Friday, October 14, 2016

No Clear Directional Bias Currently Based on My Backtests in SPY

Yesterday and today I have looked a several scans investigating MACD position, stochastics, bollinger bands, 3 month low reversal attempts, real money sentiment measures, and the balance I would have to say is mildly bearish looking forward a couple months from yesterday's close in terms of MAX losses versus MAX gains.

Basically it seems the the intermediate term of the next couple weeks or possibly longer has some downside currents.  So real short term or swing style trading may be best on selling rallies.

It would be nice to see more of a hard sell off in order to set up a bullish rebound attempt.  The best trades I ever see in my scans are bullish set ups after hard sell offs.  Even in down trends there are often sharp rebounds for a few days, and the weekly options offer attractive opportunities, and the upside skew can be very nice for ETF plays with wide stops and limits.


What If SPY Breaks to New Corrective Lows From Here? 10-14-16 Stock Market Update

Click on Chart to Enlarge

First let me immediately state that this study is NOT currently active.  It is a "what if" scenario.  What if SPY ends up breaking this Thursday's low with a weak close in the bottom half of the range?

That would be a reversal failure type scenario because yesterday price broke to a 3 month low, and reversed to close above it.  So basically I am posing the question of what happens if a reversal attempt from a low (with bullish divergence in sentiment) fails to form a low which holds.

What the table demonstrates is that a break to new lows would suggest a bearish skew looking forward based upon historical comparisons from the last 21 years of SPY history.

The skew is pretty strong out to 2 months later, with greater MAX declines on average than MAX gains.  Looking at the 2 week forward time frame, there have been some significant declines following these breaks.  If trading the options, the maximum expected value play for my system involving at the money options, would be to buy an ATM put with 2 weeks until expiration, and set a limit order gain of 200% to exit the trade.  The win rate on the past instances was only 44% and the Kelly Bet was low at about 15%.  So it would justify only a smaller position but the expected value was strong at ~32%.

What about the equity side of things?  Well first off there is a difference between just shorting on a stop order to new lows versus letting the break to new lows happen and then waiting to see if prices are able to again reverse higher.  The backtesting suggests a stronger downside skew if prices break to new lows AND close in the lower half of the range.  So that is the scenario the above stats are for, and what the strategy would be here on the ETF side.

Based upon my system and backtests, if we entered short SPY at the close of a failure day as discussed in the above paragraph, and set a stop order 7.75% above prices in SPY, and a limit gain order 7.75% below prices, but exited at the close 9 trading days later if the orders weren't hit, it would produce a nice 2.36% expected value.  That is using SPY unleveraged.  The back test stats justify 3x leverage on the short side in this case though.  And the win rate was 64% on the past 25 instances.

So we could expect about a 7% gain on average using the 3x short ETF like SPXU or SPXS.  

That is a very nice set-up for profiting quickly in a down market.  

So if there are questions on this let me know.  But if you use alerts, you could set an alert on SPY if price moves to 211.20 or lower to make sure you understand that price is breaking to new lows and this set-up will be in play if the day's close is weak (below the mid point of the day's range).


Thursday, October 13, 2016

Another Small Sample Study Showing Potential For Sizeable Sell Off - 10-13-16

Click on Chart to Enlarge

Today as of the open, SPY is opening below the lower bollinger band.  With no other filters, this is a neutral occurrence in terms of forward returns based upon backtesting.

Adding a stochastics fast line to be below 20 still shows a pretty neutral forward skew of max gains vs losses.  

But the table above adds just one filtering criteria which is that the previous day closed UP.  In our current case SPY made a small close up yesterday and closed above the bollinger band.  Now today we are seeing an oversized gap relative to the range in order for it to open below the lower bollinger band.  

In the past instances, the table demonstrates another significant skew to the downside in the short term.  In this case, as with the study on MACD and stochastics crossing down on the same day, we see some pretty sizable short term sell offs beyond random MAX draw downs.

Now most of the losses occurred in the first 3 days proportionally.  So, I am just making further noted here that we could see a swift decline.  But that decline would very likely produce some back test studies which offer bullish trade opportunities.

Again this is a smaller study, and stocks certainly could rally right fro mthe open, but be aware of the context here and possibility for sell off.


Tuesday, October 11, 2016

Another Small Sample Bearish Study Triggering Today

I ran a scan which looked at

  • daily 14,3 stochastics crosses down from above 50
  • weekly MACD is down
  • daily MACD is down
The results showed 20 instances going back to Sept 1995.  The MAX loss was 4 times as large as the MAX gain over the next 2 weeks.  This signal is occurring today.

When I added the conditions of the daily MACD line being below 0, the results were also very bearish.  And when I added the additional condition of today being a cross down on the daily MACD (from below 0 still) there were only 4 instances going back the 21 years.  

Dates were

8/27/1998 - ~9.5% loss within 3 days
10/4/2005 - ~2.5% loss within 3 days
8/19/2015 - ~12.5% loss within 3 days
9/26/16 - sideways trade for 2 weeks

So check these on your chart.  The most recent one 2 weeks ago led to nothing of significance.  Just a historically unusual sideways pattern.

But the other 3 had massive skews to the bearish side.  About 10:1 skew to the downside over the next 2 weeks.

I know it is small sample size.  But may be of note.

Some other studies looking at volume increases and price declines similar to today show a mild to moderated bullish picture for 1-2 weeks on average.

So it seems like the most probable scenario is a modest rebound over the next 1-2 weeks back up into the range we've been in.  But there may be a greater than typical risk for a quick sell off of large scale.

I will keep this under watch and post if there is relevant info to follow.


Investment Banks Look Like They Will Sell Off Hard In Coming Months - 10-11-16

Click on Chart to Enlarge

This chart is Bank of America, BAC and is a weekly chart with the stochastics visible beneath.  Here we have basically a picture perfect pattern formation of a major top reversal.
A few points of observation:

  • Notice the decline off the 2015 highs created a larger (and faster rate of decline) decline than any pullback during the bullish run since 2011.  This creates what Gann referred to as an overbalancing after the end of a price trend.
  • The broad market indices like SP500 and Nasdaq have made new all time highs in 2016, but see that BAC has not rallied back to new all time highs.  I call this a non-confirmation of the broad market trend and implies future weakness in the stock.
  • The rally since the lows in the winter, has taken longer to form than the preceding decline, both from the all time highs, and the secondary lower top in Nov 2015.  This price action implies a new downward trend in that the larger and faster moves are in the downward direction.  the explosive, higher rate of change moves have been to the downside.

Click on Chart to Enlarge

This chart is a daily chart of BAC, so it is zoomed in some.  Further observations....
  • Notice the overlapping nature of the move up since February.  Boundary lines for the move up since February would form a rising wedge type pattern.
  • The rally is now peaking in an area of prior price congestion and multiple gaps downs from the winter decline.  This could be a chart/price based overhead resistance zone.
  • The MACD on this chart shows a major bearish divergence with price at the price peak yesterday.  This divergence at a lower high (than the beginning of the prior declining move) implies exhaustion and completion of a price move at a lower high, again indicating a downtrend likely to continue.
  • The recent poke to new highs and reversal below the old high, creates a mini failed breakout or double top, which may be a final smart money selling point as they sell into the break to new highs and attempt to unload right at the top.
From this point, a high quality top reversal pattern would be unlikely to make any further advance.  If a major price decline is to unfold, it would be ideal or not uncommon for price to gap down through the rising red trend line of the recent leg up since this summer bottom.

Assuming there is a sharp break of the red trend line, it would be quite common for price to rally up to very near the price level where it broke the trend line.  That would often occur with a daily stochastics rally to above the 50 mark.  And the most common short term reversal candlestick to occur at that backtest point on the chart would be a bearish engulfing pattern based upon my observations.

This quality of pattern will likely retrace back to at least the mid point of the recent rise.  It would take a major correction or bear market environment to fulfill a complete retracement of the rally since February.  But given the current position of the broad stock indices, the bull market could be topping here.

Similar comments apply to charts of SCHW, MS, AMTD, ETFC, CFG, C.

This will be interesting to observe in context of the upcoming US election, and a potential reversal of the long period of no interest rate increases by the Federal Reserve.  The next FOMC meeting and announcement of interest rate changes is November 2nd.  Election day is the following week on the 8th.

So it will be interesting to see if the Fed goes ahead and makes the rate change at the upcoming meeting just prior to the election.  But if not possibly wait and make the change either after the election or after inauguration.

From the observation of these price patterns on major national financials, it seems like the time for an "event" catalyst to bring the next major wave down.  


Thursday, September 29, 2016

MACD Whipsaw This Week - BackTest Results

The daily MACD this week made a couple whipsaw crosses.  It crossed up on Friday, then down on Monday, then up again on Tuesday.  This type of cross in one direction followed the next day by a cross in the other direction is not very common.

Going back to 1995 in SPY I only found 4 instances where there was an up/down/up pattern on three consecutive days.  The back test results did not seem significant.  No major skew in the forward returns, and small sample size.

When I look at just the down/up pattern on back to back days, I found 21 prior instances before this week.  In this case there was a definite bearish skew, with some large decliners in there, and the forward return showed a bearish skew with the forward MAX loss being over twice the size of the MAX gain at time frames from 3 days out to 1 month.

When I filtered the results for the occurrences where the MACD line was below 0, the results were even more bearish (this condition matches the current market) but with only 8 instances.  Only 1 out of the 8 had MAX losses less than 3% in the following month.

When I filtered for the weekly MACD being in a down position instead of the daily being below 0, the results were roughly similar with a pronounced bearish skew.

So, the first couple days since the signal in the last post, have not been typical of the strong negative skew suggested by that small sample size study.  But here again we see a piggy back study with a strong bearish skew, but small sample size.

Anything can happen, but I currently see no really great reason to abandon the previous trade currently.

As a side note, when I reversed the signal order for the whipsaw to up/down on consecutive days, there were only 13 prior instances, and the results were shorter term bearish, but more positive looking forward several months. 

My take away here is still to favor the downside for upcoming weeks.


Monday, September 26, 2016

New Put Option Trade - Stocks Could Decline Dramatically The Next 1-2 Weeks

Click on Table to Enlarge

The table here shows the results of the following back test criteria:

  • daily MACD crosses down today
  • daily 14,3 stochastics crosses down today from above 50
  • weekly MACD is in the down position (fast line below slow line)
A relatively simple scan, but shows the impact of a market that is having shorter term technical sell signals with the larger current pointing down.

Four trading days later all 8 instances closes lower.  The minimum was 0.5% and the maximum was over 10%.  But looking at the MAX gains versus loss over the coming days and weeks, I rarely see anything this strong. 

The instances are not that high, but I don't think that means this sample is insignificant.  Some studies are rare, with small sample size, but are obviously notable events.

I purchased both a 214 strike Nov 18th expiration put as well as a 214 strike Oct 7th expiration put.  Based on the stats, this Friday's expiration certainly could be used as well.  The stats are about par with the 2 week stats as far as option % returns expected.

In this case, the past instances suggest a limit order of 160% gain on the put option would produce a very high expected value and win rate.  While the sample size is small, I am going ahead and placing that limit order.

I am going to wait for a couple days before placing the limit order for the Nov put.  I think the 199 level on SPY could be broken by November.

Trade Idea:
Buy the SPY Oct 7th expiration 213 strike put at the open tomorrow.

Additionally, a short/inverse trade could be entered here.  Using a 7.75% limit gain order and stop loss order (can enter as an OCO if available), would produce the maximum expected value based on the past test.