Go to ... Page 1 | Page 2 |

Unusual Draw­down Risk

by John P. Huss­man, Ph.D., Huss­man Funds

In order to esti­mate likely returns and risks in the finan­cial mar­kets, our gen­eral approach is to iden­tify a set of his­tor­i­cal instances that match cur­rent con­di­tions on a broad range of impor­tant dimen­sions (in prac­tice, using an "ensem­ble approach" that ran­dom­izes over scores of sub­sets of his­tor­i­cal data). We then look at var­i­ous fea­tures of that clus­ter, includ­ing the aver­age return that fol­lowed over var­i­ous hori­zons, the deep­est loss over var­i­ous hori­zons, and the over­all spread of those out­comes. In gen­eral, the clus­ters include a mix of both pos­i­tive and neg­a­tive out­comes, result­ing in mod­er­ate esti­mates of expected return and mod­er­ate esti­mates of risk. In some cases, the aver­age return across the clus­ter of instances is very pos­i­tive, and the indi­vid­ual instances show few neg­a­tive out­comes at all. That sort of con­di­tion jus­ti­fies a very aggres­sive invest­ment posi­tion. In con­trast, since the late-1990's, the aver­age returns of the clus­ters have been quite poor, with a pre­pon­der­ance of neg­a­tive out­comes in his­tor­i­cal instances hav­ing sim­i­lar characteristics.

There have been a few excep­tions, includ­ing the bulk of 2003 (and on the basis of the ensem­ble meth­ods we presently use, the period between early-2009 and early 2010 — see Notes on Risk Man­age­ment for details on our unpleas­ant "miss" dur­ing that period). But gen­er­ally speak­ing, mar­ket con­di­tions since the late-1990's have sup­ported a defen­sive invest­ment stance much more often than is typ­i­cal on a his­tor­i­cal basis. Of course, the near-zero total return of the S&P 500 since the late-1990's, cou­pled with two sep­a­rate mar­ket losses of more than 50%, is a reflec­tion that on aver­age, con­cerns about poor return and high risk dur­ing this period have been well-placed.

In review­ing mar­ket con­di­tions this week, what strikes me most is the pat­tern that emerges when we look across var­i­ous hori­zons, from 2 weeks out to 18 months. When we exam­ine the aver­age 2-week out­come that has his­tor­i­cally fol­lowed peri­ods that clus­ter with present con­di­tions, the aver­age out­comes are neg­a­tive, but not strik­ingly so. Specif­i­cally, the expected return is in the low­est 26% of all his­tor­i­cal obser­va­tions, but that aver­age return is only about –1%, a fig­ure that is over­whelmed by typ­i­cal short-term noise. That's another way of say­ing that guess­ing the market's out­come over the next cou­ple of weeks is like guess­ing the throw of a very slightly biased pair of dice.

But the pro­file starts to change sig­nif­i­cantly as we move out the invest­ment hori­zon. Look­ing out 5 weeks, for exam­ple, the prospec­tive return falls into the low­est 8% of his­tor­i­cal obser­va­tions. Now, this could cer­tainly change on the basis of shifts in var­i­ous mar­ket con­di­tions, but here and now, the 5-week hori­zon is more defen­sive than we've seen in the other 92% of his­tor­i­cal data.

Most strik­ing, though, is what we observe on the basis of prospec­tive draw­down (the deep­est loss the mar­ket expe­ri­ences within a given hori­zon) look­ing out over the com­ing 18 months. On that front, the present draw­down esti­mate is in the worst 1.5% of all his­tor­i­cal observations.

Keep in mind the dis­tinc­tion between the draw­down and the return over a given period. The draw­down over an 18-month period is the deep­est loss expe­ri­enced by the mar­ket from the cur­rent point to the low­est point within that hori­zon, even if the deep­est loss occurs fairly early in that win­dow. In con­trast, the return over a given period is mea­sured from the start­ing point to the end­ing point. Impor­tantly, once we observe con­di­tions that asso­ciate with a sig­nif­i­cant risk of draw­down, we can almost always find some point later on that pro­vides a bet­ter entry oppor­tu­nity to accept mar­ket risk.

Ele­vated Mar­kets and Draw­down Risks

The chart below iden­ti­fies peri­ods in recent years where we reported mar­ket con­di­tions as being at least "over­val­ued" and "over­bought" in these weekly com­men­taries. Those two con­di­tions alone aren't enough, by them­selves, to put the mar­ket in a "hard-negative" sit­u­a­tion, but even those two tend to be enough to invite draw­down risk. The over­val­ued, over­bought peri­ods are shaded in blue on the chart below. The red lines indi­cate the deep­est draw­down expe­ri­enced by the mar­ket over the fol­low­ing 18 months (right scale), while the blue line charts the S&P 500 (left scale). Notably, even with weakly neg­a­tive con­di­tions — over­val­ued and over­bought — the mar­ket has typ­i­cally moved lower at some point in the next 18 months, wip­ing out all inter­ven­ing gains. That sur­ren­der of inter­ven­ing gains usu­ally begins with a very hard and unex­pected ini­tial loss that takes out the bulk of upside progress within a period of a few days or weeks. This is a gen­eral pat­tern that we also see through­out mar­ket history.

Of course, our present con­cerns are based on a smaller and more neg­a­tive sub­set of con­di­tions that we've seen even less fre­quently — presently fea­tur­ing not just "over­val­ued" and "over­bought" con­di­tions, but adding over­bull­ish sen­ti­ment, mod­est but clear upward pres­sure on short-term and some longer-term yields, and an "exhaus­tion syn­drome" (a com­bi­na­tion of "whip­saw" con­di­tions cou­pled with falling earn­ings yields — see Goat Rodeo ), which have his­tor­i­cally had a par­tic­u­larly hos­tile after­math, includ­ing a small set of his­tor­i­cal plunges that include 1987, 2000 and 2008.

Go to ... Page 1 | Page 2 |

Read more from the author/contributor here.

Comments

Comments are closed.