Model Accuracy
0%
Random Forest 5-CV
F1 Score
0.00
Weighted average
FinBERT Acc.
87.3%
Financial NLP model
AUC-ROC
0.000
Binary classification
Total Samples
0
Training corpus
Sentiment polarity vs stock returns (7-day)
Sentiment
Return %
End-to-end pipeline
AGG ActiveTW
collect
NLP
preprocess
AI
finbert
AGG
aggregate
FE
features
RF
rf model
OUT
predict
Baseline comparison
SentimentEdge (RF+FinBERT)
--
Technical only
63.4%
Sentiment only
58.2%
Random baseline
50.0%
Sentiment distribution
43%
POS
35%
NEU
22%
NEG
Confusion matrix
--
TP — Rise ✓
--
FN — Missed
--
FP — False+
--
TN — Fall ✓
Feature importance
human_sentiment
--
tweet_length
--
rsi_proxy
--
word_count
--
has_cashtag
--
Live tweet feed
Prediction simulator
Sentiment score+0.62
RSI 14-day58
Price vs SMA-201.03
Volume change %+12%
Next-day signal
↑ RISE
Confidence: 78%
FinBERT vs models
FinBERT (ours)
87.3%
RoBERTa-base
81.2%
BERT-base
79.4%
VADER
72.1%
TextBlob
68.9%
BERT-base-uncased · 110M params · 12 layers
Fine-tuned: Financial PhraseBank · 4,840 sentences
Fine-tuned: Financial PhraseBank · 4,840 sentences