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: