After the opening, when a player judges that straightforward continuations are insufficient to win, she may make a shobute, a play (te) that puts the whole game (shobu) in the balance. Shobute are typically invasions.
Shobute usually involve some risk, but they are not necessarily overplays.
For an example, see the Sakata - Fujisawa game on the thick plays in the endgame page.
JF I have called shoubute a meltdown move before. Others have used all-or-nothing move. It's a move on which you gamble the whole game.
tderz I see here an analogy to option (derivative) theory. The funny difference is that both players always have to close one deal (finish the 1 game, in time), whereas in options you could accept (take) your small, yet definitive loss.
Because in Go it doesn't matter, whether you lose by a margin of 0.5 or a million, the player behind can offer the one ahead a new contract ("rolling through" in options) by changing game strategy. The player ahead has to accept. Under current tournament or Go rules (there is no backgammon dice nor jubango) s/he cannot opt out and walk away with the small, sure win.
While in financials this might only be a way to hide your current losses (if not properly measured) and possibly to increase them largely over time ...
... in Go the player just behind becomes better by this new deal (if all previous evaluations were indeed correct), because the chance for a win increased.
Think of player A offering B a deal where B would lose by a pre-defined margin (even if small) - B tears the contract and offers a highly structured note (incomprehensible to both), a derivative with ultra-high volatility and an expiration over an hour ...
Player A cannot refuse because "There is no simple procedure to turn a clear lead into a victory" and the game must be finished, they have to play a game (contract).
togo: Another point to be made is, that a risky case of shobute nearly necessarily contains a big failure. This can be derived from the fact, that an optimal play results in a draw (or a 0.5 win, if komi is set in tournament manner), so there was an error. Also it is very likely, that there is a calm way to play optimally. This is not proven to my knowledge, but it can be conjectured from some basic facts, mainly the multiple miai-ish structure of the game. Calm play induces that only small losses are possible.
This results in the conclusion, that either one player unintentionally forced the other to play violently, or this other player unintentionally started playing violently, or this other player made an overplay or this other player intentionally chose to play risky. The latter may be an unusually big experiment, a risky style, or a winning strategy based on chance.
Of all this cases only the experiment might be not rated as a failure of some kind, depending on circumstances. Also of course the main goal might be to win a match, not playing the best possible go - then the chance based winning strategy might be rated as sound, but it might be rated a failure to allow such a big chance play.