Friday, January 23, 2009

Relative Volume and Volatility: Understanding Who is in the Market



This is an important post on a topic I rarely see discussed. As background, you might want to check out my posts on tracking large traders in the market and the volume realities that every short-term trader should know. The reason volume analysis is important is because the majority of volume--and the shifts in volume--can be attributed to the participation or non-participation of large market participants. These institutional traders have the resources to move markets, so it is critical to understand when they are relatively active and inactive in the markets you're trading.

I took a look at the last 14 days of trading in the ES (S&P emini) futures contract and found that the volume of each 30-minute period during the day correlated with the size of the high-low price range for that period by .78. That means that about 60% of the variance in volatility can be accounted for simply by knowing the volume traded during that period. When we identify whether volume is high or low for a period, we're also able to make an estimate of likely volatility for that period.

Above we see the median volume for each 30-minute period in ES for the last 14 trading sessions (top chart) and the median high-low percentage price range for those periods. The correlation between volume and volatility is obvious. Also obvious is that volume and volatility shift substantially within the trading day. Indeed, the median volume of the 11:30 AM CT - 12 Noon CT bar is less than half of the median volume for the market's opening 30-minute bar. The median price range per 30-minute period--the total price movement--is 40% smaller for the midday period compared with the opening one.

This changing volume/volatility dynamic within the trading day has important implications for the intraday trader, including the placement of stops and price targets. I recently corresponded with a rookie trader who placed his stops and price targets a fixed number of points from this entry. While this provided the illusion of managing risk, it was ineffectual. The fixed stop point was hit simply by random price movement during the busy periods of the day, taking the trader out of good trades. Conversely, during slow periods, the trader's price targets were never hit, leading to reversals of his paper profits. The trader sought me out for help in dealing with his emotions, but in fact his problem was his failure to adapt to shifts in volatility.

This same problem occurs among even relatively sophisticated traders who bring expectations of price movement from the last day or two to the present trading day. They anticipate more or less movement than the market is actually giving them, leading to premature exits from good trades on busy days and failure to take profits on trades when markets are slow.

So how can we estimate the volatility of the day that we're currently trading, so that we can adapt accordingly? The idea of relative volume is that you compare the volume for the recent time period with the average volume for that same time period over the past X days. For example, I know that the median volume for the time period of 12:30 - 1 PM CT is 113,785 contracts, with a standard deviation of 53,262 contracts. If I see volume for that period exceed 200,000 contracts during the trading day, I know that this is a significant jump in volume relative to the recent past. It alerts me that institutional traders are relatively active in this market, so I will pay particular attention to whether they are buying or selling (by analyzing NYSE TICK and Market Delta for that period). It is out of such surges in participation that breakout moves and trends are often born.

Conversely, if I know that if the opening half hour of trade averages 259,412 contracts with a standard deviation of 66,817 contracts and see that the current opening period has only traded 150,000 contracts, I know right away that large traders are not dominating this market. This will lead to reduced volatility, and it typically leads to a choppier trade, as market makers push the market to and fro to make their scalping profits. A market dominated by market makers trades very differently from a market dominated by prop traders and fund portfolio managers. By recognizing relative volume, you can also identify who is relatively active in the marketplace--and that will provide you with valuable clues as to how much--and what type of--opportunity is present.

Note: If there is sufficient interest, I can post real time relative volume info via Twitter and as part of the indicator updates on the blog. The RSS subscription to Twitter is free of charge; the weekly indicator updates typically appear on Monday prior to the market open.
.

19 comments:

SSK said...

EXCELLENT. Thanks! Something that will help me for sure. Best, SSK

Magik said...

I would love to see the volume updates via twitter.

Fred said...

I would love to see the volume updates via twitter too.

Nice post !


Thanks !

leo00o83 said...

I'd love to get these twitter updates and why not make this an indicator as well?

markus said...

Good post.
Adapting to the changing volatility is really a challenge. My edge in trading is mostly good execution. So if the market I am trading becomes very volatile perfect execution does not matter so much any more and my p/l becomes worse.

Cheers,
Markus

MGTrader said...

Great post Brett. There was an excellent Relative Volume spike that really helped me on Friday. The ES was into its 4th bracket (30 minute bar closing at 1300 EST) of rotation (balance). I was impatiently waiting for a breakout. I have a relative volume in Market Delta set to 200 trading days. Suddenly, the relative volume for that period shot to 147%. It immediately put me on the alert that a breakout was probable.

I posted a chart of this in my blog (http://mgtrader.blogspot.com/) if you'd like to check it out.

MG

michael said...

brett , how many days of lookback are you using to calculate avg or median volume for a given period.....

also, what do you prefer...the median or the avg.

michael

adan said...

i'm impressed, and yes, i too would like the volume updates!

wow, great stuff, thank you!

Don C said...

Good post.

Beyond looking at relative volume within the same market/instrument, looking at relative volume between related markets (ie markets with a fundamental causal relationship) may be worthwhile as well.

Time structure of volume dynamics between related markets like GC/EU, ES/TY, ES/BD(here you are splicing across not only the usual equities/rates dynamics, but also session timeframe) can reveal some useful insights.

Both short-term and longer term analysis throws up interesting perspectives. But note regime shifts that cause market inter-relationships in volume dynamics to change in the longer timeframe.

Brett Steenbarger, Ph.D. said...

Thanks for the interest and perspectives. Michael, I use median for this particular purpose with a 3 to 4 wk lookback. Median does a better job at minimizing impact of outliers--

Brett

Eric said...

Thanks for the info, it would be great if real time volume is posted.

SSK said...

HELLO MGTrader, Congrats on your blog! Nice work within the beta version. I am spending time reconfiguring some charts too. Is it possible to upload the definition to your chart that has the relative volume on the 30 minute time frame to charthub? For the life of me, I cant figure out how to get that configured correctly on my chart. I like the fact that you can see that on the fly. If you can, Great, if you could help me understand how to program that, that is great too! Thanks a million. My website is www.tamta.net or my email is daytradingfutures@tamta.net
Thanks again, and keep up the good work.
Good trading to you, Best, SSK

Mark said...

Hello Brett,

Good post, I especially like the 'time-relative' volume idea...

This post makes me really curious about something I was already wondering about reading your posts:

Whenever you talk about volume in your posts you always seem to focus on the volume of the futures and not the cash.

Could you please share your thoughts on this subject with us?

Regards,

-Mark-

sunyata said...

Dr. Brett,

Since I do not have access to futures data, do you think that an analysis of the SPY's volume would be sufficient?

Best,
Jake

Noam said...

Hi Dr.

How many "days back" do you use for calculating the relative volume?

Brett Steenbarger, Ph.D. said...

Hi Noam,

At least 20 days for calculations--

Brett

Noam said...

Thank you very much

S L said...

Relative volume update would be greatly appreciated. Thks.

TREND ZONE said...

Hello Brett,
I have a question, it's not regarding the current topic, but it's something you've talk in the past. I'm interested in the cumulative tick (nyse), and I can not find it on any site, nor can I find it in my trading platform. Can you help me with a link. thank you