Paratrade Systems Weekly Research News

24 Apr

Does Volume Matter?

I am constantly reading market commentary that describes rallies with light volume as being less meaningful than those with heavy volume. I remember once losing a ton of money in a Japanese (Yen) bond position. When I called the broker who sold me the bonds, he tried to console me by saying that the decline had occurred on very light volume. I guess he was trying to say that only a few traders had sold the bond so when everyone else woke up and paid attention they would fix it all. Of course, that never happened.

Study I

If f the S&P declines and volume is less than yesterday’s volume then do we have a better reason to buy the market? This is easy to test. If we bought the market and held it for one day after it fell and volume was lower than yesterday’s volume then we would have accumulated 789 S&P points. Without the volume test, we would have captured 1624 points (since’05). If we waited for volume to be higher than yesterday we would have made 835 points. Clearly, a volume filter of this type does no good.

We can see why this is true in the chart below. The plot shows you the correlation between (yesterday’s volume)/(volume from 2 days ago) vs the next day’s price change. There is no correlation as you can see. The average value (though it’s somewhat volatile) is near zero.

volume blog1

The reason is because volume correlates heavily with the distance between high and low. The greater the day’s range the more volume we get so volume alone tells us nothing new.

volume blog 2

This correlation (between volume and range) is consistently elevated well above zero proving that volume and range are very similar. We are not here to test if a narrow range on a day has any forecasting value. (It actually does help a little.) Thus we shall change our variable to “relative volume” (volume/true-range) to see if it acts as a valuable filter. I will also show you the results with a pure volume filter in case you’re still not convinced.

Study II

We shall compare the returns of buying after a down day when relative volume is up above its 20-day average versus when it is below that average:

Volume blog 3

The only time high relative volume helped you was in the relentless ’14-’15 rally. Surprisingly the RV filter helped you avoid most of the ’08-’09 mess.

Now we can look to see if this helps on the short side. Since ’05, the low volume filter worked better than the high volume one. The former made 405 points while the latter earned minus 98 points. I wouldn’t exactly recommend this as a great way to make consistent money on the short side. The equity curve is very unstable.

Study III

We can look at multiple days to see if we get a clearer picture. First, we need a filter to compare it to. Here are the rules:

  1. Buy if RSI (len)<50
  2. Sell if RSI (len)>80
  3. Sell Short if RSI (len>50
  4. Cover short if RSI (len)<20

I will compare this with simply adding the additional volume test. I want to see low volume when RSI is weak to buy, and low volume when RSI is high to get short. Does that help? In the past twelve years, the rule set above made you money rather consistently:

volume blog 4

This is using a 3-day look-back and yes, it’s the best of all the lengths tested from 1-10. Here’s how it looks after we added a volume filter:

volume blog 5

Not so good…Now we can add a relative volume filter:

volume blog 6

This is not a tradeable system but it makes a point: Using a relative volume approach can improve your signals a little. Using a pure volume filter does you no good. Volume alone rarely tells you anything new. You might as well just look at the day’s range on your price chart.

Add Comment