticking up - 0011. XML7 shows the results of the other hex possibilities.

Columns starting with D are the number of days the market was in each state. The X columns add up to the LogRet for each row. The D columns add up to 40.

Interesting that one could have done pretty well over this difficult period by staying away from X3L7. X3 looks like the weakest consistently with other lengths. X7 X0 and XC consistently do OK. The day distribution is a little skewed because of the small sample size. XM usually stays around 11% of the days. XF always has the most days - except for small samples like here.

My latest article is about FoxForce5

The posts I've done here after the original one, documents the thought process.

There is an Elite Quantitative forum that I will probably use to make future posts.