ADMK won this seat with just 34.9% of the votes. It is just 3.4 points ahead of TVK. The resignation of the ADMK MLA who is also the contestant for TVK has thrown unique challenges for all three political parties in this race. Booth-level data shows how fragile the win was: if just 5% of ADMK voters follow the now TVK candidate, TVK wins. It's just 1 in 20 votes polled for ADMK in April 2026.
This is not a prediction. We do not have statistically sound pre-poll surveys, and even if we had such a thing, there are chances that it can go the other way. Whoever wins on October 9th, the more interesting question would be where the votes move: which booths held, which booths shifted, and towards whom? A party can win and still lose ground, or lose and still grow. This article sets up what to look for, so the results tell us something beyond who won the by-election in Madurantakam.
Vote share, Madurantakam, April 2026
All valid votes polled
Three parties finished within 5 points of each other: ADMK 34.9%, TVK 31.5% and DMK 30.1%. About two in three voters chose someone other than the winner.
Out of 285 booths, 110 booths were decided by less than a 5% vote margin.
ADMK won 130 booths and it won them big. Only 1 in 5 booths were close calls, compared to half for TVK and DMK. A typical booth win was around 10%, which is double what TVK and DMK managed.
It also has 6 stronghold booths where it won by 37%-42%.
This chart explains the puzzle: it was big wins but a slim overall lead. ADMK's vote is concentrated in clusters of booths where it dominates, and it's weak elsewhere. This is what makes this seat worth the watch. If it holds those clusters, then the base is real, and if they break, then we learn who those votes really belonged to.
TVK edged past DMK by just 1.4 points overall. In a head-to-head count, TVK outpolled DMK in 158 of 285 booths, regardless of who won the booth. Neither party had a clear edge anywhere.
We can't predict how many voters will switch parties and follow their MLA, but we can test how many it would take. Holding DMK's and other parties' votes constant, the chart moves a growing share of ADMK's votes to TVK.
The bar for TVK is remarkably low. Every ADMK voter who follows the former MLA takes a vote away from ADMK and adds one to TVK, so the gap closes twice as fast. TVK needs just 5% of ADMK's voters to overturn the April result, and that's before counting the advantage ruling parties usually enjoy in by-elections.
That's exactly why a TVK win, by itself, proves little. The real signal is the margin. A narrow TVK win would mean the defection barely moved voters. A large win with a substantial vote loss from ADMK, especially where they were strong, would mean the candidate really did help TVK gain more votes, and if a large win results from pulling in votes from all corners, then it means TVK as a party itself is growing. And if TVK falls short despite such a low bar, that would be the biggest story of all.
Test: win or lose, does AIADMK keep its 6 stronghold booths and stay above 30% overall? If it does, the base belongs to the party. If the strongholds fall, those votes belonged to the candidate or the reliable party voters are gradually moving towards TVK. This is a test for the party leadership.
Test: booth data can show where votes moved, not why, so the test looks at where TVK gains. In booths where DMK was strong and AIADMK weak in April, the former MLA had few AIADMK voters to bring. If TVK gains ground there too, it points to a pull toward TVK beyond the defection. If TVK's gains are confined to AIADMK's old strongholds, with AIADMK dropping by a similar amount, the votes simply followed the defection, whether voters were drawn by her or by Vijay.
Test: Congress, VCK, MDMK and IUML, all recent DMK partners, are backing TVK in this by-election. If DMK holds near its April share of 30.1%, the claim is weaker than it sounds. If DMK drops, the booths where it falls will show which ally votes mattered.
Part 2 will compare the October 9 results booth by booth against these tests, and show which narratives held up and which didn't.