A scorecard is only honest if you can see what it leaves out. These are the public forecasts I made and deliberately kept out of the graded corpus — one by one, with the reason for each. Some were wrong. One was right and still excluded. None of them are hidden.
The oldest objection to any track record is the simplest: you only show your hits. It is the right objection. A scorecard is worth exactly nothing if the person keeping it also decides, after the fact, which calls count — because then 'accuracy' just measures how good they are at quietly dropping the misses. So the integrity of the JYOTINT corpus does not rest on the calls that are in it. It rests on this page: the calls that are not, named openly, with the reason for each. If the exclusions are principled and visible, the ledger means something. If they are arbitrary and hidden, it means nothing. This essay walks the whole [excluded list](/excluded), one call at a time.
First, the rule, so you can hold me to it. A call is excluded only for reasons fixed in advance and stated in public — ethics, unfalsifiability, responsibility, register, or low information — and never because it turned out wrong. The proof that the rule is real and not a fig leaf is that the wrong answers are still in the corpus. The [Maharashtra 2024 advisory](/intel/maharashtra-2024) is a documented MISS — 288 seat-by-seat calls, graded against the Election Commission result, scored against me. I built it a flagship. A person hiding his misses does not build his biggest miss a flagship. Read this page knowing that the misses are not what I'm leaving out.
Begin with the one exclusion that is purely moral. In November 2023 I posted a reading on [Israel and Gaza](https://x.com/vijayjyotish/status/1723410406201315828). It is sealed and public and I will not touch it, because rendering a live mass-casualty war into a hit/miss tickbox for my own credibility is grotesque, and I am not going to do it. There is a methodological reason too — much of it speaks to a twenty-year horizon and 'the spirit of a people,' which no window can falsify — but the methodological reason is not why I exclude it. Ethics is. Some things should not be scored even if they could be, and a person who cannot say that out loud should not be trusted with the things that can.