There is a post poll forecast at
http://www.lensonnews.com/ (GVLN) that looks similar to 543.
If NDA crossing 300 involves scoring 15 seats in Tamil Nadu, I will consider pre-poll NDA crossing 300 an unlikely outcome. Even counting all seats with an outside chance in TN (south Madras, Vellore, Coimbatore, Kanyakumari, Salem, Dharmapuri, Erode, Virudhunagar), it does not cross 8 seats. To believe in 15, one might need to smoke some very potent stuff.
No point in treating such projections credibly and then blaming it on EVM fraud after results. GVLN's work on the feasibility of EVM fraud is a very credible one and he (and his colleagues) went through a lot of personal hardship for that. So it would be good even for him to keep up that credibility and do not put out such numbers.
Exit polls not trending to the final results by itself is not an indication of EVM fraud (demonstrating the feasibility of that through studies like GVLN's are just enough).
In 2004 exit poll results, trends in TN, AP, and the decline of BJP in UP that started even in 1999, were all missed. The complete route of ADMK was very obvious. AP was a caste issue of other OBC vs Khamma (who were thought to have got max benefits under TDP) - a friend from AP explained this and cautioned about all polls about CB Naidu's 'developer image' even before the elections. Looking back, it is a wonder how exit polls missed these very visible trends.
In 2009, a close contest was not on the cards at all with an Advani leading BJP after Pak/Jinnah secular fiasco, infighting/internal sabotage in rajasthan, Rajnath Singh screwing UP further by becoming friendly with SP (possibly for safety in his own seat). In fact it is a surprise that BJP crossed 100 and held on to most of its strongholds (except rajasthan).
In 2014, I trust in Modi and Amit Shah (having been fooled earlier in 2004 and 2009). But some trends are not encouraging already. There was a deccan chronicle article about 90% polling in Muslim areas in western UP (area including Muzzafarnagar) and 45-55% only polling in Hindu areas. I checked the overall turnout in western UP in the first phase of polling, and that is 65%. Assuming 25% muslim population, 90% of 25% and 55% of 75% adds upto total 65%. Assuming Muslim consolidation behind one candidate (the one likely to be runner up to BJP otherwise), this is cutting it too close - assuming 10 lakh total voters, it is 2.25Lakh muslim votes and 4.125 Hindu votes. If Muslims voted one party, this will need BJP hitting 77% of Hindu votes. Keep in mind that the average muslim population in UP is 25%, and turn out started going down after the first western UP round - final phases was about 55% only.
At close, Amit Shah (while still holding onto 50 seats tally for BJP) was talking of BSP doing better and coming second to BJP.
Why does UP show such low polling %? Is that because poor migrant labour mostly being away from their places and thus never able to vote? One anecdote - a low level painting contractor in Bangalore had all Gorakhpur boys in his employ - they probably never get to vote in their life.