Initial Balance Breakouts: By the Numbers

One of the questions we get asked time and again is how to play breakouts of the initial balance (IB) range. Of course, baked into those questions is the assumption that initial balance breakouts should be played. But is that true? If so, when and how? In an important way, this study is the logical extension to the study I published (and recently updated!) which looked at the IB range itself. Well, now I’m back with another update to the breakout study. Here we’ll quantify and try to interpret the when, how and how much of what happens outside the IB range in a pair of perennially popular instruments.

So keep reading, amigos, and we’ll examine the hard cold facts and let them guide our approach to this situation. Then maybe we’ll see whether, in the words of Yogi Berra, that particular fork in the road is one worth taking.

Here we go…

Methodology

As always, I like to keepin’ simple and this’s one that you Pro Members can play along with pretty easily. I used the Acme IB Breakout market study from the Market Study Pack to export 2 years’ of data in each of the 2 instruments examined. Then I fired up Excel® to help focus and visualize the data. The instruments examined here are the the ES, and the CL, and only the RTH session was included in each. Some specifics you’ll want to know about:

exported data is 1 minute bars, RTH sessions only
the IB duration – set to the traditional 60 minutes for each instrument
we’re looking for when the IB range was first broken to the high side and/or low side
we’re looking for when price reached reached it’s maximum extension from the IB high or low after the breakout
we’re looking for how far price was from the IB high or low at its maximum excursion
the current graphs and discussion is on top, the prior sample graphs are down below

Even though the study design is pretty straightforward, I need to call out a couple caveats. First, keep in mind that whether price reentered the IB after the breakout and how far it tended to travel was not considered. This could be important if you choose to initiate a position on an IB breakout as it could help you decide where to place your stops (or otherwise manage your risk). Next, the range of the IB or the day was not taken into account either. As I like to say, as an independent trader, you are your own research department. And these are facets you should examine and factor into your own trading plan. I’ll leave it to you, good sirs and madams, to dot and cross those particular “i”s and “t”s.

[hr_invisible]

[box style=”rounded”]NOTE: the graphs below are very wide because of the large ranges, so make your browser as wide as possible before clicking on them.[/box]

[hr_invisible]

And Now… the Data

[iconbox title=”Symbol: ES” icon=”x-office-spreadsheet.png”]

Sessions in sample: 515
Sample period: November 5, 2012 – November 3, 2014

[/iconbox]

In graphs 1a and 1b we have the distribution of the number of minutes after the open that the IB was first broken to the upside and downside. The x-axis shows the possible range of minutes, the y-axis shows the frequency of the breakage for each possible minute. A 60 minute IB for this study, so the first time the IB could be broken is in the 61st minute after the open and the last possible time is 405 minutes after the open, which is the last minute of the RTH session for the ES. Perhaps not so shockingly, the IB was broken most often soon after being set.

—

Like below, graphs 2a and 2b show the distribution of the number of minutes after the breakout that price took to reach its maximum excursion beyond the IB high and low. On the x-axis we have minutes, on the y-axis shows the frequency of each possible duration.

But here’s where the plot starts to thicken a bit. As before there’s a pretty clear pattern here. You’re most likely to get to max excursion on either a high or low breakout within the first 30 minutes afterward. If not, you may have to hang on and incur substantial time risk to get rewarded in those secondary clusters in the last hour of trading. But if you compare the graphs from the previous sample to these you’ll see there’s a bit more uniformity to the 0-30 minute range of the distributions, also a secondary cluster around the 30 minute mark. And also, on the high side, there’s more meat in the area out to 60 minutes. So whaddya make of that?

—

In the caboose position are graphs 3a and 3b. They show the distribution of the max excursion after the breakout, with the x-axis as the number of points, the y-axis the frequency of that max excursion. Said simply, these graphs show the distribution of maximum potential rewards after the breakout. This time as before notice how the maximums are clustered down toward the left side of the graph?

Then and Now – Discussion

In both samples graphs 3a and 3b tell the most important tale. They define the possible range of rewards for ANY risk taken on this particular play. If we use the 80/20 rule (aka the Pareto Principle), then we see the most frequent (and thus most probable) rewards are relatively small. So roughly 80% of the time, as the data shows us, the maximum likely reward is going to be 6 points or less on a breakout to either side. Interesting… there was a bit less on the upside available previously. Now both sides are the essentially the same, potential rewards-wise.

Next, graphs 2a and 2b show us when our most likely maximum reward most likely is achieved. Again, it’s all about understanding the probabilities, and here there is a tight cluster of maximum excursions right after the breakout. The latest sample shows you’re most likely going to see the bulk of your reward for risk taken within about 45 minutes of the breakout. Compare that to 30 minutes previously (below). Keep in mind the axoim that time = risk. If you’re not seeing what you expect in terms of reward within about 30-45 minutes of the breakout, perhaps consider whether a muy rapido and (hopefully) profitable exit from your position is the best course of action. One of the keys to long term survival for any species is to be an opportunist – remember to take what you can, when you can, while you can. Or as my wise and grizzled uncle used to say, “kid, don’t be a dick for a tick.”

Both samples show that IB breakouts most frequently occur fairly near the time when the IB range is actually set. But the ES has gotten a bit slower to break out as shown by the second sample. Now, as opposed to then (below), if breakout does not occur in about 120 minutes of the session open, it becomes less and less likely to happen.

For a trader it’s all about risk and reward, right? So the message for the taking here is that if you’re thinking you can make your intraday bread with this play, maybe it’s possible and maybe not. Like so many things in life and trading – it depends. But what was true before remains so. You can likely improve your long term success in this situation by giving this kind of trade a pretty short leash. The data says if it doesn’t perform in the time frames suggested, then it’s just not likely to. If it does perform later and you miss it? Meh. So what? A big part of trading is learning to be OK with being wrong, learning to be OK with missing out when you didn’t see the opportunity coming. There is no way to predict the future (that I am aware of ;-}). Personally, I’d consider this more of a “gravy trade” and not one to look to for to make my daily bread. Besides, when you put bread together with gravy and a little chipped beef you have S.O.S.! Not my favorite meal.

[hr_invisible]

[iconbox title=”Symbol: CL” icon=”x-office-spreadsheet.png”]

Sessions in sample: 513
Date Range: November 5, 2012 – November 3, 2014

[/iconbox]

Once again in graphs 1a and 1b we’re visualizing the number of minutes after the open it took for price to break out of the IB. And again as with the ES, we see the so-called long pole (most frequent time) of breakout in the minute following the establishment of the IB. If you think about it, this makes sense. The one-hour IB was established because many floor traders have to execute their market-on-open orders by the end of the first hour of trade, and of course there is often going to be a rush of trades as the hour comes to a close. Just as before, the x-axis is the minutes and the y-axis is the frequency. Also as before, the distribution is skewed heavily to the left, with the vast majority of breakouts to the high or low side occuring in the first 90 minutes of trade, or about 30 minutes after the IB range is established. What is different this time compared to last, is the very pronounced spike and cluster right around the 90 minute mark on both the high and low sides. There was a similar topographic feature on the previous run, but it was not as pronounced.

—

Graphs 2a and 2b also show the distribution of time it took for price to reach its maximum excursion, after the breakout occurred. The x-axis again shows the minutes it took to reach the breakout high or low, and the y-axis shows the frequency for each zenith or, er, acme. Sorry, couldn’t resist that one. So if you’ve made your play and you’d like to get paid, the numbers say it’s most likely to happen in the first 45-60 minutes after the breakout. In the prior run, the weight of the graph was concentrated just a bit to the left, say 30-45 minutes. Could it be that CL is becoming a more patient instrument? Maybe. But you might not want not be. But if you have the fortitude to tough it out in your position to near the end of the session, you may bet a crack at a favorable exit there too. But the numbers imply that’s more like gambling than trading. Time = risk in any position.

—

Graphs 3a and 3b show again the maximum theoretical rewards for this trade. As in the ES sister graphs, the x-axis shows the amount of each possible excursion in the data set, and the y-axis shows the frequency of each possible excursion. Whereas in the prior samples the “Pareto majority” of the reward distribution was about 60 ticks or less on either the high or low side. Here we see they can now extend out to around 80 ticks, which is about a 33% increase in the potential (read: reasonably sane) reward target area. That’s pretty significant when you consider that each CL point (100 ticks) is worth US$1000 per contract. This somewhat subtle change in character is also reflected in the change observed in the updated tick-by-tick volume analysis.

Then and Now – Discussion

Really, what is notable here is much the same as with the ES. Though I will say that this trade may be a bit more rewarding over the long haul in the CL instead of in the ES. That is if you’re not slinging large lots around. CL slippage is generally much worse than with the ES. It just has far less liquidity and doesn’t quiver back and forth over the same prices as it moves along in the way ES does. On the other hand, in the CL, it may be more psychologically difficult to execute consistently. In the CL, the reward range is much greater, moves faster and demands quicker decisions, and the time-to-max-reward distribution is somewhat less clear cut. Its volatility can be downright brutal if you’re caught wrong-footed at the wrong time. So this play may be more suitable for a more experienced trader with well-developed intuitions and discipline. Meaning you have to know when to hold ‘em and when to fold ‘em and be OK folding ‘em when it becomes clear you are holding the weak hand. Said differently, I can see this being the kind of trade that is all to easy to hold too long too often.

Only you can decide whether this is one you should consider adding to your playbook. You’re ultimately responsible for assessing your own development and improving your skills and trading capacities. But this one may have potential if executed with discipline. But then I think that’s true for trading in general. It’s a pretty fast-moving instrument that requires a very high level of attention and focus. I often say the ES is like a semi truck and the CL is like a sports car. Different vehicles for different purposes. Each places different demands on its drivers.

—

Ready for Some Next-Level Stuff?

I’ll through out some take-home thoughts and suggestions for further refinement and stratification.

This is a pretty general study to explore the topmost layers of this subject on a couple of the most commonly-traded instruments. If your favorite is not one of them, you can easily replicate this with a little number-crunching elbow grease and seat time in front of Excel.
We didn’t take into account specific market contexts, such as above- or below-average volume or range days, or general market contexts such as whether the market is at or near new long-time highs or lows.
What would the changes look like if you split this data into just the last year only? How much of the changes are due to the most recent activity? How would you adapt your trading with this knowledge?
Lastly, we didn’t look at how the situation changes when we shorten or lengthen the IB duration. Naturally, as the IB is shortened the range will include less of the day’s range and so breakouts will be more forceful and extended. So, a natural follow-up question is how does the reward potential change when using a 30 minute IB? I guarantee it does, and in a meaningful way. But to find out exactly how meaningful, you’ll need to get your own curiosity on.

Until next time… trade em’ well.

[hr]