WEBVTT

1
00:00:00.020 --> 00:00:00.180
Hello,

2
00:00:00.280 --> 00:00:02.143
welcome back to Papers with Backtest podcast.

3
00:00:02.482 --> 00:00:05.465
Today we dive into another Algo trading research paper.

4
00:00:05.688 --> 00:00:06.125
Hi there.

5
00:00:06.586 --> 00:00:06.727
Yeah,

6
00:00:06.746 --> 00:00:09.250
today we're tackling a really interesting one from 2016.

7
00:00:09.406 --> 00:00:09.812
That's right.

8
00:00:09.828 --> 00:00:10.227
It's called

9
00:00:10.648 --> 00:00:12.492
Timing Smart Beta Strategies.

10
00:00:12.914 --> 00:00:13.391
Of course,

11
00:00:13.875 --> 00:00:14.297
buy low,

12
00:00:14.656 --> 00:00:15.273
sell high.

13
00:00:16.094 --> 00:00:17.141
It's by Rob Arnott,

14
00:00:17.422 --> 00:00:17.938
Noah Beck,

15
00:00:18.250 --> 00:00:20.344
and Vitaly Kelesnik from Research Affiliates.

16
00:00:20.641 --> 00:00:23.938
And it really digs into something I think a lot of us wonder about.

17
00:00:24.719 --> 00:00:26.719
Can you actually improve your results by,

18
00:00:27.328 --> 00:00:27.562
you know,

19
00:00:27.688 --> 00:00:28.781
actively timing smart...

20
00:00:28.892 --> 00:00:30.094
beta or factor tilts.

21
00:00:30.213 --> 00:00:30.654
Exactly.

22
00:00:30.794 --> 00:00:34.355
Is it really feasible to like buy these things low and sell them high?

23
00:00:34.418 --> 00:00:36.340
That's the core question they're exploring.

24
00:00:36.341 --> 00:00:37.598
And they look at quite a few things,

25
00:00:37.645 --> 00:00:37.918
don't they?

26
00:00:37.941 --> 00:00:39.824
Like eight different smart beta strategies.

27
00:00:40.004 --> 00:00:40.145
Yeah.

28
00:00:40.207 --> 00:00:41.246
Things like fundamental index,

29
00:00:41.684 --> 00:00:42.168
equal weight,

30
00:00:42.668 --> 00:00:43.207
low vol,

31
00:00:44.668 --> 00:00:46.004
quality dividend weight.

32
00:00:46.707 --> 00:00:47.363
A pretty good mix.

33
00:00:47.629 --> 00:00:48.645
And also eight factors,

34
00:00:48.910 --> 00:00:49.316
value,

35
00:00:49.379 --> 00:00:49.785
momentum,

36
00:00:49.895 --> 00:00:50.316
size.

37
00:00:50.598 --> 00:00:50.832
Right.

38
00:00:50.988 --> 00:00:53.098
And they look at factors as long,

39
00:00:53.191 --> 00:00:53.395
short.

40
00:00:53.496 --> 00:00:54.117
portfolios,

41
00:00:54.177 --> 00:00:58.101
whereas the smart beta stuff is long only compared to a cap weighted benchmark.

42
00:00:58.159 --> 00:00:59.241
That's an important distinction.

43
00:00:59.440 --> 00:00:59.644
Okay.

44
00:01:00.323 --> 00:01:01.163
And the main argument,

45
00:01:01.183 --> 00:01:01.487
I guess,

46
00:01:01.526 --> 00:01:08.675
is that you can potentially do better by leaning into factors or strategies that look cheap compared to their own history.

47
00:01:08.933 --> 00:01:10.425
That's pretty much the central idea.

48
00:01:10.597 --> 00:01:10.737
Yeah.

49
00:01:11.487 --> 00:01:12.534
Emphasize the cheaper ones,

50
00:01:12.675 --> 00:01:13.972
dial back the expensive ones.

51
00:01:14.112 --> 00:01:14.300
Okay.

52
00:01:14.612 --> 00:01:16.675
So let's get into the training rules and results.

53
00:01:17.565 --> 00:01:20.378
The paper starts by warning about just chasing performance,

54
00:01:20.425 --> 00:01:20.644
right?

55
00:01:21.159 --> 00:01:21.644
It mentions

56
00:01:22.080 --> 00:01:25.763
Revaluation alpha versus structural alpha.

57
00:01:25.764 --> 00:01:26.062
Uh huh.

58
00:01:26.605 --> 00:01:27.687
That's a key concept here.

59
00:01:27.906 --> 00:01:33.734
Revaluation alpha is basically gains you get just because the market decides to pay a higher price,

60
00:01:33.735 --> 00:01:35.617
a higher multiple for the same thing.

61
00:01:35.656 --> 00:01:36.672
It's often not sticky.

62
00:01:36.851 --> 00:01:37.093
You know,

63
00:01:37.094 --> 00:01:37.914
it can reverse.

64
00:01:38.101 --> 00:01:39.398
So it's like the price goes up,

65
00:01:39.461 --> 00:01:42.086
but the underlying substance hasn't necessarily changed that much.

66
00:01:42.226 --> 00:01:42.726
Exactly.

67
00:01:42.961 --> 00:01:48.492
Whereas structural alpha is the outperformance that's left over after you account for those valuation changes.

68
00:01:49.039 --> 00:01:51.086
It's hopefully more about the inherent.

69
00:01:51.268 --> 00:01:54.191
quality or characteristic of the strategy or factor itself.

70
00:01:54.290 --> 00:01:54.832
That makes sense.

71
00:01:54.873 --> 00:01:58.333
So just because something did well recently doesn't mean it'll keep going.

72
00:01:58.376 --> 00:01:59.712
It might just have gotten expensive.

73
00:02:00.056 --> 00:02:00.580
Precisely.

74
00:02:01.517 --> 00:02:04.861
The people warns this is a big trap in smart beta and factors,

75
00:02:05.080 --> 00:02:06.759
just like anywhere else in investing.

76
00:02:07.244 --> 00:02:14.087
They give examples like value looking pricey back in 77 or size near its peak popularity in 81.

77
00:02:14.228 --> 00:02:16.806
And their point is a lot of people are timing these things already,

78
00:02:17.228 --> 00:02:17.994
but maybe badly.

79
00:02:18.644 --> 00:02:19.824
Buying after a good run,

80
00:02:20.004 --> 00:02:21.284
selling after a bad one.

81
00:02:21.364 --> 00:02:21.504
Yeah,

82
00:02:21.585 --> 00:02:23.524
effectively buying high and selling low.

83
00:02:23.985 --> 00:02:25.165
The classic mistake.

84
00:02:25.227 --> 00:02:27.047
So if chasing winners is dangerous,

85
00:02:27.305 --> 00:02:28.766
what's the alternative they propose?

86
00:02:29.047 --> 00:02:29.243
Well,

87
00:02:29.305 --> 00:02:31.110
they focus on relative valuation.

88
00:02:31.547 --> 00:02:38.227
The idea is that how expensive or cheap a strategy is compared to its own history can give you clues about its future performance.

89
00:02:38.352 --> 00:02:38.508
Okay,

90
00:02:38.509 --> 00:02:40.243
so not just cheap in absolute terms,

91
00:02:40.305 --> 00:02:42.727
but cheap relative to its own normal range.

92
00:02:42.837 --> 00:02:43.446
Exactly.

93
00:02:43.649 --> 00:02:47.337
They find that this relative valuation seems to be a useful signal for timing.

94
00:02:47.880 --> 00:02:52.623
Strategies or factors that are historically cheap tend to do better going forward and vice versa.

95
00:02:53.006 --> 00:02:53.385
Interesting.

96
00:02:53.404 --> 00:02:55.447
So it's a contrarian signal fundamentally.

97
00:02:55.748 --> 00:02:56.045
It is.

98
00:02:56.451 --> 00:02:58.310
But there's a caveat.

99
00:02:58.513 --> 00:03:00.592
They do warn that if you get too aggressive,

100
00:03:00.631 --> 00:03:03.076
like betting everything on just the single cheapest thing,

101
00:03:03.678 --> 00:03:07.521
you might hurt your diversification and your risk adjusted returns,

102
00:03:07.553 --> 00:03:08.521
your Sharpie ratio.

103
00:03:09.021 --> 00:03:10.615
So moderation is probably key then.

104
00:03:10.740 --> 00:03:13.928
Maybe tilting rather than making huge all or nothing bets.

105
00:03:14.068 --> 00:03:15.521
That seems to be the implication.

106
00:03:15.584 --> 00:03:15.740
Yeah.

107
00:03:16.146 --> 00:03:16.834
Moderate tilts.

108
00:03:17.004 --> 00:03:18.126
based on valuation.

109
00:03:18.385 --> 00:03:18.645
Okay,

110
00:03:18.706 --> 00:03:20.048
let's talk about how they tested this.

111
00:03:20.327 --> 00:03:25.592
They compared trend chasing versus a contrarian approach using some hypothetical portfolios.

112
00:03:25.671 --> 00:03:25.889
Right.

113
00:03:26.053 --> 00:03:26.374
First,

114
00:03:26.475 --> 00:03:27.631
they set up a baseline,

115
00:03:27.756 --> 00:03:28.499
just simple,

116
00:03:28.733 --> 00:03:32.358
equally weighted portfolios of the smart beta strategies and the factors.

117
00:03:32.741 --> 00:03:34.022
Diversification helps here,

118
00:03:34.381 --> 00:03:35.006
lower risk,

119
00:03:35.288 --> 00:03:36.616
often better Sharper ratios.

120
00:03:36.850 --> 00:03:38.178
Simple but effective sometimes.

121
00:03:38.272 --> 00:03:40.147
Then they simulated a trend chaser,

122
00:03:40.413 --> 00:03:41.553
someone who invests in the,

123
00:03:41.881 --> 00:03:42.085
say,

124
00:03:42.344 --> 00:03:45.185
three best performers based on the last year or three years,

125
00:03:45.224 --> 00:03:45.624
five years,

126
00:03:45.665 --> 00:03:46.605
10 years of returns.

127
00:03:46.765 --> 00:03:48.406
Following the momentum basically?

128
00:03:48.644 --> 00:03:50.027
And how did that strategy do?

129
00:03:50.609 --> 00:03:51.070
Not well.

130
00:03:51.507 --> 00:03:53.249
Across all those look back periods,

131
00:03:53.585 --> 00:03:57.546
the trend chasing strategy actually underperformed the simple equal weight approach,

132
00:03:57.890 --> 00:04:00.460
both for smart beta and for factors.

133
00:04:00.882 --> 00:04:01.132
Really?

134
00:04:01.429 --> 00:04:02.023
Underperformed,

135
00:04:02.148 --> 00:04:03.429
even following the winners?

136
00:04:03.476 --> 00:04:03.663
Yep.

137
00:04:04.023 --> 00:04:04.585
And worse,

138
00:04:04.663 --> 00:04:06.648
because you end up concentrating your bets,

139
00:04:06.882 --> 00:04:08.132
the risk actually went up,

140
00:04:08.304 --> 00:04:09.445
lower Sharpe ratios.

141
00:04:10.264 --> 00:04:13.808
So that intuitive ride the winner's idea just didn't seem to work here.

142
00:04:14.169 --> 00:04:14.349
Wow.

143
00:04:14.446 --> 00:04:14.649
Okay.

144
00:04:14.669 --> 00:04:15.610
So what about the opposite?

145
00:04:16.071 --> 00:04:17.270
Investing in the losers.

146
00:04:17.950 --> 00:04:19.235
The contrarian approach.

147
00:04:19.431 --> 00:04:19.571
Ah,

148
00:04:19.634 --> 00:04:20.876
now that's where it gets interesting.

149
00:04:21.477 --> 00:04:26.235
They simulated investing in the three worst performers over those same past periods.

150
00:04:26.376 --> 00:04:26.642
Yeah.

151
00:04:26.845 --> 00:04:30.063
And that contrarian strategy generally beat the trend chaser,

152
00:04:30.329 --> 00:04:31.251
often by quite a bit.

153
00:04:32.432 --> 00:04:37.537
It frequently even beat the simple equal weight portfolio in terms of raw return.

154
00:04:37.877 --> 00:04:41.420
So buying the laggards actually paid off better than buying the leaders.

155
00:04:41.779 --> 00:04:42.584
In their tests,

156
00:04:42.685 --> 00:04:42.920
yes.

157
00:04:43.404 --> 00:04:43.521
Now,

158
00:04:43.584 --> 00:04:44.646
sometimes for factors,

159
00:04:44.647 --> 00:04:48.787
the Sharpe-Ray ratio wasn't necessarily better than equal weight because you are less diversified,

160
00:04:49.131 --> 00:04:51.209
but the overall return boost was often there.

161
00:04:51.537 --> 00:04:54.099
And did it matter which time frame they used for worse performers,

162
00:04:54.131 --> 00:04:55.678
like one year versus five years?

163
00:04:55.834 --> 00:04:56.443
Interestingly,

164
00:04:56.553 --> 00:04:56.678
no.

165
00:04:57.184 --> 00:05:01.128
The contrarian approach tended to outperform regardless of whether they looked at the past one,

166
00:05:01.167 --> 00:05:01.386
three,

167
00:05:01.449 --> 00:05:01.687
five,

168
00:05:01.688 --> 00:05:02.269
or ten years.

169
00:05:02.851 --> 00:05:03.210
Actually,

170
00:05:03.351 --> 00:05:06.175
the three-year laggards often gave the best results going forward.

171
00:05:06.550 --> 00:05:08.996
That's quite a strong finding against performance chasing.

172
00:05:09.292 --> 00:05:10.175
It really is,

173
00:05:10.300 --> 00:05:15.441
and they even did a Fama-French factor analysis on the difference between the contrarian and trend strategies.

174
00:05:15.566 --> 00:05:16.222
What did that show?

175
00:05:16.863 --> 00:05:18.300
It suggested the difference had,

176
00:05:18.488 --> 00:05:18.738
you know,

177
00:05:18.847 --> 00:05:20.550
positive exposure to value,

178
00:05:20.753 --> 00:05:22.222
negative exposure to momentum,

179
00:05:22.253 --> 00:05:23.347
which kind of makes sense.

180
00:05:23.457 --> 00:05:25.972
But even after accounting for those known factors

181
00:05:26.228 --> 00:05:28.510
there was still some unexplained alpha left over.

182
00:05:28.731 --> 00:05:28.950
Okay,

183
00:05:29.032 --> 00:05:32.172
so let's circle back more directly to valuation.

184
00:05:32.696 --> 00:05:35.555
They looked at the link between how expensive something is,

185
00:05:35.938 --> 00:05:37.297
its relative valuation,

186
00:05:37.797 --> 00:05:39.040
and its later performance.

187
00:05:39.102 --> 00:05:39.461
They did.

188
00:05:39.961 --> 00:05:41.243
Using a blend of metrics,

189
00:05:41.305 --> 00:05:41.680
PE,

190
00:05:41.922 --> 00:05:42.782
price to sales,

191
00:05:42.891 --> 00:05:43.672
price to dividend,

192
00:05:43.750 --> 00:05:44.407
price to book,

193
00:05:44.844 --> 00:05:45.954
all relative to the market,

194
00:05:46.391 --> 00:05:48.907
they found a pretty clear negative relationship.

195
00:05:49.047 --> 00:05:52.047
Negative meaning expensive is bad for future returns.

196
00:05:52.157 --> 00:05:52.797
Exactly.

197
00:05:53.224 --> 00:05:57.105
High relative valuation tended to be followed by lower subsequent returns,

198
00:05:57.546 --> 00:05:58.984
and low relative valuation,

199
00:05:59.207 --> 00:06:00.667
being cheap compared to history,

200
00:06:01.144 --> 00:06:03.410
tended to precede higher future returns.

201
00:06:03.746 --> 00:06:05.527
And this connects back to performance chasing,

202
00:06:05.589 --> 00:06:05.808
right?

203
00:06:06.105 --> 00:06:07.128
Winners get expensive,

204
00:06:07.207 --> 00:06:08.089
losers get cheap.

205
00:06:08.386 --> 00:06:09.253
That's the link they make,

206
00:06:09.285 --> 00:06:09.550
yes.

207
00:06:10.144 --> 00:06:12.753
Good past performance often drives up valuations,

208
00:06:12.816 --> 00:06:13.925
making things expensive,

209
00:06:14.207 --> 00:06:15.925
while poor performance can make them cheap,

210
00:06:16.207 --> 00:06:17.863
potentially setting them up for a rebound.

211
00:06:18.316 --> 00:06:18.425
So,

212
00:06:18.894 --> 00:06:20.035
to really nail this down,

213
00:06:20.457 --> 00:06:22.113
they did an out-of-sample test

214
00:06:22.656 --> 00:06:25.518
Using these relative valuations without looking ahead.

215
00:06:25.659 --> 00:06:25.899
Uh-huh.

216
00:06:26.460 --> 00:06:29.503
They essentially simulated making decisions in real time.

217
00:06:29.925 --> 00:06:30.526
At each point,

218
00:06:30.581 --> 00:06:31.862
they'd look at the available history,

219
00:06:32.308 --> 00:06:36.612
figure out which factors or strategies were cheap or expensive relative to their own past,

220
00:06:36.933 --> 00:06:38.433
and then form portfolios.

221
00:06:38.847 --> 00:06:42.018
How did they compare cheap versus expensive in that test?

222
00:06:42.472 --> 00:06:49.143
They compared investing in the three cheapest versus the three most expensive based on that historical relative valuation measure.

223
00:06:49.656 --> 00:06:53.240
Standardized it so they could compare across different factors and strategies.

224
00:06:53.420 --> 00:06:55.101
And the results of that out-of-sample test?

225
00:06:55.160 --> 00:06:55.902
Pretty compelling.

226
00:06:56.242 --> 00:06:58.207
Over the almost 40-year period,

227
00:06:58.246 --> 00:07:07.496
the portfolio of the three cheapest smart beta strategies and the portfolio of the three cheapest factors both significantly outperformed the equal weight approach and the cap weighted market.

228
00:07:07.637 --> 00:07:08.590
And the expensive ones?

229
00:07:08.918 --> 00:07:10.152
They tended to underperform.

230
00:07:10.574 --> 00:07:14.605
It really supports the idea that relative valuation has predictive power,

231
00:07:14.887 --> 00:07:15.824
even out of sample.

232
00:07:16.262 --> 00:07:18.871
But you mentioned earlier diversification still matters.

233
00:07:19.504 --> 00:07:22.247
How does that fit with focusing on just the three cheapest?

234
00:07:22.868 --> 00:07:23.327
Good point.

235
00:07:23.849 --> 00:07:27.271
They also looked at a strategy they called tilted diversification.

236
00:07:27.970 --> 00:07:29.775
Instead of just the three cheapest,

237
00:07:30.134 --> 00:07:31.939
you hold all the strategies or factors,

238
00:07:32.040 --> 00:07:37.103
but overweight the cheaper ones and underweight the expensive ones on a sliding scale.

239
00:07:37.337 --> 00:07:38.197
And how did that compare?

240
00:07:38.525 --> 00:07:40.853
It often produced better risk-adjusted returns,

241
00:07:40.931 --> 00:07:43.806
a better sharp ratio than just holding the three cheapest,

242
00:07:43.931 --> 00:07:46.243
even if the absolute return was slightly lower.

243
00:07:46.940 --> 00:07:49.820
It suggests that while timing based on valuation is useful,

244
00:07:50.301 --> 00:07:53.340
you probably don't want to abandon diversification completely.

245
00:07:53.723 --> 00:07:54.500
Moderation again.

246
00:07:55.106 --> 00:07:59.082
So finding that balance between the timing signal and keeping diversification benefits.

247
00:07:59.145 --> 00:07:59.699
Precisely.

248
00:08:00.004 --> 00:08:00.160
Now,

249
00:08:00.223 --> 00:08:02.106
the paper does acknowledge something important,

250
00:08:02.184 --> 00:08:02.840
data mining.

251
00:08:03.121 --> 00:08:04.043
How does that factor in?

252
00:08:04.199 --> 00:08:04.340
Yeah,

253
00:08:04.387 --> 00:08:05.184
they're upfront about it.

254
00:08:05.246 --> 00:08:06.668
Like a lot of investment research,

255
00:08:06.731 --> 00:08:11.652
finding factors or strategies often happens after they've already had a good run in the historical data.

256
00:08:12.043 --> 00:08:13.793
So they might look amazing in backtests,

257
00:08:14.012 --> 00:08:14.152
but...

258
00:08:14.368 --> 00:08:17.190
part of that could be because they were identified because they looked amazing.

259
00:08:17.272 --> 00:08:17.731
Exactly.

260
00:08:18.252 --> 00:08:22.455
And that good run might have included a period where they became more expensive,

261
00:08:22.495 --> 00:08:24.299
that revaluation alpha we talked about.

262
00:08:24.838 --> 00:08:29.463
So by the time a factor is discovered or an index is launched,

263
00:08:29.862 --> 00:08:31.971
it might already be priced relatively high.

264
00:08:32.159 --> 00:08:33.424
Do they show any evidence of this,

265
00:08:33.581 --> 00:08:35.627
like performance fading after discovery?

266
00:08:35.706 --> 00:08:36.034
They do.

267
00:08:36.471 --> 00:08:38.752
They present evidence of what they call phantom alpha,

268
00:08:39.315 --> 00:08:42.596
comparing the excess returns of these strategies and factors before

269
00:08:42.608 --> 00:08:45.932
for their index launch or academic publication versus after.

270
00:08:46.192 --> 00:08:47.150
And returns dropped off.

271
00:08:47.514 --> 00:08:48.033
Generally,

272
00:08:48.135 --> 00:08:48.412
yes.

273
00:08:49.033 --> 00:08:52.299
Returns tended to be lower post-launch or post-publication.

274
00:08:52.721 --> 00:08:54.221
Could be data mining bias,

275
00:08:54.518 --> 00:08:56.885
could be that the effect gets arbitraged away,

276
00:08:57.322 --> 00:09:01.463
or it could be that the earlier returns included unsustainable valuation gains.

277
00:09:01.807 --> 00:09:05.088
That's a really crucial point for anyone looking at back-tested strategies.

278
00:09:05.166 --> 00:09:05.713
Absolutely.

279
00:09:05.963 --> 00:09:11.510
Always question how much of the past performance is real structural alpha versus just luck.

280
00:09:11.705 --> 00:09:12.612
or revaluation.

281
00:09:12.814 --> 00:09:14.328
Did they look internationally at all?

282
00:09:14.909 --> 00:09:17.030
Does this valuation timing work outside the U.S.?

283
00:09:17.270 --> 00:09:19.952
They did briefly look at developed ex-U.S.

284
00:09:20.011 --> 00:09:20.530
markets.

285
00:09:20.850 --> 00:09:23.511
The results were maybe a bit weaker.

286
00:09:23.675 --> 00:09:27.612
The difference between cheapest and most expensive wasn't as dramatic as in the U.S.

287
00:09:27.909 --> 00:09:28.690
Any thoughts on why?

288
00:09:29.136 --> 00:09:35.393
They suggest maybe the shorter data history internationally makes it harder to reliably judge historical valuation norms.

289
00:09:36.128 --> 00:09:36.331
Also,

290
00:09:36.456 --> 00:09:37.581
post-financial crisis,

291
00:09:37.628 --> 00:09:39.050
there was a big flight to safety,

292
00:09:39.268 --> 00:09:44.018
which might have pushed up valuations for certain less risky factors globally.

293
00:09:44.457 --> 00:09:46.139
potentially distorting things for a while.

294
00:09:46.440 --> 00:09:48.081
But the general tendency was still there.

295
00:09:48.420 --> 00:09:49.803
Contrarian beating trend.

296
00:09:50.143 --> 00:09:50.561
Generally,

297
00:09:50.784 --> 00:09:51.065
yes.

298
00:09:51.405 --> 00:09:54.905
The cheapest still tended to outperform the most expensive,

299
00:09:54.944 --> 00:09:56.952
just maybe not by as much as in the U.S.

300
00:09:57.053 --> 00:09:57.288
data.

301
00:09:57.428 --> 00:09:57.592
Okay,

302
00:09:57.631 --> 00:09:58.592
so pulling it all together,

303
00:09:59.092 --> 00:10:01.475
what's the key takeaway trading rule from this paper?

304
00:10:01.975 --> 00:10:03.131
I think it boils down to this.

305
00:10:03.709 --> 00:10:05.491
Be a contrarian based on valuation.

306
00:10:05.866 --> 00:10:08.225
Don't just chase past performance periodically,

307
00:10:08.256 --> 00:10:08.975
maybe annually.

308
00:10:09.475 --> 00:10:11.006
Look at your smart beta strategies,

309
00:10:11.007 --> 00:10:12.069
your factor exposures.

310
00:10:12.459 --> 00:10:13.772
See which ones are trading cheap.

311
00:10:13.901 --> 00:10:15.163
relative to their own history.

312
00:10:15.442 --> 00:10:16.403
And lean towards those.

313
00:10:16.704 --> 00:10:16.903
Yeah.

314
00:10:17.223 --> 00:10:18.364
Overweight the cheap ones,

315
00:10:18.485 --> 00:10:20.587
underweight or avoid the expensive ones.

316
00:10:21.028 --> 00:10:25.954
It suggests a systematic way to potentially add value beyond just holding a static mix.

317
00:10:26.477 --> 00:10:28.016
So timing is possible,

318
00:10:28.055 --> 00:10:28.915
the paper argues,

319
00:10:29.376 --> 00:10:33.040
but it's about buying low based on historical norms,

320
00:10:33.587 --> 00:10:35.602
not buying high based on recent returns.

321
00:10:35.790 --> 00:10:37.165
That captures it perfectly.

322
00:10:37.290 --> 00:10:39.524
Consider the price you're paying relative to history.

323
00:10:40.071 --> 00:10:42.571
Don't just extrapolate recent alpha blindly.

324
00:10:43.113 --> 00:10:44.814
This has been a really insightful deep dive.

325
00:10:44.894 --> 00:10:48.099
Lots to think about for implementing factor and smart beta strategies.

326
00:10:48.178 --> 00:10:48.580
Definitely.

327
00:10:48.779 --> 00:10:53.627
Thinking about valuation as a whole other dimension beyond just the factor definitions themselves.

328
00:10:53.822 --> 00:10:56.369
Thank you for tuning in to Papers with Backtest podcast.

329
00:10:56.549 --> 00:10:58.931
We hope today's episode gave you useful insights.

330
00:10:59.377 --> 00:11:01.471
Join us next time as we break down more research.

331
00:11:01.892 --> 00:11:03.299
And for more papers and backtests,

332
00:11:03.330 --> 00:11:06.455
find us at hgtps.paperswithbacktest.com.

333
00:11:06.939 --> 00:11:07.517
Happy trading.