mindmoneymarkets

Combo templates - When two is better than one

In a past post I blogged about the advantages of combining a short term mean reversion indicator with an intermediate term indicator of the same bent. Click here to view the post. 

As a recap, the DVI/B combo method helped smooth the ride with the longer term DVI modulating the shorter term and more erratic DVB. These indicators are available from DV Indicators.

This year the DVI will have had its work cut out in modulating the shorter term indicators like the DVB which came unstuck in August (though have enjoyed a good month or so recently). 

To help assess this better I've created two templates for the DV Indis Research package. The general rule is that you only trade if both indicators are signalling in the same direction. You stay flat otherwise. 

Template 1 Combines a short term and intermediate term indicator. 

Template 2 Combines a short term indicator with a longer term unrelated trend type indicator. 

Combo 1 - Short term and intermediate term indicators. 

I've used the template to test out various combination options. The first idea presented a few months ago was the DVB/ DVI combination. This helped soften the blow of the DVB's poor performance, but did not avoid it entirely. Following my recent post analysing the components of the DVI, I have been focusing my trading on the DVIM component and stripping out the DVIS. This combo option improves results further. 

Finally, using the DVO instead of the DVB has been another way to lift results. 

Combo 2 - Short term and longer term trend or skew indicators.

Logically combining separate indicators should offer better diversification and this indeed seems to be the case. The longer term trend filter, the DLTI has been a great companion for shorter term MR indicators helping to reduce drawdown and keep a good level of upside. 

Short term mean reversion isn't quite dead yet, especially when its used in conjuction with a wiser, slower moving companion. 

Templates available soon. 

Equity curve surfing & other filters

Following my last post on the struggles the DVB indicator is having, a reader reminded me of this excellent post Jeff Pietsch's market rewind blog. 

Worth a read certainly. The main highlights from this post for me are these three suggestions: 

1. Know what environment you are trading in (reverting/ trending).

2. Trading methods themselves revert (addition by David V.).

3. Track your methodology's equity curve like it was a stock.

In the context of the current market, I thought it would be a good idea to put some flesh on these ideas applied to the DVB Indicator. 

DVB Through the crisis

I've been out of action trading wise for the last month. I got married at the start of the month and enjoyed an active yet refreshing honeymoon in Peru for 3 weeks (We're not beach people!). 

Even in the Peruvian Andes I couldn't fail to miss the headlines from Greece and the ensuing market volatility. 

It looked good time for Mean Reversion type strategies which is typical when you’re aware from your screen. So I was keen to find out how a basic DVB strategy would have fared recently.

The basic strategy is to go long on a reading below 0.5 and short on a reading above 0.5.

Here’s how things have panned out recently using this method on the SPY (S&P 500 ETF). 

Not too shabby based on a rather simplistic eyeballing of the charts. 

 

DVB & TSI Ranking Part II

Following on from my last post looking at TSI rankings to choose the market to trade DVB, I thought I'd follow up on a couple of the ideas mooted in my last post. 

The general idea is to trade the DVB on the market with the highest TSI score (Between SPY, QQQ and IWM).

I noted that the market with the highest TSI tended to have the best returns on the long trades (as you might expect). Following a similar theme, when the IWM (Russell 2000) had the highest TSI, short trades as a whole under performed, especially on the IWM. 

So I looked at a hedge on short trades when the IWM was top. The theory runs that if the IWM has the

Yet Another Mean Reversion Indicator (Part 2)

In my first post on this topic I looked at whether intraday range could be used as an environment filter to help predict the favourability of Mean Reversion type strategies. 

Specifically I looked at the ratio of the low compared to the close as a % of the day’s range. Let’s call this the Low-Close Range. There is nothing magic or original in this calculation and I’ll explore this point a little more at the end of this missive.

The general logic was that days where the buyers manage to push up the market from the low are indicative of a strong bull market.

In other words, what percentage of a day’s range is taken up by the difference between the low and the close? If the bulls manage to put a big gap between the low and the close then is this a sign of bullishness?

Specifically, is this the sort of behaviour more likely to be indicative of trend following markets?

Is the reverse true? If there are periods with low % scores on average would this be better for mean reversion?

The candle highlighted below had a close to low difference that was 87% of the day’s range.

 

A day with a low % or zero would mean that the day pretty much sank from open to close with the day closing exactly at the low.

Test 1 – Average Low Close Range.

The first test is to look at the average score from X number of days. Since 1998 the 20 day average daily Low Close Range is 55%, that is, 55% of the day’s range is taken up by the difference between the low and the close on average. This makes sense because markets on average have an upwards bias.

So the test is based around an average score of 55%.

The chart below gives you an idea of what I thought might be having an impact. Prior to the recent turbulence, the 20 day average Low Close Range crept up to its highest level ever – extreme trend following!

The Mean Reversion condition is based on the DVB indicator. Go long on scores below 0.5 and short above 0.5.

All data up to March 9^th:

DVB Basic Long/ Short:

 

 Now here’s the performance of the same strategy if the average Low Close Range % is above 55%.

The strategy is still profitable, but the average day is pretty uninspiring with short trades loss making over the last 12 months. Long trades don’t seem to be too affected though, possibly as a function of the near relentless bull market prior to Feb. Since 2008 tough, the >55% condition hasn’t been great for mean reversion with sub standard average days.

 

Looking at sub 55% shows interesting results

 

Such periods are more volatile, but they are also more profitable on average especially since 2008. Long trades in the last 12 months have shown little difference here. 

Conclusion - It's not a massive effect, but since 2008 it does appear that Mean Reversion strategies perform better when intraday day trading ranges are more volatile as measured by the Average Low Close Range %. Most of these day's ocurr during a down trend which could be explaining much of these results anyway. 

Test 2 – Standard Deviation of the Average Low Close Range.

Going back to the blue line chart above of the average low to close % of range, it was apparent that prior to the recent turmoil, the average % was unusually high. Could there be something in looking at the average compared to a standard deviation away from the norm?

For this test I performed a 100 period standard deviation of the 20 period average Low Close range %. As its an average bound by 0 and 100 there's not much deviation, but there are still some interesting results.

As there is a low standard deviation I look at what happens when the 20 period average is 0.5 standard deviations above normal.

As a reminder, here's the baseline DVB performance on SPY: 

Average 0.5 Standard Deviations above normal: 

There aren't many days meeting this condition, but there are hints of an impact. 

If the average Low to close as % of range is 0.5 Standard Deviations above its typical level, it has been poor for the longs and good news for the short trades. 

Here's the opposition condition, when the Low to close as % of range is 0.5 Standard Deviations below its typical level. 

Slightly better for long trades and it shaves a few % of the average day & strike rate for short trades. 

Conclusions

There is not smoking gun in these studies, but there may be enough to warrant their addition to panel of indicators for position sizing. 

One reader commented that the low to close as a % of range was picking up on similar mechanics to many of David Varadi's indicators such as the DVB. He was right of course. However, in the course of this investigation I found a new appreciation for how the DVB (and other, similar indicators) works and why it works for certain markets and not others. I hope this little study offers something useful to you as well. 

For the record, here's an update of the DVI/ DVB Combo method as of yesterday's close: 

 

DVI/DVB + Seasonality Update

In December I took at look at using seasonality filters to improve on systems such as the DVI/ DVB Combo method. With financial markets starting to get a little more interesting, it's time for an update. 

You can find the original post here

First let's take a look at the performance of the DVI/ DVB Combo method. This basically goes long or short when both the short term DVB indicator and the longer term DVI indicator are singing from the same hymn sheet. Roughly the last year, no commission etc. 

The Combo method flat lined while markets went up and up and up, but performance picked up recently with the return of volatility. 

Now let's have a look at how those seasonality filters are helping or hindering. 

The returns look back around 14 months. 

These returns look back around 100 trading days: 

The seasonlity filters still seem to be doing their job quite well, primarily by keeping a long side bias. It would be interesting to see how they progress if we swing to bear market territory though. 

Yet Another Mean Reversion Indicator (Part 1)

With markets finally erm reverting this week, it's a good time to update on the success of the most basic of mean reversion strategies - namely buying after a down day and selling after an up day. 

'Mean Reversion' is a nebulous theme, but the core idea is to sell when the market is too hot and buy when its over sold. 

Mean Reversion (MR) started to become successful following the tech bubble collapsed and went into over drive during the credit crunch. Since then, things have become a little more pedestrian. This true for both the simple upday/ downday method and more sophisticated methods such as those using the RS2 or DVB. In geneneral the (former) inexortable uptrend didn't do these pull back strategies any favours. 

Here's the latest scores on the doors up until Monday 21st. 

Long after a down day on SPY (No commission or slippage). 

Since March 2008: 58% return/ Ave day 0.18%/ 2.22% Standard Deviation. 
Last 12 months:    13.3% return/ Ave day 0.13%/ 1.33% Standard Deviation. 
Last 5 months:      6.68% return/ Ave day 0.22%/ 0.81% Standard Deviation. 

The chart below shows this short only strategy since March 2008. 

Short after an up day on SPY (No commission or slippage). 

Since March 2008: 37% return/ Ave day 0.09%/ 1.59% Standard Deviation. 
Last 12 months:    -7.2% return/ Ave day -0.05%/ 0.88% Standard Deviation. 
Last 5 months:      -8.85% return/ Ave day -0.13%/ 0.66% Standard Deviation. 

The chart below shows this short only strategy since March 2008. 

 

SPY Buy and hold. 

Since March 2008: 21% return/ Ave day 0.03%/ 1.90% Standard Deviation. 
Last 12 months:    20.5% return/ Ave day 0.08%/ 1.08% Standard Deviation. 
Last 5 months:      14.78% return/ Ave day 0.15%/ 0.68% Standard Deviation. 

Short selling has not been a fruitful strategy in the recent up trend. You would have had to very carefully pick your spots to make any sort of return in the last 6-12 months. 

A different approach

One thing that struck me about the recent uptrend was the number of times the bulls were able to drag the index up off the lows. Could this be a better way of looking at activity rather than simple close to close comparisons?

I tested a simple calculation as follows: 

(Close-low) / (High-low)

In other words - What percentage of the day's range is accounted for by the difference between the close and the low? The further the day closes away from the low, the closer this percentage will be to 100%. The measure doesn't look at how today's close compares to yesterday's it's all about the abilty of the buyers to drive intraday action higher. 

I doubt this is an original calculation, but I still found it of interest. 

I used a score of 55% as being the dividing line. Below 55% is negative intraday action and above 55% is positive intraday action. The average day since 2008 is 55%, arguably 50% should have been the cut off. 

Let's have a look at a MR strategy using the following rules: 

Buy on a close below 55%:

Since March 2008: 76% return/ Ave day 0.22%/ 2.17% Standard Deviation. 
Last 12 months:    25.36% return/ Ave day 0.26%/ 0.37% Standard Deviation. 
Last 5 months:      11% return/ Ave day 0.37%/ 0.74% Standard Deviation. 

I won't bore you with another chart, but the returns from going short above 55% are similar but better to the returns from the simple short MR strategy (It has lost less money recently)

This different way of looking at MR improves on the simple upday/downday measure and still leaves room for further application. It's not a massively dramatic improvement on simple MR, but it is an improvement nonetheless 

Next...

Can this calculation be used as an environment filter for other MR type strategies? It seems so when looking at the average scores over different periods. In other words if a period has a high average low to close % of the day's range, it's indicative of a bull market. 

More to come soon. 

Mean Reversion - Theory

I spoke too soon about being settled back in. I'm nearly there but apparently my broadand supplier thought it was foolish for me to expect the stated 10 working days maximum set up time to be any quicker than 20 working days.... New supplier on the job.

An another note, I finished reading The guts and glory of day trading the other day and a thought struck me. It's not a bad read but its clear that a bunch of the traders featured simply got lucky riding the 90s bull market on leverage, then were lost when online dog food companies stopped doubling over night.

One thing that stood out were traders' reaction to the popping of the tech bubble. There were many quotes from the guys who were then paper millionares about how they shrunk their time frames to go in search of their trading edge. People who used to hold tech stocks for as long as the ride lasted started to look intraday, saying they had to look at the 60/ 15 minute/ 5 minute and even tick charts to find an edge. Holding over night certainly became a no no.

Around this time (2000), mean reversion rose to prominence (buy after down day rather than buying on strength etc).

A few theories have been put forward for why mean reversion started working so abrubtly post 2000 and why trend following worked so well before it such as the introduction of decimalisation. Now, file this under the obvious if you wish, but it seems to me that Mean Reversion kicked in because those day traders started focusing on the very short term, the good times were over and after being scared whitless with the (shock) stock market going down, any buying was done in a single day, rather than being spread over multiple days. Thus each day became a mini event with all the buying (selling) demand spent in a single day

Forward to 2003, trend followig started to kick in and traders (more cautiously) started to enjoy trend following again... Until 2008 when they were firmly reminded that stock markets do not only go one way.

So Mean Reversion may be a function of a shock factor, jolting investors out of their trend following easy life into grabbing quick gains when they can and not taking any overnight risk.

Once again, this may seem very obvious, but I wonder if this may explain why its hard to time when MR will switch on and off. It may not be a technical factor but a psychological one. 2010 has been a half decent year for MR, possibly thanks to the jolt enduced by the European sovereign crisis.

This isn't a scientific examination, but I looked at the returns of a Mean Reversion type strategy (the DVIB Combo) following the 1987 crash.

For around a year after the crash a MR type strategy was slightly more profitable until the trend followers got their mojo (confidence) back. Perhaps the MR switch in 2000 was so big because there were so many day traders (thanks to the internet) who were then able to become shorter term traders, again thanks to the internet.

Anyhoo, perhaps this theory is not exactly new, but it's helped me look at MR slightly differently and may shape the way I frame some more quant based investigations.