Iqs Factors
This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA
Overview
Download & View Iqs Factors as PDF for free.
More details
- Words: 1,127
- Pages: 7
IQS Research Brief Do Factors Persist Anymore? April 29, 2009
In this IQS Research Brief, we take a look at the persistence of factors in this latest environment. Historically, factors had persistence – if a factor was skillful in forecasting returns today, it was likely the factor would still be skillful 4 weeks from now. Similarly, if a factor had no skill today, it probably would not have any skill in the near future. Is this still true? We measure skill using the information coefficient (IC), which is the rank correlation between the model scores and the subsequent returns. The better the agreement, the higher is the IC. To determine the level of persistence, we calculate the correlation between the IC now and the IC 1, 4 and 8 weeks from now. The more they move directionally together, the more positive the correlation. The more they disagree, the more negative the correlation. Positive correlation indicates persistence; negative correlation indicates reversal. Slow moving fundamental factors are expected to persist. Let’s take a look at the numbers below showing the “persistence” for the 5 IQS factor categories from 2004 through May 2007 and since June 2007. The attached pages show graphs of persistence across time (36 week moving average) for each factor category at the four-week persistence level. One-Week Factor Persistence Chart Momentum Value June 2004 – May 2007 June 2007 -
-.060
.022
.133
Improving Financials -.045
.193
.085
.119
.166
.153
Balance Sheet
Sentiment
-.046
Four-Week Factor Persistence Chart Momentum Value June 2004 – May 2007 June 2007 -
Balance Sheet
.091
.148
-.004
Improving Financials .150
-.115
-.092
-.049
-.012
Sentiment -.004
.107
IQS Brief – Do Factors Persist Anymore?
Eight-Week Factor Persistence Chart Momentum Value June 2004 – May 2007 June 2007 -
Balance Sheet
-.018
.089
.104
Improving Financials .152
-.010
-.043
-.156
-.132
Sentiment .033 .003
Observations: One-Week factor persistence for very few factors in the past, but persistence is now realized for all 5 factor categories! Four-Week and Eight-Week factor persistence was positive in the past, but now they are negative! Improving Financials is expected to have longer-period persistence, which it consistently had at four-week and eight-week, but not recently. Momentum does not have persistence over long periods, and wouldn’t expect it to. Value is expected to have persistence over long periods, and did so in the past, but not recently. Conclusions: Quantitative weighting schemes need to look back at the past to help forecast the future, with the assumption that persistence exists. With negative persistence recently, the better plan is to weight factors opposite what has performed well by incorporating the persistence information above into the process. Based on the results above, shorter-term trading models would have performed well since June 2007, while longer-term models would have performed better previously.
Graphs of Four-Week Persistence Correlation Calculations of 36 Week moving average Momentum Category Model 4 Week Lag (36 Week moving average) 0.50 0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40 -0.50 9/
6/
11
10
00 /2
9 00
9
9
8
7
7
07
00 /2
/2
8 00
8 00
8 00
0 /2
00 /2
/2
/2
/2
/6
9 3/
12
4 9/
3 6/
2 3/
00 /2
7
6
6
06
00 /2
0 /2
00 /2
30
29
28
24
/ 11
8/
5/
2/
00 /2
6
5
5
5
4
05
00 /2
0 /2
00 /2
23
22
21
17
/ 11
8/
5/
2/
00 /2
00 /2
00 /2
00 /2
16
15
14
10
/9
/ 11
8/
5/
2/
11
IQS Brief – Do Factors Persist Anymore?
Value Category Model 4 Week Lag (36 Week moving average) 0.50
0.40
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
9/ 6/
11 10
00 /2
9 00
00 /2
/2
8 00 8 00 8 00 0 /2
9 9
8
4
5
5
5
05
6
6
6
06
7
7
7
07
00 /2
/2
/2
/2
/6
9 3/ 12 4 9/ 3 6/ 2 3/
0 /2
00 /2
00 /2
00 /2
30
24
28
29
/ 11 8/ 5/ 2/
0 /2
00 /2
00 /2
00 /2
23
17
21
22
/ 11 8/ 5/ 2/
00 /2
00 /2
00 /2
00 /2
16
/9
10
14
15
/ 11 8/ 5/ 2/
11
IQS Brief – Do Factors Persist Anymore?
Balance Sheet Category Model 4 Week Lag (36 Week moving average) 0.50
0.40
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
9/ 6/
11 10
00 /2
9 00
00 /2
/2
8 00 8 00 8 00 0 /2
9 9
8
4
5
5
5
05
6
6
6
06
7
7
7
07
00 /2
/2
/2
/2
/6
9 3/ 12 4 9/ 3 6/ 2 3/
0 /2
00 /2
00 /2
00 /2
30
24
28
29
/ 11 8/ 5/ 2/
0 /2
00 /2
00 /2
00 /2
23
17
21
22
/ 11 8/ 5/ 2/
00 /2
00 /2
00 /2
00 /2
16
/9
10
14
15
/ 11 8/ 5/ 2/
11
IQS Brief – Do Factors Persist Anymore?
Improving Financials Category Model 4 Week Lag (36 Week moving average) 0.50
0.40
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
9/ 6/
11 10
00 /2
9 00
00 /2
/2
8 00 8 00 8 00 0 /2
9 9
8
4
5
5
5
05
6
6
6
06
7
7
7
07
00 /2
/2
/2
/2
/6
9 3/ 12 4 9/ 3 6/ 2 3/
0 /2
00 /2
00 /2
00 /2
30
24
28
29
/ 11 8/ 5/ 2/
0 /2
00 /2
00 /2
00 /2
23
17
21
22
/ 11 8/ 5/ 2/
00 /2
00 /2
00 /2
00 /2
16
/9
10
14
15
/ 11 8/ 5/ 2/
11
IQS Brief – Do Factors Persist Anymore?
Sentiment Category Model 4 Week Lag (36 Week moving average) 0.50
0.40
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
9/ 6/
11 10
00 /2
9 00
00 /2
/2
8 00 8 00 8 00 0 /2
9 9
8
4
5
5
5
05
6
6
6
06
7
7
7
07
00 /2
/2
/2
/2
/6
9 3/ 12 4 9/ 3 6/ 2 3/
0 /2
00 /2
00 /2
00 /2
30
24
28
29
/ 11 8/ 5/ 2/
0 /2
00 /2
00 /2
00 /2
23
17
21
22
/ 11 8/ 5/ 2/
00 /2
00 /2
00 /2
00 /2
16
/9
10
14
15
/ 11 8/ 5/ 2/
11
In this IQS Research Brief, we take a look at the persistence of factors in this latest environment. Historically, factors had persistence – if a factor was skillful in forecasting returns today, it was likely the factor would still be skillful 4 weeks from now. Similarly, if a factor had no skill today, it probably would not have any skill in the near future. Is this still true? We measure skill using the information coefficient (IC), which is the rank correlation between the model scores and the subsequent returns. The better the agreement, the higher is the IC. To determine the level of persistence, we calculate the correlation between the IC now and the IC 1, 4 and 8 weeks from now. The more they move directionally together, the more positive the correlation. The more they disagree, the more negative the correlation. Positive correlation indicates persistence; negative correlation indicates reversal. Slow moving fundamental factors are expected to persist. Let’s take a look at the numbers below showing the “persistence” for the 5 IQS factor categories from 2004 through May 2007 and since June 2007. The attached pages show graphs of persistence across time (36 week moving average) for each factor category at the four-week persistence level. One-Week Factor Persistence Chart Momentum Value June 2004 – May 2007 June 2007 -
-.060
.022
.133
Improving Financials -.045
.193
.085
.119
.166
.153
Balance Sheet
Sentiment
-.046
Four-Week Factor Persistence Chart Momentum Value June 2004 – May 2007 June 2007 -
Balance Sheet
.091
.148
-.004
Improving Financials .150
-.115
-.092
-.049
-.012
Sentiment -.004
.107
IQS Brief – Do Factors Persist Anymore?
Eight-Week Factor Persistence Chart Momentum Value June 2004 – May 2007 June 2007 -
Balance Sheet
-.018
.089
.104
Improving Financials .152
-.010
-.043
-.156
-.132
Sentiment .033 .003
Observations: One-Week factor persistence for very few factors in the past, but persistence is now realized for all 5 factor categories! Four-Week and Eight-Week factor persistence was positive in the past, but now they are negative! Improving Financials is expected to have longer-period persistence, which it consistently had at four-week and eight-week, but not recently. Momentum does not have persistence over long periods, and wouldn’t expect it to. Value is expected to have persistence over long periods, and did so in the past, but not recently. Conclusions: Quantitative weighting schemes need to look back at the past to help forecast the future, with the assumption that persistence exists. With negative persistence recently, the better plan is to weight factors opposite what has performed well by incorporating the persistence information above into the process. Based on the results above, shorter-term trading models would have performed well since June 2007, while longer-term models would have performed better previously.
Graphs of Four-Week Persistence Correlation Calculations of 36 Week moving average Momentum Category Model 4 Week Lag (36 Week moving average) 0.50 0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40 -0.50 9/
6/
11
10
00 /2
9 00
9
9
8
7
7
07
00 /2
/2
8 00
8 00
8 00
0 /2
00 /2
/2
/2
/2
/6
9 3/
12
4 9/
3 6/
2 3/
00 /2
7
6
6
06
00 /2
0 /2
00 /2
30
29
28
24
/ 11
8/
5/
2/
00 /2
6
5
5
5
4
05
00 /2
0 /2
00 /2
23
22
21
17
/ 11
8/
5/
2/
00 /2
00 /2
00 /2
00 /2
16
15
14
10
/9
/ 11
8/
5/
2/
11
IQS Brief – Do Factors Persist Anymore?
Value Category Model 4 Week Lag (36 Week moving average) 0.50
0.40
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
9/ 6/
11 10
00 /2
9 00
00 /2
/2
8 00 8 00 8 00 0 /2
9 9
8
4
5
5
5
05
6
6
6
06
7
7
7
07
00 /2
/2
/2
/2
/6
9 3/ 12 4 9/ 3 6/ 2 3/
0 /2
00 /2
00 /2
00 /2
30
24
28
29
/ 11 8/ 5/ 2/
0 /2
00 /2
00 /2
00 /2
23
17
21
22
/ 11 8/ 5/ 2/
00 /2
00 /2
00 /2
00 /2
16
/9
10
14
15
/ 11 8/ 5/ 2/
11
IQS Brief – Do Factors Persist Anymore?
Balance Sheet Category Model 4 Week Lag (36 Week moving average) 0.50
0.40
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
9/ 6/
11 10
00 /2
9 00
00 /2
/2
8 00 8 00 8 00 0 /2
9 9
8
4
5
5
5
05
6
6
6
06
7
7
7
07
00 /2
/2
/2
/2
/6
9 3/ 12 4 9/ 3 6/ 2 3/
0 /2
00 /2
00 /2
00 /2
30
24
28
29
/ 11 8/ 5/ 2/
0 /2
00 /2
00 /2
00 /2
23
17
21
22
/ 11 8/ 5/ 2/
00 /2
00 /2
00 /2
00 /2
16
/9
10
14
15
/ 11 8/ 5/ 2/
11
IQS Brief – Do Factors Persist Anymore?
Improving Financials Category Model 4 Week Lag (36 Week moving average) 0.50
0.40
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
9/ 6/
11 10
00 /2
9 00
00 /2
/2
8 00 8 00 8 00 0 /2
9 9
8
4
5
5
5
05
6
6
6
06
7
7
7
07
00 /2
/2
/2
/2
/6
9 3/ 12 4 9/ 3 6/ 2 3/
0 /2
00 /2
00 /2
00 /2
30
24
28
29
/ 11 8/ 5/ 2/
0 /2
00 /2
00 /2
00 /2
23
17
21
22
/ 11 8/ 5/ 2/
00 /2
00 /2
00 /2
00 /2
16
/9
10
14
15
/ 11 8/ 5/ 2/
11
IQS Brief – Do Factors Persist Anymore?
Sentiment Category Model 4 Week Lag (36 Week moving average) 0.50
0.40
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
9/ 6/
11 10
00 /2
9 00
00 /2
/2
8 00 8 00 8 00 0 /2
9 9
8
4
5
5
5
05
6
6
6
06
7
7
7
07
00 /2
/2
/2
/2
/6
9 3/ 12 4 9/ 3 6/ 2 3/
0 /2
00 /2
00 /2
00 /2
30
24
28
29
/ 11 8/ 5/ 2/
0 /2
00 /2
00 /2
00 /2
23
17
21
22
/ 11 8/ 5/ 2/
00 /2
00 /2
00 /2
00 /2
16
/9
10
14
15
/ 11 8/ 5/ 2/
11
Related Documents
Iqs Factors
April 2020 5
Iqs October
June 2020 5
Iqs Vix Low
May 2020 9
Factors
November 2019 53
Factors
August 2019 77
Iqs Vix S&p
June 2020 10More Documents from "Zerohedge"
Pitchbook Private Equity Breakdown 4q 2009
June 2020 2
Annaly Capital
May 2020 5
Petition_signees July 16
May 2020 7
Profit Act
May 2020 9
Calsters
April 2020 5