There are certainly examples of transcription factors that regulate genes in response to a stimulus, and the binding of the factor to the genome correlates to both up and down regulated genes.
There is an excellent paper from the Lis and Kraus labs that used a relatively recent technique, GRO-seq to monitor transcription in breast cancer cells genome wide in response to estrogen signalling. The authors employed a novel (in my opinion) hidden Markov model to identify transcripts, including those not yet annotated. This paper is one such example of a factor whose binding correlates to both increased and decreased transcription with similar kinetics, though the mechanism through which a factor may be both negative and positive, seemingly simultaneously, is unclear.
The paper may be of interest from a biological, technical, and analytical perspective