A N OTE ON V OLATILITY P UMPING

Enrico Schumann

1 Introduction 2.0
8

u−1 = 0.2

u−1 = 0.2
1.5 ● ●

● ● ●
● ●
●
●
●

● ●
● ● ●

● ● ●
●

● ● ●
6
Volatility pumping is a trading strategy which, in
● ● ●
● ● ● ● ● ● ●
● ● ●

1.0 ● ● ● ●
● ● ● ● ●
● ● ● ● ● ● ● ● ●
● ●
● ●

4
● ● ● ● ● ●
● ● ● ● ● ● ● ● ● ● ● ● ●
● ● ● ● ● ● ● ● ● ● ●
● ● ● ● ● ● ● ●
● ● ● ● ● ● ● ● ● ●
● ● ● ● ● ● ●
● ● ● ● ●
● ● ● ● ● ● ●
● ● ● ● ●
● ● ● ● ●
●

essence, tries to exploit the fact the even though the

●

0.5 2 ●● ●●●
●● ● ●●● ●
●
●
●
●
●
● ● ●
● ●
●
●
● ● ●●●
● ●
●
●
● ●
●
●
●
● ●
● ● ●
●
●
● ●

●●●
●●●●●●●● ●●●●●●

0.0 0
correct measure to gauge the performance of a finan-

20

40

60

80

100

20

40

60

80

100
cial asset is the geometric mean return, its expecta-
tion at any point in time is its arithmetic mean re- 4 3.0
turn, which is always equal to or higher than the ge- 2.5

u−1 = 0.2

u−1 = 0.2
3
2.0
ometric mean. (For a much better introduction, see
●
●
●

2 ●
● ● ● ●
● ● ● ● ●
●
● ● ● ●
● ● ● ●
●
● ● ●
1.5 ● ● ● ●
● ● ● ●
● ● ● ●
● ● ● ● ●
● ● ●
● ●
●
●
● ● ●
● ● ● ● ●
● ● ● ● ●

1.0
● ●
●
● ● ● ● ● ● ● ● ● ● ● ● ●
● ● ● ● ● ● ● ●
● ● ● ● ● ● ●

Luenberger (1998).)
● ● ●

1 ● ●
● ● ● ● ●
● ● ● ● ● ● ● ● ● ●● ● ●
● ● ● ● ● ●● ●●● ● ●
● ● ● ●●
● ● ● ● ● ● ● ● ● ●

0.5
● ● ● ●
● ● ●● ● ● ● ● ● ● ● ● ●
● ● ●● ●
● ●● ● ● ●●● ● ●●● ● ● ● ● ●●
●●●
● ● ●●●
● ● ● ●●
●●●

0 0.0
To illustrate this point, assume two assets, one

20

40

60

80

100

20

40

60

80

100
which stays constant in value (cash), whereas the
other either rises by u, or goes down by d = 1/u 2.0 1.0 ● ● ●

●
● ● ● ●

(both with probability .5). Assume u = 1.2, that is 0.8

●
● ●
● ● ●
u−1 = 0.2

u−1 = 0.2
●

1.5 ● ●

●
●

●
● ●
●
●
● ●

●
● ● ●

● ● ●
● ● ●

● ● ●
●

●
●

0.6 ● ● ●
● ●

we have a rather volatile asset. Figure 1 shows sev-

● ● ● ● ● ● ●
● ● ● ● ● ● ● ● ● ●

1.0 ● ●
●
●
● ●
● ●
● ●
●
● ●
● ● ●
● ●
● ● ● ●
●
● ●
● ● ●

●
● ●
●
●
● ● ● ● ● ● ●
● ● ● ● ●
● ●
●
●
●
●
● ●
● ●
0.4 ● ● ●
● ●
● ● ●
●
● ●

●
●
● ● ● ● ●

eral sample paths of this asset. (The price in period 0

●

0.5 ● ●
● ● ●
● ● ●
● ●
● ● ●●●● ●
● ● ● ●● ● ●●
● ●
● ●
●
●
● ● ●● ●● ●●●●●
●
● ●
● ● ●
● ● ●
●
●
● ● ●
● ● ●
●
●

0.2
0.0 0.0
equals 1.)
0

20

40

60

80

100

20

40

60

80

100
1.5 8

1.0
6 Figure 2: Sample series for risky asset with cash (50-
4 50 portfolio). The dotted line represents the portfo-
0.5 2
lio, the solid line the asset.
0

20

40

60

80

100

20

40

60

80

100

3.5 2.5 which is greater than zero. To exploit this, one can (in
3.0 2.0
2.5
1.5
theory, of course, costlessly) shift one’s wealth back
2.0
1.5 1.0 and forth between the riskless asset and the risky
1.0
0.5
0.5 one, thus increasing one’s exposure to the risky as-
0.0 0.0
set when is has gone down, and decreasing it when
0

20

40

60

80

100

20

40

60

80

100

it has risen. This resembles much the idea of ’buying

1.2 1.0 low and selling high’. Unfortunately, it is also akin to
1.0 0.8
0.8 the infamous ’martingale strategy’, by which a gam-
0.6
0.6
0.4 0.4 bler doubles the stakes after each loss so to get even
0.2 0.2 with just one lucky trial.
Figure 2 shows how this strategy can work out in
0

20

40

60

80

100

20

40

60

80

100

’practice’ when rebalancing is costless. Here 50% of

the portfolio are invested into the risky asset, the rest
Figure 1: Sample series for risky asset. is kept as cash. There are some important caveats:
First, the returns are assumed to be independent. De-
The average geometric return of the risky asset is
pendence (say, correlation) is also important in the
1 1
d u =1
2 2 cross-section. In general, volatility pumping can be
applied not just to one asset and cash, but to many
However, the expected return in any period is the risky assets simultaneously (in theory, this works
arithmetic mean, given by much better than with cash). Volatility pumping will
1 1 not work nicely, though, when applied to correlated
(d − 1) + (u − 1) = .03̄, assets which move up or down in lockstep. Second,
2 2

1
 2.5 2.5
2.0 2.0
u−1 = 0.01

u−1 = 0.02
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0 0.0

20

40

60

80

100

20

40

60

80

100
2.5 2.5
2.0 2.0
u−1 = 0.05

u−1 = 0.1
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
0

20

40

60

80

100

20

40

60

80

100
2.5 2.5
2.0 2.0
u−1 = 0.15

u−1 = 0.2

1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
0

20

40

60

80

100

20

40

60

80

100
Figure 3: Mean path of 5,000 simulations of rebalanc-
ing strategy (solid line). The dotted line represents
the 10th quantile to give an indication for the risk in-
volved.

the higher the volatility, the better. For low volatil-

ity assets, the results may be less impressive. Third,
all results rely on features which are certainly valid
’in the long run’, but this does not necessarily imply
desirable return properties of this strategy in finite
samples.

2 Implications for trading strate-

gies
Volatility pumping may be also applicable to trad-
ing strategies. In fact, trading strategies on several
assets may be less correlated than these assets them-
selves. Simulations indicate that at least for only one
risky asset and cash, volatility must be very high to
produce reasonable results (see figure 3). Thus, to
produce profits, a considerable amount of risk must
be borne. . . .
Further research has to be done on rebalanced
portfolios of several risky assets (or trading strate-
gies).

References
Luenberger, David G. (1998). Investment Science. Ox-
ford University Press. Oxford.