19th Ave New York, NY 95822, USA

A Possible Edge On High Volume Days

LOREM IPSUM DOLOR SIT AMET

In this article I will show that there is the possibility of a profitable edge when trading breakout stocks on high volume days.

However, the edge requires an ability to foresee such high volume in advance. This is no easy task but the data suggests it could be an interesting line for future research.

High Volume Days Trading Strategy

The rules of this high volume trading strategy are simple.

We are going to buy a stock on the open if the day’s turnover (volume * close) is more than twice as large as the 100-day average and the stock closed the previous day at a new 10-day high.

Similarly, we are going to short a stock on the open if the day’s volume is more than twice as large as the 100-day average and the stock closed the previous day at a new 10-day low.

Trades are held for two days and closed on the open of the third day. Turnover is used here as an alternative to volume which can be affected by stock splits etc.

Buy Rules:

  • Yesterday’s close was the highest close in 10 days
  • Daily turnover is > twice the 100-day average
  • Buy on open

Sell Rule:

  • On open of third day

Short Rules:

  • Yesterday’s close was lowest close in 10 days
  • Daily turnover is > twice the 100-day average
  • Short on open

Cover Rule:

  • On open of third day

Trade Example

The following chart shows a trade example in RRD from July 2016. Our entry is made when turnover (volume * close) crosses the 100-day average and when the previous day is a 10-day high.

RRD trade example high volume days

System Rationale

The idea behind this trading system is the belief that a good percentage of stock returns come from high volume days and price movement that occurs right before a high volume day might be instructive of future price movement.

For example, if a stock goes on a 10-day rally right before a big event like an earnings announcement, that performance may reveal important information relating to the earnings announcement itself.

It could be the case that insiders are buying the stock in anticipation of a good earnings report. Or that rumours have surfaced indicating that the earnings report will be better than expected. There is in fact plenty of research in this area with many papers on SSRN. This one for example.

System Results

This trading system described was run on historical end of day data from Norgate between 1/2010 and 1/2017 on a universe of S&P Mid-Cap 400 stocks with commissions of $0.01 per share. Starting capital was $100,000 split between 10 positions.

We recorded an average profit per trade of 0.34%, annual return of 19.3% and a CAR/MDD of 2.07 with a win rate of 56%.

Here is the equity curve:

high volume days equity curve amibrokerSome Initial Thoughts

The idea behind this system is that stocks breaking out right before a high volume day are likely to continue in that direction once the high volume day takes place.

Results appear to back up this thesis and show a profitable edge but there is big problem with using the system we created. We are trading on the open before we know how much trading volume there is going to be. Thus, there is the problem of look-ahead bias.

We cannot reasonably trade this system as we don’t know when a high volume day is going to take place in advance. Furthermore, when this system is backtested with a trade delay of one day, it loses all its profit potential.

However, I have chosen to illustrate this system because of one fact. In some situations, we will have some idea of what volume will be like.

Consider earnings announcements as the primary example.

If we know a stock is going to announce earnings tomorrow, we can reasonably assume that the stock is going to generate a lot more interest than usual and therefore volume is going to be higher than usual. And if the stock in question just closed at a new high then maybe smart investors know something we don’t.

Thus, it could be said that to trade this system we do not need to predict price movement, but trading volume.

Example Of TUP High Volume Setup

In the following cherry-picked example you will see that TUP closes at a new 10-day high on the 24th April right before a high volume day. The high volume day takes place on the 25th and you can see that the stock surges higher.

Our system bought on the open of the high volume day and then sold two bars later for a profit of 2.78% after fees.

TUP example high volume days

Normally we could not predict that the 25th of April would have been a high volume day so we could not have made this trade using price data alone. However, as it turns out TUP was scheduled to report earnings on the 25th.

Since we know that TUP is reporting earnings on the 25th we know it is likely to be a high volume day. Thus, we in fact could have bought on the open or even on the previous day’s close expecting that the event is likely to cause above average trading volume.

The rationale is as follows: we know TUP is reporting earnings tomorrow and we know the stock has rallied in the days running up to the announcement. So we go long expecting a high volume day and a positive earnings reaction.

Final Thoughts

The trading idea presented here shows that there might be a possible edge when trading high volume days. Such high volume days often include earnings announcement days or other major news events that are scheduled into the trading calendar.

However, it should be said that the results are so far based on a lot of assumptions which means that it is probably not a good idea to go out and trade this method.

For example, once you take out all the scheduled events like earnings reports it could be that the strategy loses its edge. This would make it worthless to trade. Furthermore, even if an event is scheduled in we cannot guarantee that volume will be more than twice the average.

Overall, it seems unlikely that all you need to do to make money is to follow the trend right before scheduled news events.

To test this strategy more thoroughly it is necessary to obtain more data. Earnings announcements data, for example, would give the ability to isolate trades only to scheduled earnings days.

Such data can be obtained from places like Zacks Research and EventVestor. This could be an interesting line of future research.

Example Amibroker Code

/*Code is provided as an example and for informational purposes only. Full disclaimer and risk warning applies. JBMarwood.com. */
SetFormulaName("High Volume Days");
SetOption ( "CommissionMode", 3 ); 
SetOption( "CommissionAmount", 0.01 );
SetOption( "InitialEquity", 100000 );
SetPositionSize( 10000, spsValue );
MaxPos = 10;
SetOption( "MaxOpenPositions", MaxPos );
SetBacktestMode( backtestRegular );

PositionScore = Ref(RSI(10),-1);
Turnover = C*V;

HVBuy = Turnover > 2*(Ref(MA(Turnover,100),-1)) AND C1<Ref(LLV(C1,10),-1); 
HVSell = Turnover > 2*(Ref(MA(Turnover,100),-1)) AND C1>Ref(HHV(C1,10),-1);

Buy = HVBuy;
Sell = 0;

Short = HVSell;
Cover = 0;

ApplyStop(stopTypeNBar,stopModeBars,2);

BuyPrice = ShortPrice = O;
SellPrice = CoverPrice = O;

Simulations in this article made with Amibroker using data from Norgate.


Comments (6)

Hi,

if you have intraday data, you could backtest opening position with the following trigger: intraday cumulated turnover (for current trading day) > average over 100 past days?

Good idea. Actually, a lot of the volume will probably take place on the open. I remember reading an article once about estimating the daily volume based on the first minute of the day.

I looked at this myself a couple of weeks ago. However, what I found on flipping through the charts was that you might get a spike up in volume and gap up but it would fall over the next day or two later. Tend to suspect short sellers or hopefuls are pushing the price up. On the flip side, you look for these spikes, wait a day or so (intra-day wait a couple of hours after open to make sure it is going down) then go short. These spikes are particularly evident when the stock has been going sideways for a while before spiking.
Its also important to take note of the overall trend
Maybe a % profit target would work, though it would be tricky and perhaps lose out on profits following exit.

Fading the big spikes seems like it would be a profitable strategy from looking through the charts but testing hasn’t shown much potential so far. This seems to back up the efficient market theory relating to the pricing of new information.
That being said, the strategy does work a lot better in small cap names especially when the stock is up more than 50 or 60 per cent in just a few days. One of my own strategies is similar to this, albeit it is a risky one.

Hello, thanks for the strategy. I have question. I do not understand part of the code specifically turnover = C * V. How can I open position at open when I don’t know its close or volume yet? Should be there a close and volume the day before?

This is explained in the article.

Leave a comment