Jarvis V2

v2.0.0
0/7 Agents
Uptime:
BQ
API
Live
Loading market data...

Regression Analysis

Statistical regression models, feature importance, residual diagnostics, and prediction accuracy

Last Model Run

2026-02-15 09:41 EST

1,248 observations

Model Quality Overview

R-Squared

0.8743

Variance explained

Adj. R-Squared

0.8591

Penalized for features

Mean Absolute Error

0.0234

Avg absolute deviation

Root Mean Sq. Error

0.0312

Penalizes large errors

F-Statistic

57.82

p-value (F-test)

< 0.0001

Durbin-Watson

2.04

Feature Importance

Relative contribution of each predictor to the Elastic Net regression model

1Options Flow Sentiment
p=1e-418.7%
2Volume
p=2e-416.2%
3Congressional Trades
p=8e-413.4%
4Short Interest
p=0.001412.1%
5Sector Momentum
p=0.002110.9%
6Insider Activity
p=0.00399.4%
7Market Cap
p=0.00677.8%
8Macro Indicators
p=0.01126.3%
9PE Ratio
p=0.02415.2%

Residual Analysis

Mean Residual

0.00032

Std. Deviation

0.0298

Skewness

-0.142

Kurtosis

3.21

Jarque-Bera Normality Test

JB = 4.87 | p = 0.0876

NORMAL

Residual Distribution

n=12
n=38
n=124
n=287
n=341
n=264
n=112
n=48
n=22
-0.090.00+0.09

Multicollinearity Check

Variance Inflation Factors — VIF > 5 concerning, VIF > 10 highly correlated

VIF < 55 – 10VIF > 10

Macro Indicators

11.42HIGH

Sector Momentum

6.31WARN

Short Interest

4.89OK

Market Cap

3.47OK

Insider Activity

2.74OK

PE Ratio

2.12OK

Congressional Trades

1.92OK

Volume

1.83OK

Options Flow Sentiment

1.56OK
!

High Multicollinearity Detected

Macro Indicators — consider removing or combining these features to reduce coefficient instability.

!

Moderate Collinearity

Sector Momentum — monitor for inflated standard errors. Ridge or Elastic Net regularization recommended.

Prediction vs Actual

Recent out-of-sample prediction accuracy — 5-day forward returns

Directional Accuracy

70% (7/10)

TickerPredicted Dir.Actual Dir.Pred. Mag. (%)Actual Mag. (%)Error (%)Result
NVDALONGLONG+4.2+3.80.4CORRECT
AAPLLONGLONG+1.8+2.10.3CORRECT
TSLASHORTSHORT-3.1-4.71.6CORRECT
METALONGSHORT+2.4-1.23.6MISS
MSFTLONGLONG+1.5+1.30.2CORRECT
AMZNLONGLONG+2.9+3.40.5CORRECT
GOOGLSHORTLONG-0.8+0.61.4MISS
AMDLONGLONG+5.1+4.30.8CORRECT
JPMLONGLONG+1.2+1.00.2CORRECT
PLTRLONGSHORT+3.6-2.15.7MISS

Model Comparison

Performance across regression methodologies — best model highlighted

Active Model: Ridge

OLS

R-Squared0.8743
MAE0.0234
RMSE0.0312
AIC-4,821
BIC-4,756
Train Time1.2s

Ridge

BEST
R-Squared0.8812
MAE0.0218
RMSE0.0295
AIC-4,889
BIC-4,824
Train Time1.4s

Lasso

R-Squared0.8651
MAE0.0251
RMSE0.0337
AIC-4,762
BIC-4,701
Train Time1.3s

Elastic Net

R-Squared0.8789
MAE0.0223
RMSE0.0301
AIC-4,871
BIC-4,808
Train Time1.8s
Train/Test Split: 80/20Cross-Validation: 5-foldRegularization (alpha): 0.001L1 Ratio: 0.50
Pipeline active — next retrain in 4h 12m