What Is a “Reflecting Boundary”? The Idea Behind Milton Berg's Market-Bottom Signals

The drunkard and the wall

Milton Berg's paper The Boundaries of Technical Analysis, published in the Journal of Technical Analysis in 2008, opens by granting the academic case in full. Five-day changes in the S&P 500 form a bell curve, and a bell curve is what randomness looks like. Then the paper turns to an example from Stephen Jay Gould's book Full House.

A man leaves a bar dead drunk and stands on the sidewalk. The wall of the bar is on one side of him and the gutter is on the other. Each stagger goes in either direction with equal odds. Let him stagger long enough and he ends in the gutter every time — not because anything pulls him there, but because the wall will not let him go the other way.

“In a system of linear motion structurally constrained by a wall at one end, random movement, with no preferred directionality whatsoever, will inevitably propel the average position away from a starting point at the wall.”

Gould called the wall a reflecting boundary. Milton Berg's claim is that markets have them too.

“We posit that rigorous technical analysis can identify areas of 'reflecting boundaries' in the capital markets. The direction of stock price movements can therefore be predicted in advance despite the perceived random nature of their daily and weekly moves.”

That is the resolution of the paradox. The day-to-day motion can be random and the destination still knowable — if you can tell when prices are standing against a wall.

A wall you cannot see, only bump into

The paper is careful about what a boundary is not. It is not a price level you can calculate in advance, not a calendar date, not a line on a chart. It can last for years or for days, and there can be one boundary or a staircase of them at successively higher or lower prices. Milton Berg makes no attempt to explain why any particular boundary exists.

“In our view, the primary causes of stock price movements are too diverse, complex, and hidden to be analyzable. What we as technicians attempt to do is recognize the symptoms that lead and accompany directional movement of stock market prices.”

Even valuation gets this treatment: a very cheap price, the paper suggests, is one of the factors that helps form a boundary, not the boundary itself. So how do you find a wall you cannot see? By watching for the moment a crowd hits it.

“Technical indicators do not reveal causes of market movement. They simply indicate the proximity of a reflecting boundary.”

The question the paper refuses to answer

Back to the bell curve. From January 1928 through the end of 2007, the paper counts 21,165 five-day periods in the S&P 500. In 150 of them, 0.71% of the total, the index gained 8% or more; in 138, or 0.65%, it lost 8% or more. To a statistician those are the tails of a random distribution and nothing else.

Milton Berg looks at the same tails and asks a different question.

“Why would buyers be willing to pay 8-24% more for a diversified portfolio of stocks than they were willing to pay five days prior? Why would sellers be willing to accept 8-24% less than they were willing to receive five days prior? We do not care to know the answer. We care that it is a good question.”

The only information a price carries is what investors were willing to pay. When thousands of them suddenly pay far more, or accept far less, than they would have a week earlier, that is aberrant behavior — and aberrant behavior is what a crowd hitting a wall looks like. The timing cannot be forecast; that is what random means. But once it has happened, in the paper's words, the market “inevitably propels away from that boundary.”

Price and time first, then volume

A raw five-day surge of 8% is not a signal on its own. The paper's appendix lists all 150 of them, scattered across bear-market rallies as well as bottoms. The signal comes from adding time. Milton Berg keeps only the surges that arrive four, five or six days after the market's lowest close of the previous 90 days. A surge that comes later is ignored, because the market is no longer standing at the wall. A surge one to three days after the low is also thrown out, because a thrust that quick after a low is, in the paper's word, suspect.

What survives is a list of 20 dates since 1929. According to the report, using nothing but price and time, this rule signaled within four to six days of the historic lows of November 13, 1929; June 1, 1932; February 27, 1933; June 26, 1962; May 26, 1970; October 3, 1974 and August 12, 1982 — and within days of each leg of the triple bottom of July 2002, October 2002 and March 2003. The gains that followed the signals of 1962, 1974, 1982 and 2002–2003, again according to the report's table, ran from 53% to 80%. The table is printed in full, including four signals between 1930 and 1942 that arrived within a week of a top in the whipsaw markets of the Depression, when the index held only 90 stocks.

Then the left tail, with a third ingredient. The paper takes every five-day decline of 8% or more, keeps only those within one day of a six-month low, and adds volume: the five-day average of daily volume must be the highest in 250 days, something seen on just 2.04% of days since 1929. Sellers accepting 8% less than a week earlier, at a six-month low, in the heaviest trading of the year — that is capitulation with numbers on it. The report calls the result “a viable capitulation-defining indicator” and lists where it fired: within four days of the lows of November 13, 1929, October 19, 1987 and July 23, 2002, and near the final low of June 26, 1962.

Why this is the foundation, not the model

Nothing above is the MB Edge model. The five-day rules are public, reproducible examples that Milton Berg chose because anyone can check them against the record. The model itself combines many indicators and waits for clusters of them, as described in What Is Turning Point Analysis?

But the paper explains why the model is built the way it is. All of its precision sits at the low, because a reflecting boundary is a place and a crowd hits it in a window of days. That is why the model's buy signals are dated to a single day, and why it is content to do nothing for long stretches when no boundary is in sight. How that becomes one position — 100% S&P 500 or 100% Treasury bills — is covered in Inside the MB Edge Model.

Frequently asked questions