Finding — Vote efficiency: who converted the cheapest, who wasted the most ​
IUML needed 71,000 votes per seat won. BJP needed 1.47 million. NTK wasted all 1.97 million of their votes. DMK wasted 7.6 million in losing races — about the same as TVK's 7.6 million wasted.
Votes per seat (across major parties) ​
| Party | Statewide vote % | Total votes | Seats won | Votes per seat |
|---|---|---|---|---|
| IUML | 0.29% | ~142K | 2 | 71,232 |
| CPI(M) | 0.60% | ~294K | 2 | 146,908 |
| TVK | 34.92% | 17.2M | 108 | 159,502 |
| CPI | 0.66% | ~327K | 2 | 163,244 |
| DMK | 24.19% | 11.9M | 59 | 202,189 |
| AIADMK | 21.21% | 10.5M | 47 | 222,599 |
| VCK | 1.09% | ~540K | 2 | 270,028 |
| PMK | 2.17% | 1.07M | 4 | 267,686 |
| INC | 3.37% | 1.66M | 5 | 332,262 |
| AMMK | 0.86% | ~425K | 1 | 425,345 |
| DMDK | 1.20% | ~589K | 1 | 589,500 |
| BJP | 2.97% | 1.47M | 1 | 1,467,024 |
| NTK | 4.00% | 1.97M | 0 | infinity (wasted) |
| IND | 1.06% | ~525K | 0 | infinity |
| NOTA | 0.41% | ~200K | 0 | (not a party) |
IUML is the most efficient party in TN 2026
IUML contested 2 seats, won 2 seats, needed only 71,232 votes per seat. They picked their fights — minority-Muslim ACs — and won both. Tightest target/spread strategy of any party in TN.
Wasted votes by party (votes cast in losing seats) ​
| Party | Wasted votes | Wasted % of party total |
|---|---|---|
| DMK | 7,627,583 | 64% |
| TVK | 7,587,869 | 44% |
| AIADMK | 6,589,489 | 63% |
| NTK | 1,972,537 | 100% (zero seats) |
| BJP | 1,418,536 | 97% |
| INC | 1,329,825 | 80% |
| PMK | 740,112 | 69% |
DMK and TVK wasted similar absolute votes
DMK wasted 7.63M votes; TVK wasted 7.59M — almost identical absolute waste. But TVK got 108 seats from their effort, DMK only 59. TVK was twice as efficient at converting non-wasted votes into seats.
The two parties that wasted everything ​
- NTK contested all 234 ACs and won zero. Every single one of their 1,972,537 votes went into a losing campaign. 100% wastage.
- IND candidates collectively got 524,744 votes across 232 ACs they contested. No Independent won. 100% wastage.
These are the two "presence-only" entities in TN 2026 — voters who chose them sent a signal but got no representation.
BJP's worst case ​
BJP contested 33, won 1. The 1 they won (their lone victory) absorbed ~50K of their 1.47M total votes. The remaining ~1.42M votes were wasted in 32 contests where they came 3rd or 4th. Their effective "votes per seat" of 1,467,024 is the worst of any major party in TN 2026.
This reflects the NDA seat-sharing calculation that failed: BJP got allocated 33 seats expecting NDA wave + AIADMK transfer. The TVK wave neutralised the anti-DMK vote that would have gone to NDA candidates.
How TVK was so efficient ​
TVK got 34.92% statewide → 108 seats → 159,502 votes/seat. Compare to DMK (202K) and AIADMK (223K). The mechanism:
- High win rate (46.4%) — TVK won close races more often than they lost them
- Concentrated wins — TVK's 108 wins had above-average margins (median ~25K vs TVK statewide ~159K/seat efficiency)
- No alliance arithmetic — TVK ran solo, didn't lose seats to allies, got every vote they earned
The "votes per seat" math benefits parties that:
- Pick fights they can win (IUML's 100% win rate at 2 seats is the extreme case)
- Win those fights by efficient margins (not too narrow, not blowouts)
- Don't waste effort on lost causes (TVK only contested 233 — they had a near-universal strategy)
Vote-share spread per party (compactness) ​
This tells you whether a party ran "high everywhere" or "high in some, low in others":
| Party | Median % | P10-P90 spread | Std dev |
|---|---|---|---|
| INC | 28.19 | 10.5 | 4.6 — most compact |
| DMK | 31.80 | 11.7 | 5.1 |
| BJP | 21.98 | 15.3 | 6.1 |
| TVK | 33.73 | 19.4 | 7.5 |
| PMK | 28.16 | 16.7 | 9.0 |
| AIADMK | 28.84 | 21.0 | 8.3 — most variable |
| NTK | 3.49 | 4.0 | 1.7 |
- INC + DMK are the most compact campaigns — same vote share in similar shape across ACs.
- AIADMK + PMK have wider variance — depends heavily on AC-specific community alignment.
- TVK is broader than DMK but not as variable as AIADMK — wave reached most places, just unevenly.
What this says structurally ​
- Two-party mature dynamics produce stable vote-share spreads (DMK ≈ 5pp std dev)
- Wave-party like TVK produce moderate variance (7.5pp std dev) — they hit most places but not all
- Caste-bloc parties like PMK produce high variance (9pp std dev) — strong where bloc lives, absent elsewhere
- Niche parties (IUML) produce extreme efficiency in small target zones
Cross-check ​
- All numbers from
s3://tnelection2026/results/curated/year=2026/... - Wasted votes = vote total in ACs where party != winner
- Vote-share spread = polars
.std()and.quantile(0.10)/.quantile(0.90)per party - Code:
pipelines/mega_insights.py(S7 + S15 sections)