P300 waveforms using Cyton-Daisy

I am using a Cyton-Daisy 16-channel EEG acquisition system. I am trying to use this system along with post-processing on Matlab for a P300 experiment.

In my experiment I have a green square image as the frequent stimulus and a red circle as the rare stimulus (appearing with probability of 1/7). Each image is shown for 750ms with the same amount of time between images. I am currently not using any stimulus markers (via photodiodes etc) as recommended in other posts. I am also not recording/saving the EEG signals continuously. Rather, I save the EEG signals for only the 750ms during which the images are displayed and not during the time when the inter-stimulus grey background is displayed.

I have attached a snippet of the main part of Matlab code that is acquiring the raw EEG data using LSL. After acquiring all the data, I pass each epoch data through a 8th order HPF with fc = 0.1 Hz and a 36th order LPF with fc = 15 Hz. I have attached the averaged ERP plot as well.

Now I understand that using photodiodes taped to a corner of the screen for generating stimulus markers and recording continuous EEG signals would be much more accurate. I would probably do that eventually but at first I wanted to get idea as to whether using this method I can at least start seeing P300 waveforms to some extent. If not, why not? The Rare stimulus response has a slight peak around 300 ms and is higher than the frequent stimulus response for at least the 1st and 2nd channels. Is there any way like for example, starting to record the EEG a few milliseconds before the 'imshow' is performed, can help get better ERPs?