I built my own insider trading sentiment extraction.
Here’s some results, 3+ years of the DJIA,
I plot the prior 10 days and the following 10 days to the sentiment event as averages of % change in price compared to event date close price.
Approx. 1200 sentiment dates (many thousands of Form 4 SEC filings coalesced down to single events per security per date).
Evaluation is a simple scale:
-3 strongly bearish
-1 weakly bearish
1 weakly bullish
3 strongly bullish
Interesting to note that there is a general up trend on all events.
That there appears to be a bit of a leak on the strong bullish events.
And that the sellers tend to sell during a strong up trend.
Data looks like this:
Symbol,Date,Sentiment,Prior10,Prior9,Prior8,Prior7,Prior6,Prior5,Prior4,Prior3,Prior2,Prior1,Event,After1,After2,After3,After4,After5,After6,After7,After8,After9,After10 AA,2010-01-25,3,19.09,16.53,19.39,22.46,8.92,12.14,10.90,9.66,9.58,6.88,0.00,-6.00,-6.07,-4.97,-6.66,-9.36,-10.68,-6.29,-4.10,-5.12,-9.22 AA,2010-03-01,-2,-1.09,2.03,-0.23,3.28,2.19,2.27,1.72,1.72,-0.94,-1.88,0.00,-0.08,0.00,-0.47,0.23,0.94,3.99,3.60,2.74,1.95,2.50 AA,2010-04-26,3,7.36,3.91,5.18,3.53,4.21,3.31,0.45,-0.90,-0.68,-0.98,0.00,1.88,1.43,-2.93,-2.03,-0.90,-3.01,-5.03,-9.17,-9.77,-13.60 AA,2010-10-04,-2,-7.01,-7.78,-6.07,-7.78,-3.42,-3.08,0.77,-0.34,0.85,-0.26,0.00,0.94,-1.62,0.26,2.14,0.77,6.41,6.75,8.97,10.43,8.38
After examining the data some more, I reran the plots to use the Median of the price % changes rather than the averages. This plot below is of a secondary group of securities, NOT the DJIA.
The list from the Google Screen criteria:
AI, ALJ, ARII, AWR, AZZ, BANC, BFK, BGG, BLW, BOE, CATO, CFNL, CHI, CHY, CII, CRK, CSGS, DCOM, DUC, EDE, EHI, ERC,ESIO, ETJ, ETV, FFBC, FFC, FPO, GSM, HAFC, HPI, HPS, HQH, HTD, HTGC, HYT, HYV, JFR, JGT, JRO, JSN, KBALB, KND, KNL, LBAI, MCA, MGU, MMU, MTSC, MUE, NEA, NMZ, NP, NPI, NPM, NPP, NPT, PCN, PDT, PFL, PFS, PMO, PNNT, RGS, RNP, SHLM, TAXI, TCAP, TDF, TEI, TSRA, VTA, WIW