WEBVTT

1
00:00:00.311 --> 00:00:02.211
How much of each
fund should you hold?

2
00:00:04.463 --> 00:00:06.273
Start with 4
building blocks:

3
00:00:06.273 --> 00:00:08.583
stocks, bonds, gold,
and real estate,

4
00:00:08.583 --> 00:00:11.267
because the solver can
only choose from the

5
00:00:11.267 --> 00:00:12.203
funds you list.

6
00:00:13.003 --> 00:00:15.403
That makes the universe
the first real decision.

7
00:00:16.001 --> 00:00:19.087
Portfolio Optimizer
measures each fund's return,

8
00:00:19.087 --> 00:00:21.981
volatility, and relationship
with the others.

9
00:00:22.901 --> 00:00:24.985
It searches for weights
that best satisfy

10
00:00:24.985 --> 00:00:25.951
a chosen objective.

11
00:00:26.821 --> 00:00:29.717
As an example, optimize 4
building blocks using 10

12
00:00:29.717 --> 00:00:30.701
years of history.

13
00:00:31.487 --> 00:00:34.041
Then choose 10 years
of daily history,

14
00:00:34.041 --> 00:00:36.327
starting at the
beginning of 2016.

15
00:00:37.017 --> 00:00:38.777
Through the end of 2025.

16
00:00:38.777 --> 00:00:41.582
Everything the solver knows
about these funds comes

17
00:00:41.582 --> 00:00:42.517
from this window.

18
00:00:43.253 --> 00:00:45.453
Mode determines the
shape of the answer.

19
00:00:46.373 --> 00:00:48.967
One optimal mix returns
a single set of weights,

20
00:00:48.967 --> 00:00:51.182
while the frontier
returns the whole risk

21
00:00:51.182 --> 00:00:52.533
and return curve instead.

22
00:00:53.313 --> 00:00:54.693
Start with one optimal mix.

23
00:00:55.437 --> 00:00:57.887
The objective defines
what the solver should

24
00:00:57.887 --> 00:00:59.057
maximize or minimize.

25
00:01:00.157 --> 00:01:02.934
The selected objective
seeks the most return per

26
00:01:02.934 --> 00:01:05.422
unit of volatility,
measured here against a

27
00:01:05.422 --> 00:01:06.637
risk-free rate of 0%.

28
00:01:07.657 --> 00:01:10.119
Other objectives
target low volatility,

29
00:01:10.119 --> 00:01:11.697
equal risk, or tail loss.

30
00:01:13.146 --> 00:01:15.742
Open Advanced settings
to inspect constraints

31
00:01:15.742 --> 00:01:17.126
and estimation controls.

32
00:01:17.906 --> 00:01:20.748
A constraint narrows what
the solver may propose,

33
00:01:20.748 --> 00:01:23.126
by banning short
selling or capping risk.

34
00:01:23.746 --> 00:01:26.407
Shrinkage pulls noisy historical estimates toward a

35
00:01:26.407 --> 00:01:28.546
common value, which
steadies the weights.

36
00:01:29.426 --> 00:01:31.576
This first run leaves
all of them inactive.

37
00:01:34.018 --> 00:01:36.629
Each fund also has its
own settings layer,

38
00:01:36.629 --> 00:01:39.863
where a modifier applies
leverage or an expense drag

39
00:01:39.863 --> 00:01:41.231
to that fund's history

40
00:01:41.231 --> 00:01:43.158
before the solver
ever sees it.

41
00:01:43.838 --> 00:01:45.258
This run uses no modifiers.

42
00:01:46.783 --> 00:01:49.733
Now find the mix that best
satisfies this objective

43
00:01:49.733 --> 00:01:51.063
over the chosen window.

44
00:01:51.983 --> 00:01:53.807
Run the solver to
measure every fund

45
00:01:53.807 --> 00:01:55.733
and search the
available combinations.

46
00:01:57.022 --> 00:02:00.313
The answer puts
everything in two funds:

47
00:02:00.313 --> 00:02:02.782
about 41% stocks
and 59% gold.

48
00:02:03.822 --> 00:02:05.882
Bonds and real
estate receive 0.

49
00:02:06.882 --> 00:02:08.832
Optimizers call that
a corner solution.

50
00:02:09.322 --> 00:02:11.772
Only funds that improve
this window's objective

51
00:02:11.772 --> 00:02:12.502
enter the mix.

52
00:02:13.758 --> 00:02:16.933
The mix shows an
expected 15% a year

53
00:02:16.933 --> 00:02:19.138
against 11.7% volatility.

54
00:02:20.158 --> 00:02:22.817
The return-to-volatility
ratio is 1.28,

55
00:02:22.817 --> 00:02:24.998
with no risk-free
rate deducted.

56
00:02:26.018 --> 00:02:28.889
Both figures use the same
historical window used to

57
00:02:28.889 --> 00:02:29.958
select the weights.

58
00:02:30.792 --> 00:02:32.372
Save any run you
want to keep.

59
00:02:33.052 --> 00:02:34.233
It stores the universe,

60
00:02:34.233 --> 00:02:35.567
the window, the objective,

61
00:02:35.567 --> 00:02:37.825
and the weights, so it
can be reopened later

62
00:02:37.825 --> 00:02:38.852
or compared against.

63
00:02:39.677 --> 00:02:42.472
Now constrain volatility
to no more than 10% a

64
00:02:42.472 --> 00:02:44.295
year,
while keeping the funds,

65
00:02:44.295 --> 00:02:45.997
window, and objective fixed.

66
00:02:46.837 --> 00:02:48.257
Add the ceiling, then rerun.

67
00:02:49.268 --> 00:02:51.808
Same funds, same window,
one constraint.

68
00:02:52.748 --> 00:02:54.660
Bonds enter at about 18%,

69
00:02:54.660 --> 00:02:58.561
and the expected annual
return the solver can reach

70
00:02:58.561 --> 00:02:59.708
falls to 12.7%.

71
00:03:00.068 --> 00:03:01.924
The saved run
stays as it was,

72
00:03:01.924 --> 00:03:03.224
so you can update it,

73
00:03:03.224 --> 00:03:04.028
or keep both.

74
00:03:05.689 --> 00:03:08.880
Hand any answer to Portfolio
Backtest to inspect

75
00:03:08.880 --> 00:03:10.209
its historical path.

76
00:03:10.909 --> 00:03:13.518
The allocation can also
become a state inside the

77
00:03:13.518 --> 00:03:14.689
Strategy Builder tool.

78
00:03:15.387 --> 00:03:17.787
The solver used
nominal total returns

79
00:03:17.787 --> 00:03:20.967
before tax and trading
costs to fit these weights

80
00:03:20.967 --> 00:03:23.367
over 10 years when
gold ran strongly.

81
00:03:24.537 --> 00:03:26.773
These weights are a
property of that window,

82
00:03:26.773 --> 00:03:27.637
not of the funds.

83
00:03:28.437 --> 00:03:30.268
Change the dates
and the optimal mix

84
00:03:30.268 --> 00:03:31.437
can move substantially.

85
00:03:32.057 --> 00:03:35.083
Choosing one optimal mix
answers a single question

86
00:03:35.083 --> 00:03:36.657
about the selected window.

87
00:03:37.477 --> 00:03:40.138
Efficient Frontier mode
answers a broader one:

88
00:03:40.138 --> 00:03:41.757
every efficient mix at once.

89
00:03:42.575 --> 00:03:45.372
Run it, and the solver
traces the best return

90
00:03:45.372 --> 00:03:47.423
available at every
level of risk,

91
00:03:47.423 --> 00:03:49.474
from the lowest
volatility mix to

92
00:03:49.474 --> 00:03:50.655
the highest return.

93
00:03:51.766 --> 00:03:54.324
The star marks the mix
with the highest return

94
00:03:54.324 --> 00:03:56.326
per unit of volatility
on the curve.

95
00:03:57.146 --> 00:03:59.409
Every point on it is
a different trade-off,

96
00:03:59.409 --> 00:04:01.146
and no single fund
sits above it.

97
00:04:01.746 --> 00:04:02.876
The gap between a fund

98
00:04:02.876 --> 00:04:05.086
and the curve is what
diversification buys.

99
00:04:05.970 --> 00:04:08.004
The frontier explains
why the constraint

100
00:04:08.004 --> 00:04:08.920
moved the weights.

101
00:04:09.710 --> 00:04:11.437
A ceiling on
volatility is a wall

102
00:04:11.437 --> 00:04:12.170
on this plane.

103
00:04:12.870 --> 00:04:14.205
The best mix under it sits

104
00:04:14.205 --> 00:04:15.798
where the wall
meets the curve,

105
00:04:15.798 --> 00:04:17.390
further left,
with less return.

106
00:04:18.287 --> 00:04:19.851
Every run keeps its window

107
00:04:19.851 --> 00:04:21.655
and its assumptions
on record,

108
00:04:21.655 --> 00:04:24.301
so a surprising allocation
can be reproduced

109
00:04:24.301 --> 00:04:26.647
and challenged instead
of argued about.

110
00:04:27.308 --> 00:04:28.947
List the funds,
choose a window,

111
00:04:28.947 --> 00:04:30.228
and define the objective.

112
00:04:31.208 --> 00:04:32.887
Read the proposed
weights against

113
00:04:32.887 --> 00:04:34.058
their fitted trade-off.

114
00:04:34.308 --> 00:04:36.706
Save the run, add one
constraint or change the

115
00:04:36.706 --> 00:04:37.748
window, and compare.

116
00:04:39.320 --> 00:04:41.560
Choose Take a tour for
a guided walkthrough.

117
00:04:42.120 --> 00:04:43.670
You can dismiss
it at any time.

118
00:04:46.061 --> 00:04:47.892
Open Docs for every setting,

119
00:04:47.892 --> 00:04:50.901
objective, constraint,
and methodology detail.

120
00:04:52.567 --> 00:04:54.474
Let the solver
propose a mix for the

121
00:04:54.474 --> 00:04:55.587
objective you define.

122
00:04:56.527 --> 00:04:58.209
Then inspect its
fitted evidence

123
00:04:58.209 --> 00:05:00.207
and test the proposal
against history.