Network analysis identifies the system; the regime model reads its state.
Members: UNP, WAB, GBX, TRN, CSX, NSC
Step 1 · Who moves together
daily return correlation
UNP
WAB
GBX
TRN
CSX
NSC
UNP
1.00
0.49
0.35
0.41
0.72
0.82
WAB
0.49
1.00
0.50
0.50
0.46
0.47
GBX
0.35
0.50
1.00
0.63
0.36
0.36
TRN
0.41
0.50
0.63
1.00
0.40
0.40
CSX
0.72
0.46
0.36
0.40
1.00
0.77
NSC
0.82
0.47
0.36
0.40
0.77
1.00
High correlation means these names are one bet, not several. That is what makes the system thesis true and what makes owning all of them concentration rather than diversification.
Step 2 · Who moves first
|corr| > 0.088 is the ~5% significance line
same day corr
partner→UNP best lag
UNP→partner best lag
read
NSC
0.82
-0.099lag 4
-0.080lag 1
NSC leads by 4d
CSX
0.72
+0.081lag 5
-0.106lag 1
UNP leads by 1d
GBX
0.35
-0.074lag 4
+0.030lag 5
no lead-lag
WAB
0.49
+0.069lag 5
-0.074lag 2
no lead-lag
TRN
0.41
+0.058lag 5
+0.053lag 5
no lead-lag
A partner that leads UNP is a signal. Daily lead-lag correlations are usually tiny and often noise — treat anything near the threshold as a hypothesis to re-test on fresh data, not a finding.
Step 3 · What state the system is in
equal-weighted basket of all 6 names
System state today
CALM
89% probability
Calm regime
39days
mean +0.07%/day · vol 1.11
Stressed regime
2days
mean +0.62%/day · vol 3.88
System basket
+53%
501 days · 22% ann vol
Fitted 2-regime Markov switching model (statsmodels). The regimes were inferred from the data — nothing was hand-labeled. Expected durations come from the estimated transition matrix, same 1/(1-p) idea as your rule-based dashboard.
Members: NVDA, AMAT, TSM, MRVL, AVGO, VRT
Step 1 · Who moves together
daily return correlation
NVDA
AMAT
TSM
MRVL
AVGO
VRT
NVDA
1.00
0.55
0.71
0.54
0.63
0.67
AMAT
0.55
1.00
0.67
0.58
0.55
0.59
TSM
0.71
0.67
1.00
0.58
0.67
0.68
MRVL
0.54
0.58
0.58
1.00
0.59
0.58
AVGO
0.63
0.55
0.67
0.59
1.00
0.61
VRT
0.67
0.59
0.68
0.58
0.61
1.00
High correlation means these names are one bet, not several. That is what makes the system thesis true and what makes owning all of them concentration rather than diversification.
Step 2 · Who moves first
|corr| > 0.088 is the ~5% significance line
same day corr
partner→NVDA best lag
NVDA→partner best lag
read
TSM
0.71
-0.116lag 1
-0.120lag 4
NVDA leads by 4d
AVGO
0.63
-0.113lag 1
-0.172lag 4
NVDA leads by 4d
VRT
0.67
-0.074lag 1
-0.131lag 4
NVDA leads by 4d
MRVL
0.54
-0.073lag 1
-0.123lag 1
NVDA leads by 1d
AMAT
0.55
-0.070lag 1
-0.142lag 1
NVDA leads by 1d
A partner that leads NVDA is a signal. Daily lead-lag correlations are usually tiny and often noise — treat anything near the threshold as a hypothesis to re-test on fresh data, not a finding.
Step 3 · What state the system is in
equal-weighted basket of all 6 names
System state today
CALM
86% probability
Calm regime
26days
mean +0.41%/day · vol 2.04
Stressed regime
8days
mean -0.25%/day · vol 4.71
System basket
+202%
501 days · 46% ann vol
Fitted 2-regime Markov switching model (statsmodels). The regimes were inferred from the data — nothing was hand-labeled. Expected durations come from the estimated transition matrix, same 1/(1-p) idea as your rule-based dashboard.
Members: SBUX, KC=F, KDP, NSRGY, DNUT, EWZ
Step 1 · Who moves together
daily return correlation
SBUX
KC=F
KDP
NSRGY
DNUT
EWZ
SBUX
1.00
-0.01
0.20
0.15
0.25
0.23
KC=F
-0.01
1.00
-0.02
0.07
0.03
0.13
KDP
0.20
-0.02
1.00
0.31
0.11
0.13
NSRGY
0.15
0.07
0.31
1.00
0.06
0.19
DNUT
0.25
0.03
0.11
0.06
1.00
0.08
EWZ
0.23
0.13
0.13
0.19
0.08
1.00
High correlation means these names are one bet, not several. That is what makes the system thesis true and what makes owning all of them concentration rather than diversification.
Step 2 · Who moves first
|corr| > 0.088 is the ~5% significance line
same day corr
partner→SBUX best lag
SBUX→partner best lag
read
KC=F
-0.01
+0.123lag 5
+0.080lag 2
KC=F leads by 5d
DNUT
0.25
+0.095lag 3
-0.045lag 4
DNUT leads by 3d
EWZ
0.23
+0.092lag 4
+0.079lag 2
EWZ leads by 4d
KDP
0.20
-0.085lag 5
+0.044lag 1
no lead-lag
NSRGY
0.15
-0.064lag 3
+0.103lag 1
SBUX leads by 1d
A partner that leads SBUX is a signal. Daily lead-lag correlations are usually tiny and often noise — treat anything near the threshold as a hypothesis to re-test on fresh data, not a finding.
Step 3 · What state the system is in
equal-weighted basket of all 6 names
System state today
CALM
89% probability
Calm regime
29days
mean +0.03%/day · vol 1.03
Stressed regime
3days
mean -0.05%/day · vol 2.48
System basket
+9%
501 days · 19% ann vol
Fitted 2-regime Markov switching model (statsmodels). The regimes were inferred from the data — nothing was hand-labeled. Expected durations come from the estimated transition matrix, same 1/(1-p) idea as your rule-based dashboard.
Members: HSY, CC=F, MDLZ, NSRGY, SBUX
Step 1 · Who moves together
daily return correlation
HSY
CC=F
MDLZ
NSRGY
SBUX
HSY
1.00
-0.19
0.50
0.27
0.13
CC=F
-0.19
1.00
-0.14
0.01
0.02
MDLZ
0.50
-0.14
1.00
0.37
0.18
NSRGY
0.27
0.01
0.37
1.00
0.15
SBUX
0.13
0.02
0.18
0.15
1.00
High correlation means these names are one bet, not several. That is what makes the system thesis true and what makes owning all of them concentration rather than diversification.
Step 2 · Who moves first
|corr| > 0.088 is the ~5% significance line
same day corr
partner→HSY best lag
HSY→partner best lag
read
CC=F
-0.19
-0.070lag 3
+0.071lag 4
no lead-lag
NSRGY
0.27
-0.069lag 4
-0.077lag 3
no lead-lag
MDLZ
0.50
+0.062lag 4
-0.049lag 4
no lead-lag
SBUX
0.13
-0.053lag 5
-0.071lag 5
no lead-lag
A partner that leads HSY is a signal. Daily lead-lag correlations are usually tiny and often noise — treat anything near the threshold as a hypothesis to re-test on fresh data, not a finding.
Step 3 · What state the system is in
equal-weighted basket of all 5 names
System state today
CALM
98% probability
Calm regime
63days
mean +0.03%/day · vol 0.99
Stressed regime
2days
mean -0.06%/day · vol 3.10
System basket
+9%
501 days · 18% ann vol
Fitted 2-regime Markov switching model (statsmodels). The regimes were inferred from the data — nothing was hand-labeled. Expected durations come from the estimated transition matrix, same 1/(1-p) idea as your rule-based dashboard.
Members: ADM, BG, ZC=F, ZS=F, ZW=F, DE, NTR, UNP
Step 1 · Who moves together
daily return correlation
ADM
BG
ZC=F
ZS=F
ZW=F
DE
NTR
UNP
ADM
1.00
0.71
0.04
0.20
0.05
0.31
0.39
0.25
BG
0.71
1.00
0.07
0.23
0.05
0.29
0.38
0.16
ZC=F
0.04
0.07
1.00
0.56
0.58
0.00
0.11
-0.07
ZS=F
0.20
0.23
0.56
1.00
0.43
0.11
0.19
0.00
ZW=F
0.05
0.05
0.58
0.43
1.00
0.06
0.11
-0.05
DE
0.31
0.29
0.00
0.11
0.06
1.00
0.25
0.39
NTR
0.39
0.38
0.11
0.19
0.11
0.25
1.00
0.09
UNP
0.25
0.16
-0.07
0.00
-0.05
0.39
0.09
1.00
High correlation means these names are one bet, not several. That is what makes the system thesis true and what makes owning all of them concentration rather than diversification.
Step 2 · Who moves first
|corr| > 0.088 is the ~5% significance line
same day corr
partner→ADM best lag
ADM→partner best lag
read
UNP
0.25
-0.097lag 4
-0.064lag 3
UNP leads by 4d
DE
0.31
-0.091lag 4
-0.065lag 3
DE leads by 4d
ZS=F
0.20
+0.079lag 1
-0.059lag 3
no lead-lag
ZW=F
0.05
-0.075lag 5
-0.038lag 4
no lead-lag
BG
0.71
-0.071lag 3
+0.073lag 1
no lead-lag
ZC=F
0.04
-0.055lag 5
-0.076lag 4
no lead-lag
NTR
0.39
-0.047lag 3
-0.037lag 4
no lead-lag
A partner that leads ADM is a signal. Daily lead-lag correlations are usually tiny and often noise — treat anything near the threshold as a hypothesis to re-test on fresh data, not a finding.
Step 3 · What state the system is in
equal-weighted basket of all 8 names
System state today
CALM
92% probability
Calm regime
8days
mean +0.02%/day · vol 0.77
Stressed regime
1days
mean +0.46%/day · vol 1.54
System basket
+39%
501 days · 14% ann vol
Fitted 2-regime Markov switching model (statsmodels). The regimes were inferred from the data — nothing was hand-labeled. Expected durations come from the estimated transition matrix, same 1/(1-p) idea as your rule-based dashboard.
Members: VMI, LNN, CTVA, DE, NTR, MOS
Step 1 · Who moves together
daily return correlation
VMI
LNN
CTVA
DE
NTR
MOS
VMI
1.00
0.45
0.33
0.41
0.15
0.23
LNN
0.45
1.00
0.36
0.50
0.18
0.29
CTVA
0.33
0.36
1.00
0.45
0.47
0.47
DE
0.41
0.50
0.45
1.00
0.25
0.36
NTR
0.15
0.18
0.47
0.25
1.00
0.59
MOS
0.23
0.29
0.47
0.36
0.59
1.00
High correlation means these names are one bet, not several. That is what makes the system thesis true and what makes owning all of them concentration rather than diversification.
Step 2 · Who moves first
|corr| > 0.088 is the ~5% significance line
same day corr
partner→VMI best lag
VMI→partner best lag
read
CTVA
0.33
-0.113lag 3
+0.051lag 5
CTVA leads by 3d
NTR
0.15
-0.104lag 3
+0.081lag 2
NTR leads by 3d
DE
0.41
-0.090lag 4
-0.076lag 2
DE leads by 4d
MOS
0.23
-0.068lag 3
-0.074lag 4
no lead-lag
LNN
0.45
-0.031lag 3
+0.090lag 1
VMI leads by 1d
A partner that leads VMI is a signal. Daily lead-lag correlations are usually tiny and often noise — treat anything near the threshold as a hypothesis to re-test on fresh data, not a finding.
Step 3 · What state the system is in
equal-weighted basket of all 6 names
System state today
CALM
86% probability
Calm regime
5days
mean +0.06%/day · vol 0.92
Stressed regime
2days
mean +0.12%/day · vol 2.02
System basket
+44%
501 days · 22% ann vol
Fitted 2-regime Markov switching model (statsmodels). The regimes were inferred from the data — nothing was hand-labeled. Expected durations come from the estimated transition matrix, same 1/(1-p) idea as your rule-based dashboard.